Thom Hartle - Active Trader Magazine - Trading Strategies Analysis Collection Vol1

Thom Hartle Trading Strategy and Analysis collectionVol. 1: 2001-2004

2 On-target trading (Active Trader, July 2001)

7 Finding value opportunities (Active Trader, Aug. 2001)

13 Know thy market: Gauging typical price behavior (Active Trader, Oct. 2001)

18 Fundamentally technical trading (Active Trader, July 2002)

22 Familiarity breeds profitability (Active Trader, Sept. 2002)

30 Bull vs. Bear The details matter (Active Trader, Nov. 2002)

36 Equity curve drawdown management (Active Trader, Feb. 2003)

40 Opening shots (Active Trader, April 2003)

43 What goes up must come down (Active Trader, May 2003)

49 What's the time? (Active Trader, Aug. 2003)

54 Trend, consolidations and unchanged volume (Active Trader, Oct. 2003)

57 The telltale spread (Active Traderr, Nov. 2003)

61 Choosing the proper time frame (Active Trader, Dec. 2003)

65 The method trader (Active Traderr, April 2004)

71 Nasdaq 100 volume and the QQQ (Active Trader, July 2004)

T rading, like any other pro-fession, is an acquired set ofskills. Granted, certain indi-viduals naturally possess

the prerequisite abilities for successfultrading: a superior memory for patternrecognition, supreme confidence andplenty of capital. As a result, they hit theground running.

However, there’s a learning curve forthe rest of us — a step-by-step process ofobtaining skills in areas such as tradeentry, trade exit and money manage-ment (taking profits and contro l l i n grisk). How to set a target for taking prof-its is an essential part of this process.

Before explaining the method of settingtargets, we’ll review the basic guidelinesof the trading process we’ll be workingwith.

First, technical analysis lends itself toa scientific approach of defining yourtrading procedure. The word lends isimportant here, because not all technicalanalysis methods can be used in a scien-tific way. For example, the relationshipbetween today’s closing price and a 10-

day moving average of the closing priceis an observable condition with only oneof three possible outcomes: Today’s closeis either greater than, less than or equalto the moving average.

The advantage of using technicalapproaches with such precise definitionsis that you can convert the definitionsinto trading rules or pro c e d u res andthen test them on historical data to deter-mine the outcome.

For example, let’s compare two tech-nical analysis concepts that point to anexpected outcome. The first is the closevs. the moving average, as describedabove. The second is the break of thetrendline referred to as the “neckline” ofthe head-and-shoulders top pattern.

The first is very specific, while the lat-ter is difficult to precisely define. Canyou precisely define the head-and-shoul-ders chart pattern in combination withthe break of the neckline in such a waythat you can create a rule and test its out-come? If you can, great. If not, you mightwant to avoid this kind of analysis.

What is the problem with not havingp recise definitions? Imagine you arereviewing charts, trying to measure the

outcome of a particular chart reversalpattern. Seeing the outcome togetherwith the pattern may bias your reviewprocess. For example, it’s possible you’llsee the pattern at market tops but fail torecognize it (or a similar pattern) in themiddle of a trend when no re v e r s a loccurs. Or you may convince yourselfthat you’d know when to trade the pat-tern when it was accurate, and know toavoid it in the situations it failed.

Hindsight is one thing, but in re a l - t i m etrading you don’t know the outcome —you can’t see the pattern and the re s u l t ,t o g e t h e r, in advance — and your judg-ment may fail to provide the sameadvantage you assumed existed whenyou did know the outcome. Such circ u m-stances can reduce your confidence andraise your anxiety level — both majorobstacles in the path of effective trading.

However, once you have precise defi-nitions, a whole new world of develop-ing trading procedures opens to you.

This idea of using precise definitions isreally the foundation of the scientific

2 www.activetradermag.com • July 2001 • ACTIVE TRADER

O N- TA R G E T t r a d i n gTaking profits is a balancing act: Hold on too long

and you risk giving back your gains. Get out too soon

and your gains might not be worth holding on to.

By using a scientific approach to solve this puzzle

you can establish intelligent profit targets that

enhance your performance. BY THOM HARTLE

TRADING Strategies

continued on p. 3


method credited to Sir Francis Bacon.The scientific method requires you tofollow certain steps toward understand-ing and finding a solution to a problem.

One reason to follow the steps outlinedin the scientific method is to remove anybias the researcher may bring to the situ-ation. We’ll approach the problem of set-

ting profit targets using the steps of thescientific method.

Step 1: State the problem. You can’tsolve a problem unless you know whatthe problem is. Our problem is to deter-mine where to take profits after a tradingsignal is triggered. Consequently, thesolution is twofold: We have to define,first, an entry signal and, second, the tar-geting technique.

Step 2: Research the problem. Look toother people’s work in the area of entrysignals and taking profits. There aremany sources for trading signals (whichwe will mention shortly), but regardingprofit-taking, we’ll use John Sweeny’sbook, Campaign Trading: Tactics andStrategies to Exploit the Markets and aresearch technique he calls measuringmaximum favorable excursion.

Maximum favorable excursion (MFE)is the positive price movement of a tradefrom beginning to end. After we enterinto a trade there will be a range of pos-sible price movement. Some trades mayhave little, or possibly zero, positiveprice movement. For example, a longentry executed at the high of the daywould have no positive price movement.On the other hand, if the entry occurredat what turns out to be the start of a sub-stantial trend, the positive price move-ment for this trade will be the best underreview. In between these two extremesof maximum favorable excursions lie theoutcomes of all of the other trades.

For example, if you bought a stock at20 and it rallied to 30, but your exit tech-nique didn’t trigger an exit until it sub-sequently dropped to 27, your profitwould be seven points but the MFEwould be 10 points. Let’s define ourentry signals.

We will use the difference between theMACD line and its signal line (referredto as the MACD histogram) to indicatetrend and trigger long and short posi-tions. (For more on the MACD, seeIndicator Insight, p. 88.) The entry signalwill be to buy when the MACD his-togram turns positive, staying long untilthe MACD histogram turns negative.The sell-short rule is the opposite: Sellshort when the MACD histogram turnsnegative and stay short until the MACDhistogram turns positive. This system isalways in the market. Now, on to Step 3.

In addition to the trading range period shown in Figure 1, the test period alsoincluded trend periods. During this downtrend, the MACD histogram signaledtwo good short trades.

FIGURE 2 TRENDING PERIOD

1,800

1,750

1,700

1,650

1,600

1,550

1,500

1,450

1,400

1,350

1,300

8 . 8 1

1 . 2 4

- 6 . 3 3

- 1 3 . 9

DowntrendMACDsystem 1

3/27/01 3/28/013/29/01 3/30/01 4/2/01 4/3/01 4/4/01 4/5/01 4/6/01

Nasdaq 100 Index (NDX), 30-minute

Source: Fibonacci Trader

At point A, the MACD histogram went positive (on the close) at a price of1,656.54, triggering a buy. The MACD histogram switched to a short positionon the close at point B at 1,673.62, for a 17.07-point profit. The peak pricewhile in the trade was 1,749.29, for an MFE of 92.75 points.

FIGURE 1 MAXIMUM FAVORABLE EXCURSION (MFE)

1,800

1,760

1,720

1,680

1,640

1,600

1,560

- . 0 4

- 8 . 6

- 1 7 . 1 5

A

B

Trading range

MFEMACD system -1

3/16/013/19/01 3/20/01 3/21/01 3/22/01 3/23/01 3/26/01 3/27/01 3/28/01

Nasdaq 100 Index (NDX), 30-minute


ACTIVE TRADER • July 2001 • www.activetradermag.com 4

Step 3: Form a hypothesis or solutionto the problem. We will assume therewill be positive outcomes from the trad-ing signals and we will measure theMFE for each signal. It may turn out theentry signal concept is flawed, which isvaluable information. Most likely, withsuch a simple approach, the entry rulesare a starting point, and by using MFEanalysis, you can review the results andrepeat the process with the new infor-mation. (The importance of testing anytrading idea cannot be stressed enough.This research ultimately saves you a con-siderable amount of money.)

Step 4: Test the hypothesis. To performthis test we will review 30-minute barsof the Nasdaq 100 from Dec. 22, 2000, toApril 4, 2001. (This is too short a timeperiod for confidence in the systemitself, but more than adequate to demon-strate the analytical concept.) TheNasdaq 100 can be traded using theNasdaq 100 tracking stock (QQQ) or theNasdaq 100 futures contact.

Figure 1 (opposite page, top) shows aseries of trades during March when theNasdaq 100 was in a trading range;Figure 2 (opposite page, bottom) showstrades during a trending period. At theclose of a bar when the MACD his-togram crosses into positive territory (abuy signal), the price bars turn greenand an up arrow is displayed. (Theentire bar is colored green even thoughthis action takes place at the close of thebar.) When the MACD histogram dropsinto negative territory (a sell signal), thebars are colored red and a down arrow isdisplayed.

For each trade we will measure theprofit or loss as well as the MFE. Forexample, at point A in Figure 1, theMACD histogram went positive on theclose at a price of 1,656.54. At point B theMACD histogram signaled a short posi-tion on the close at a price of 1,673.62, fora 17.07-point profit. The peak pricewhile in the trade was 1,749.29, for aMFE of 92.75 points.

I d e a l l y, we would like to have our ini-tial test period include a trend and somesideways price movement so we see howthe strategy performs in diff e rent markete n v i ronments. Figure 3 (top right) is adaily chart showing the period of the his-torical data base on a daily bar basis. Wecan see January was a rising trend while

t h e re was a persistentd o w n t rend duringF e b ruary and Marc h .

F i g u re 4 (bottom right)lists the performancestatistics of this trad-ing system over thetest period. Figure 5(p. 48) shows thetrade-by-trade sum-m a r y. This systemappears to be quitesuccessful but, again,the test period is tooshort to draw any re a-sonable conclusions.H o w e v e r, for our pur-poses, the first part ofour proposed solution( t h e re are positivetrade outcomes) hasbeen met. Let’s nowadd in the MFE analy-sis for each trade.

One way toapproach this analysisis to step through eachtrade on the scre e none bar at a time and

FIGURE 4 PERFORMANCE SUMMARY

Performance Results for Nasdaq 100 Index

30- D- W System From 12/22/2000 14:00 MACD System to 4/5/2001 14:30

Gross Profit 2,730.22Gross Loss -832.94Net 1,897.28

Total Trades 47.00Total Winning Trades 24.00Total Losing Trades 23.00Percent Profitable 51.06

Largest Winning Trade 266.83Largest Losing Trade -89.81Average Winning Trade 113.76Average Losing Trade -36.21Ratio Average Win/Average Loss 3.14Average Trade 75.81

Max. Consecutive Winners 4.00Max. Consecutive Profit 909.41Max. Consecutive Losers 5.00Max. Consecutive Draw Down -253.96

Source: MetaStock Professional

The system posted a profitable return of 1,897.28points (not including slippage or commissions).

The Nasdaq 100 rallied into late January and then entered a persistent downtrend during February and March.

FIGURE 3 DAILY PERSPECTIVE

2,600

2,400

2,200

2,000

1,800

1,600

140,000

Uptrend

Downtrend

26 2 8 16 22 29 1 5 12 20 26 1 5 12 19 26 2

January 2001 February March April

Nasdaq 100 Index (NDX), daily


1,448.24

continued on p. 5

log the information into Excel for fur-ther analysis. This may seem like a lot ofwork — it is — but the returns fromyour investment of time and energy aresignificant.

You gain a number of insights com-pared to asking the computer to do thework for you. First, you will feel morecomfortable with the trading approachand have more confidence, because in asense, you experience each trade. Youwill see the good times and the bad, soyou will not feel uneasy when yourstrategy experiences a pre d i c t a b l edrawdown.

Second, you will quickly realize ifyou have a rule that does not makesense, or a market situation for whichyou don’t know the appropriate action.Finally, you will bring to bear your ownintuitive skills for refining your tech-nique. None of this happens when youask the computer to do the work.However, once we have the data we willuse Excel to analyze the results.

Figure 6 (opposite page, top) displaysthe MFE results as a histogram, arrang-ing each trade by order of the size of itsMFE. The minimum value was zero,and the maximum was 359.86 points,which happened to be the second tradein the test period.

F i g u re 7 (opposite page, middle) sortsthe MFEs of long and short positions.The majority of long signals cluster fro m100 and below (with the exception oftwo large values, which are trades fro mearly January). The MFEs of the shortpositions tend to steadily increase. Bothof these observations make sense, con-sidering the market was in a downtre n dm o re than two thirds of the test period.

It appears that these trading signalsare a valid approach because of theoverall pro f i t a b i l i t y. In addition, theMFE outcomes indicate we can identifya profitable target level for most trades.Now, how should we establish a target?

Let’s return to Figure 6, which showsthe MFE of all of the trades ord e red bysize. We can use a target based on what-ever we choose from this graph, incre a s-ing or decreasing the size of the targ e tbased on the likelihood of hitting the tar-get. For example, let’s say we choose 50points as our target price. This value wasexceeded more than 33 times, or 68 per-cent of the time. Let’s see how we canalter our previous rules using this targ e t .


MACD SystemDate Time Position Price P&L P&L Accum.12/28/2000 12:00 short 2435.17 0.001/3/2001 9:30 long 2,187.07 248.09 248.091/4/2001 13:30 short 2,453.9 266.82 514.921/8/2001 14:30 long 2,281.54 172.35 687.281/12/2001 9:00 short 2,503.66 222.11 909.401/12/2001 9:30 long 2,548.94 - 45.28 864.121/12/2001 11:00 short 2,513.52 - 35.41 828.701/17/2001 8:30 long 2,599.27 - 85.75 742.951/17/2001 14:00 short 2,566.45 - 32.82 710.131/18/2001 11:30 long 2,621.14 - 54.68 655.441/19/2001 11:00 short 2,657.67 36.53 691.971/23/2001 10:00 long 2,659.16 - 1.49 690.481/24/2001 12:00 short 2,716.47 57.31 747.791/26/2001 12:00 long 2,618.14 98.33 846.121/30/2001 13:00 short 2,669.74 51.60 897.721/31/2001 9:00 long 2,713.19 - 43.44 854.271/31/2001 10:30 short 2,685.15 - 28.04 826.232/1/2001 14:30 long 2,607.17 77.98 904.212/2/2001 9:00 short 2,564.07 - 43.09 861.112/5/2001 13:30 long 2,451.03 113.04 974.152/7/2001 8:30 short 2,434.82 - 16.20 957.952/7/2001 14:30 long 2,409.66 25.16 983.112/8/2001 14:00 short 2,364.79 - 44.86 938.242/12/2001 9:00 long 2,293.92 70.87 1,009.112/13/2001 13:30 short 2,263.73 - 30.18 978.922/14/2001 13:00 long 2,284.74 - 21.01 957.912/15/2001 14:30 short 2,367.09 82.35 1,040.262/21/2001 10:00 long 2,108.95 258.14 1,298.402/21/2001 14:30 short 2,058.54 - 50.40 1,247.992/22/2001 12:00 long 2,068.31 - 9.77 1,238.222/23/2001 9:00 short 1,979.44 - 88.87 1,149.352/23/2001 13:30 long 1,986.52 - 7.08 1,142.272/27/2001 9:30 short 2,021.37 34.85 1,177.123/1/2001 13:30 long 1,877.52 143.85 1,320.973/5/2001 8:30 short 1,905.67 28.15 1,349.123/6/2001 8:30 long 1995.48 - 89.80 1,259.313/6/2001 14:30 short 1,976.31 - 19.16 1,240.143/12/2001 13:00 long 1,748.33 227.98 1,468.123/15/2001 13:00 short 1,736.44 - 11.89 1,456.233/19/2001 9:00 long 1,656.54 79.89 1,536.133/20/2001 13:30 short 1,673.62 17.07 1,553.213/22/2001 10:30 long 1,616.02 57.60 1,610.813/26/2001 11:30 short 1,708.56 92.54 1,703.353/27/2001 10:00 long 1,734.15 - 25.59 1,677.763/27/2001 13:30 short 1,715.33 - 18.82 1,658.943/30/2001 10:30 long 1,554.82 160.51 1,819.454/2/2001 12:00 short 1,525.61 - 29.20 1,790.244/4/2001 10:00 long 1,418.57 107.04 1,897.28Last Pos. Value 1,519.09 1,997.80

FIGURE 5 TRADE-BY-TRADE SUMMARY

Source: MetaStock Professional

After four consecutive winners at the beginning of the test period, the system suffered six losing trades in a row.


We will use both target and trend-fol-lowing exits for each trade. We will takehalf our profit at a target of 50 pointsand hold the remainder of the tradeuntil the MACD histogram re v e r s e sagain. Figure 8 (bottom, right) showsthe equity line for the original trend-fol-lowing positions, along with the equityline using the additional target exit. (Weare only interested in the performanceof adding the targeting scheme; theadditional necessary capital is not anissue here.)

Of course, this additional step addedm o re money, but more importantly,notice how the original equity line sys-tem peaked at trade 4 and did not sur-pass this level until trade 19. By com-parison, the targeting scheme versionpassed the previous high at trade 13.Also, the original equity line from trade19 to 26 was flat, while the targetingscheme version continued to add prof-its. Most trend-following methods tendto have lengthy periods of flat or nega-tive equity growth, and the targ e tscheme counters this drawback.

Step 5: Drawing conclusions from thetesting. Based on this testing we canconclude that the MFE approach is ana p p ropriate method for determiningprofit targets. You can use the MFE tocreate targets that suit your risk toler-ance by using larger or smaller valuesdepending on how active you want tobe. Other MFE measurements may beused for targets, such as perc e n t a g eprice movement or a ratio based on thec u r rent average true range. A l s oremember that other issues must beaddressed when designing a system orstrategy, such as slippage and commis-sions, risk and required capital.

Granted, this targeting method isbased on hindsight. So, the first stepwould be to test the approach on differ-ent (and more) historical data and see ifthe performance holds up. This is called“out-of-sample” testing, and is a verynecessary step if you wish to draw reli-able conclusions from your testing.

Finally, as with all technical tradingmethods, past profitable performancedoes not guarantee future profits, but ifyou take the time to do this type ofwork you will gain both experience anddevelop profitable techniques.Ý

This chart shows that the MFEs for the trades in the test ranged from zero to 350-plus points.

FIGURE 6 MFE RANKED BY SIZE

400

350

300

250

200

150

100

50

0

Trades101 4 7 2213 16 19 3425 28 31 4637 40 43

MFE histogram

The upper curve represents the short sale positions. The lower curve repre -sents the long positions. On average, the long positions are smaller becausethe market was in a downtrend for the majority of the test period.

FIGURE 7 MFE SORTED BY SIZE AND POSITION

400

350

300

250

200

150

100

50

071 3 5 159 11 13 2317 19 21

Short positions

Long positions

The bottom curve shows the performance of the original system. The topcurve shows the performance with the addition of a 50-point profit target.The target continues to add profits to the system while the original trend-following technique is in a flat period.

FIGURE 8 EQUITY CURVES

3,4 0 0

2,9 0 0

2,4 0 0

1,9 0 0

1,4 0 0

9 0 0

4 0 0

- 1 0 0142 6 10 3018 22 26 4634 38 42164 8 12 3220 24 28 4836 40 44

Target plus trend-following positions

Trend-followingpositions

D uring the latter part of the 1990s, successful stocktraders had one mantra: High tech is the place tobe. In reaction to the phenomenal bull market,traders quickly adapted to techniques that meas-

ured momentum, be it price, earnings or sales momentum. The goal was to hop on the backs of stocks that showed

acceleration of any kind, whether it was price- or fundamental-

based. Capital flowed to the leaders in the high-tech industrybecause the growth rates using the aforementioned measure-ments were so high. The risk, though, was great. If there wasany hint of slowdown in the acceleration of a company’s busi-ness — not even anything as serious as an actual loss — yourposition was hammered in a wink of an eye.

In the first quarter of 2000, a slowdown engulfed the entirehigh-tech sector, and these stocks— measured both individuallyand by the Nasdaq Composite —lost more than 50 percent of theirvalue over the subsequent 12months. Once seemingly invinci-ble market leaders such as CiscoSystems (CSCO), Sun Micro-systems (SUNW) and Ya h o o(YHOO), among others, plum-meted.

That’s been the headline news.But away from the front page,everything isn’t falling apart. Infact, a new theme appears to beshaping up in the market: under-valued companies. We’ll exploretechniques for identifying un-dervalued companies and meth-ods for trading them.

Figures 1 through 3 illustrate thechanges that have taken place inthe market in 2000 and early2001. Figure 1 (left) is the S&P500 Barra Growth index andFigure 2 (opposite page) is theS&P 500 Barra Value index.

7 www.activetradermag.com • August 2001 • ACTIVE TRADER

Finding VALUE OPPORT U N I T I E SWhere can you find high-potential stocks in a battered market? Adding technical

signals to basic fundamental analysis can allow you to identify value stocks poised

to move.

This index represents the stocks within the S&P 500 with price-to-book ratios greaterthan the average price-to-book ratio in the S&P 500 index.

FIGURE 1 S&P BARRA GROWTH INDEX

1000

950

900

850

800

750

700

650

600

550

500

450

400

350

300

250

S&P 500 (Barra Growth), weekly

1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1

Source: MetaStock, Equis International (www.equis.com)

TRADING & Investing

BY THOM HARTLE

The growth index is a capitaliza-tion-weighted index of the com-mon stocks within the S&P 5 0 0index with price-to-book ratioshigher than the index average (thebook value of a stock is the networth of the company divided bythe number of outstanding shares).

The value index is a capitaliza-tion-weighted index of the com-mon stocks within the S&P 5 0 0index with price-to-book ratioslower than the index average.Figure 1 shows that growth stocksfell precipitously beginning in thefirst quarter of 2000, while valuestocks, shown in Figure 2, movedto new highs over the same period.

Figure 3 (bottom right) providesa clearer picture of the relation-ships between these two indices.Here we see the growth and valueindices using a logarithmic scale,which measures how the indicesperformed on a percentage basis.You can see that the two tracked inclose fashion from 1996 until themiddle of 1998, at which pointgrowth stocks took off (A). Then, in2000, the relative strength compari-son traced out a double top andgrowth stocks fell relative to thevalue stocks.

Whether or not this current trendwill continue is difficult to know.However, because it is in place itwarrants bringing a value-orienteda p p roach to your trading andinvesting. The following approachshows how to identify value stocksand how to determine when tocommit capital to them.

Traditionally, value is defined bycomparing various “fundamentals”to the price of a stock — for exam-ple, price earnings (P/E) ratio,book value, sales ratio and so on.

A value-oriented technique dic-tates that investments should onlybe made in companies whose ratiosa re at historical low values. Thelogic is that such a company hasfallen out of favor with Wall Stre e t ,(as reflected by the low value

ACTIVE TRADER • August 2001 • www.activetradermag.com 8

continued on p. 9

The relative performance of growth and value stocks. The top chart shows a double top formed on a relative strength basis, a bearish sign for growth stocks.

FIGURE 3 S&P 500 BARRA GROWTH VS. VALUE INDEX

1.601.551.501.451.401.351.301.251.201.151.101.051.000.950.90

900

800

700

600

500

400

300

Double top

AGrowth stocks accelerate

Growth

Value

1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1


This index represents the stocks within the S&P 500 with price-to-book ratios lowerthan the average price-to-book ratio in the S&P 500 index.

FIGURE 2 S&P BARRA VALUE INDEX

690680670660650640630620610600590580570560550540530520510500490480470460450440430420410400390380370360350340330320

S&P 500 (Barra Value), weekly

1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1



The Ergodic Candlestick Oscillator (ECO) is detailed inWilliam Blau’s book, Momentum, Direction, andDivergence (1995, John Wiley & Sons). To calculate it,

two other concepts must first beexplained: “double smoothing” andthe Candlestick Indicator (CSI).

Blau uses exponential movingaverages (EMAs) extensively in hiscalculations. In an effort to createsmooth signals while minimizingprice lag, Blau used a double-smoothing technique — that is, heapplied an EMA to the raw pricedata, and then performed a secondsmoothing of the first EMA using anadditional EMA. Blau uses the stan-dard formula for an EMA, which takesa price observation, such as theclose, and multiplies it by a con-stant, called the alpha (a), which, inthe following formula, representsthe lookback period used for a simplemoving average:

a = 2/(n+1)wheren = the lookback period for

a simple moving average

In other words, the exponentialsmoothing constant (a) to approxi-mate a 20-day simple moving aver-age would be .095 (2/[n+1]).

The adjusted closing price using 1-alpha is added to yesterday’s EMAvalue, which has been multiplied byalpha. Here is the formula for theEMA:

EMA = (1-a)P + a(EMA (t-1))wherep = priceEMA (t-1) = yesterday’s EMA

The double-smoothed closingprice series (EMA Double) uses theEMA of the closing price:

EMA Double = (1-a) EMA + a(EMA Double(t-1))whereEMA Double(t-1) =yesterday’s EMA value

Blau uses the notationEMA(EMA(P,r),s) to indicate a long-term EMA (generally 26 days for the

CSI) and a short-term EMA (generally five days for the CSI).The next step is the calculation of the Candlestick

Indicator. The CSI is a ratio of the double-smoothed differ-

The ECO crossing above and below the signal line generates buy and sell signals,respectively. In this case, the ECO signal occurred eight bars before the MACDsignal.

CHART 2 PENETRATING THE SIGNAL LINE

Real Networks (RNWK), daily

ECO crosses above the signal line

Ergodic Candlestick Oscillator

MACD

A

B MACD crosses above the signal line2 / 2 2 / 2 0 0 1 2 / 2 3 / 2 0 0 1 2 / 2 6 / 2 0 0 1

Source: Fibonacci Trader (www.fibonaccitrader.com)

8 . 1 6

8 . 0 0

7 . 4 8

7 . 3 2

7 . 1 6

7 . - 0 0

1 8 . 3 6

5 . 7

6 . 9 6

- 1 9 . 6 1

. 0 5

- . 0 3

- . 1 1

- . 1 8

Even though RNWK opened on a gap-down, the CSI continued to rise and theMACD remained flat.

CHART 1 GAP-DOWN OPENING

Real Networks (RNWK), daily

CSI rises

A

MACD is flat

2 / 2 2 / 2 0 0 1 2 / 2 3 / 2 0 0 1 2 / 2 6 / 2 0 0 1


8 . 1 6

8 . 0 0

7 . 4 8

7 . 3 2

7 . 1 6

7 . 0 0

1 8 . 3 6

5 . 7

6 . 9 6

- 1 9 . 6 1

. 0 5

- . 0 3

- . 1 1

- . 1 8


MACD

Market timing: ECO

ratios), but that its management team will respond to pre s s u ref rom shareholders and take steps to improve earnings. Theprice of the stock will ultimately rise as management impro v e sits business practices.

T h e re are a number of value-oriented techniques; the one wewill look at uses historical dividend yields as a basis for value.

Value Trend Analysis is a company that publishesInvestment Quality Trends (IQ Trends), a value-orientedappraisal of stocks. It uses individual historical dividend yieldanalysis of each stock from the company’s approved list of 350“Blue Chips” to determine when a stock is undervalued (i.e.,has a historical high dividend yield), in a rising trend, overval-ued or in a declining trend. IQ Trends uses the followingrequirements to qualify stocks as blue chips:

• The dividend has been raised five times in the last 12 years.• The company carries a Standard and Poor’s Quality rank-

ing of “A.”• The company has at least 5 million shares outstanding.• At least 80 institutional investors hold the stock.• There have been at least 25 years of uninterrupted dividends.

continued on p. 11

ence between the open and close of each price bar and adouble-smoothed difference between each bar’s high andlow:

CSI = 100(Ema(Ema (close-open,r),s)/(Ema(Ema (high-low,r),s)

The CSI measures momentum based on how each barcloses relative to the range for the bar. Closes at theupper end of each bar’s range and above the openingprice indicate strength and positive CSI raw values.Closes at the lower end of each bar and below the open-ing price indicate weakness.

A trend in either direction will have persistent read-ings. The double-smoothing technique will create asmooth indicator line that rises if the market is in anuptrend and falls if prices are in a downtrend.

Intraday traders will value this indicator because thecalculation does not reference the pre-vious bars’ closes. Consequently, anylarge price gap — such as a gap open orgap close — will not cause this indicatorto fluctuate dramatically.

For example, Chart 1 is a 15-minutechart of Real Networks (RNWK). Theupper histogram chart is a 26-period CSIand the lower histogram chart is the 12-26-period moving average convergence-divergence (MACD) indicator (seeIndicator Insight, Active Trader, July2001, p. 88). Notice that at point A,there is a gap-down opening, but theMACD, which looks at the market on aclose-to-close basis, remains virtuallyflat. On the other hand, the CSI contin-ues to rise from the previous day’s read-ing because the majority of the 15-minute bars are closing near theirrespective highs, indicating strength.

The ECO is a plot of two lines. Theprimary is the ECO, which uses the CSIwith an additional smoothing by a five-period EMA, and the addition of a “sig-nal line,” which is a five-period movingaverage of the ECO (similar to thedefault nine-period moving average sig-nal line of the MACD).

Basic buy and sell signals occur whenthe ECO crosses above and below its sig-nal line, respectively. Chart 2 comparesthe ECO (upper histogram) and theMACD (lower histogram). As a short-term bottom was forming, the ECOcrossed above its signal line eight barsbefore the MACD.

IQ Trends measures an individual company’s undervalued and overvalued levels based on historical yields.

FIGURE 4 IQ TRENDS CHART

Wachovia Banc Corp. (WB)Regional bank throughout N & S Carolina and Georgia

Mid January 2001

PriceDividend

Overvalued yield line

Undervalued yield line

Dividend trend line

Earnings trend line

Source: Investment Quality Trends (www.iqtrends.com)

E r n .1 . 9 4 2 . 1 3 2 . 1 4 2 . 5 1 2 . 8 3 3 . 1 3 3 . 5 0 3 . 8 1 3 . 9 6 4 . 4 5 4 . 9 7 4 . 6 0D i v. 0 . 7 0 0 . 8 2 0 . 9 2 1 . 0 0 1 . 1 1 1 . 2 3 1 . 3 8 1 . 5 2 1 . 6 8 1 . 8 6 2 . 0 6 2 . 2 8

1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0


95

85

7570

6560

55

5046

42

38

34

302826

24

22

20

18

16151413

1211

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

24

16

8

24

16

8

.90

.80

.70

• The earnings have improved in at least seven of the last 12years.

In mid-January 2001, IQ Trends listed 30 companies asundervalued, 89 in rising trends, 116 overvalued and 114 indeclining trends. Many of the companies listed have beenaround most of the past century and are household names,such as JP Morgan Chase and Eastman Kodak. Figure 4 (p. 97)

is an excerpt from the newsletterthat illustrates the price chart,dividend yield bands and otherattributes of the company,including earnings trend.

These companies are certainlyless volatile than the typicalNasdaq stock. While the list ofblue chip stocks may not be themost exciting, they do off e ropportunities for traders andinvestors alike.

The technique to be discussedmay be primarily of interest toinvestors or longer-term traders,but since shorter-term tradersalso improve their chances ofsuccess by identifying stocksand sectors most likely to move,it will also provide valuableinformation for them. Considerthis technique as one of a num-ber of trading strategies youcould employ. In other words,diversify not only into differentmarkets, but among diff e re n ttechniques as well.

As our first criterion, we willbegin with a list of undervaluedstocks — those that have fallenin price and reached their highhistorical dividend yield. How-e v e r, we are faced with oneproblem: By putting capital intoundervalued companies, we riskthe possibility that the stockprice may languish, tradingsideways for a considerableamount of time.

H o w e v e r, use of technicalanalysis allows you to spot situa-tions that indicate price action isbeginning to move into anu p t rend. A good tool for this iscalled the Ergodic CandlestickOscillator (ECO), developed byWilliam Blau. In the charts that

f o l l o w, we will plot the ECO as a histogram and the signal lineas a dot. For more information, see “Market timing: ECO,” p. 9.

Our plan is to look for stocks from IQ Trends’ undervaluedlist and then use the ECO to determine which ones are in anuptrend. The trend turns up when the ECO histogram is innegative territory and rises to the point that the signal line isnot within the histogram. Our work will be with weekly charts.


Bank of America has traded into undervalued territory a number of times, offeringtrade opportunities at points A, B and D (confirmed by the ECO crossovers).

FIGURE 5 BANK OF AMERICA

6 8 . 0 0

6 4 . 0 0

6 0 . 0 0

5 6 . 0 0

5 2 . 0 0

4 8 . 0 0

4 4 . 0 0

4 0 . 0 0

3 6 . 0 0

3 2 . 0 0

3

- 4 . 6 8

- 1 2 . 3 5

- 2 0 . 0 3

Bank of America (BAC), weekly

Undervalued price=45

A

B

C

D

D1

B1

A1

Ergodic candlestick oscillator

7 / 5 / 9 9 1 0 / 4 / 9 9 1 / 3 / 0 0 4 / 3 / 0 0 7 / 3 / 0 0 1 0 / 2 / 0 0 1 / 1 / 0 1


S&P 500/Barra Growth and S&P 500/Barra Value Indices

In 1992, Standard and Poor’s (S&P) and Barra worked in partnership to producegrowth and value subsets of S&P’s equity indices. The S&P 500/Barra Growth andS&P500/Barra Value Indices separate fast-growing companies from slower growing,

or undervalued, companies, based upon a price-to-book value calculation (the price ofthe stock divided by the “book value,” or net worth of the company).

The price-to-book ratio captures one of the fundamental differences between com-panies classified as value companies or growth companies: Growth companies tend tohave higher price-to-book ratios than value companies. Also, price-to-book ratios tendto be more stable over time compared to alternative measures such as the price-to-earnings ratio, historical earnings growth rates or return on equity. Consequently, thegrowth and value indices experience relatively low turnover.

Companies in the growth index have higher market caps, on average, than those in thevalue index. As a result, there are many more companies in the value index than thegrowth index. As of this writing, the Growth Index had 125 companies while the Va l u eIndex listed 375 companies. More information can be found at the Standard & Po o r’s We bsite (www.spglobal.com or www.spglobal.com/indexmain500style_data.html) andB a r r a ’s Web site (www. b a r r a . c o m ) .

F i g u re 5 (opposite page) shows Bank of America (BAC).A c c o rding to the January IQ Trend, BAC was undervalued at aprice of 45 (giving it a dividend yield of 5 percent). During the

first quarter of 2000, BAC tradedbelow 45. At point A, the ECOhistogram rose above the signalline. On this bar, BAC closed at50. BAC then moved above 60, again of more than 20 percent. Inlate June 2000, BAC againd ropped into undervalued terri-t o r y, and the ECO flashed a buysignal (point B) with a closingprice of 52.50. The stock thenadvanced to 57.63 for a 5-pointgain. At point C, BAC againd ropped to below the underval-ued level. However, notice thatduring the short two-week rally,the ECO never gave a buy signal.BAC again fell back into under-valued territory at point D, andthe ECO gave a buy signal at aprice of 46.13. The stock ralliednearly 10 points after that. Inre g a rd to taking profits, you canuse a trailing stop or a hardmoney target, or take partial pro f-its at a combination of the two.

Figure 6 (top left) is DillardsInc. (DDS). The undervaluedprice for DDS is 11, or a yield of1.5 percent. DDS reached thisprice in October 2000, and theECO gave a buy signal that samemonth (point A) at a price of10.88. In mid-May 2001, the stocktraded above 18.

Our final example, Figure 7(bottom left), is HJ Heinz (HNZ).The undervalued price is 35, or ayield of 4.5 percent. HNZreached 35 twice in 2000. Thefirst time, (point A) was in thefirst quarter. The ECO gave a buysignal at 33.59, and the stockadvanced to 45.94. The secondsignal came in September 2000.The ECO flashed a new uptrendat a price of 38.94. In May 2001,the stock traded as high as 48.

Will the price/earnings ratioc o m p ression continue in theNasdaq stocks? Will value stocksbe the market for this decade? No

one knows, but broadening your methods to position yourselffor whatever happens is a good step to take. That way, you cantake advantage of the current theme driving the market.Ý


Heinz traded into undervalued territory twice last year. Each buy signal was followedby a greater than 10-point gain.

FIGURE 7 HJ HEINZ

5 2 . 0 0

4 8 . 0 0

4 4 . 0 0

4 0 . 0 0

3 6 . 0 0

3 2 . 0 0

2 8 . 0 0

4 . 7 4

- 5 . 8 2

- 1 6 . 3 7

- 2 6 . 9 3

H.J. Heinz (HNZ), weekly


A

B

A1

B1


1 / 4 / 9 9 4 / 5 / 9 9 7 / 5 / 9 9 1 0 / 4 / 9 9 1 / 3 / 0 0 4 / 3 / 0 0 7 / 3 / 0 0 1 0 / 2 / 0 0 1 / 1 / 0 1


This company reached undervalued territory for a three-month period before anuptrend began, as signaled by the ECO (point A).

FIGURE 6 DILLARDS INC.

2 0 . 0 0

1 8 . 0 0

1 6 . 0 0

1 4 . 0 0

1 2 . 0 0

1 0 . 0 0

8 . 0 0

- 5 . 8 9

- 1 5 . 7 1

- 2 5 . 5 4

Dillards Inc. (DDS), weekly


A

A1


1 / 3 / 0 0 2 / 7 / 0 0 3 / 6 / 0 0 4 / 3 / 0 0 5 / 1 / 0 0 6 / 5 / 0 0 7 / 3 / 0 0 8 / 7 / 0 0 9 / 4 / 0 0 1 0 / 2 / 0 0 1 1 / 6 / 0 0 1 2 / 4 / 0 0 1 / 1 / 0 1 2 / 5 / 0 1


W hen it comes to the price behavior of a par-ticular stock or market, many tradersthink of broad con-cepts like a tenden-

cy to trend, or relatively high or lowvolatility compared to other markets.

While such characteristics are usefulfor generalizing market behavior, morespecific, statistical measures of priceaction are better for actual trading, espe-cially for short-term traders who aremore dependent on the quality of theirentries and exits than long-term traders.

The simplest statistics can provide agreat deal of insight into a stock’s behav-ior. For instance, knowing the averagedifference between the open and low fordays the market closes up would be use-ful for improving your trade entries andstop placement. This is just one exampleof the kind of information that can begleaned from analyzing price action bar-by-bar.

We will look at some unique ways toobserve market behavior and then showhow to use this information in your trad-ing. First, we’ll walk through some sim-ple observations of market behaviorusing the QQQs, the exchange-tradedfund that tracks the Nasdaq 100 index.The same analysis, which can be done in

a spreadsheet such as Excel, can be applied to any market orindividual stock you trade.

13 www.activetradermag.com • October 2001 • ACTIVE TRADER

From March 30 to mid-June, the QQQs displayed a wide range of price behav -ior: an extended uptrend, short-term uptrends, a sideways trading range andshort-term downtrends.

FIGURE 1 GAUGING PRICE ACTION

2 9 16 23 1 7 14 21 29 1 11April May June

5 0 . 0 0

4 7 . 5 0

4 5 . 0 0

4 2 . 5 0

4 0 . 0 0

3 7 . 5 0

3 5 . 0 0

Source: CQGNet

Nasdaq 100 Tracking Stock (QQQ), daily

KNOW THY MARKET: Gauging typical price behaviorRegardless of what kind of trader you are or what approach you use,

knowing the typical price behavior for the markets you trade is essential.

H e r e ’s how to do it.

BY THOM HARTLE

The principle price behavior characteristics we’ll look atare the daily range, the close-to-close range and the differ-ences between days that close up vs. those that close down.For clarity, we will start by analyzing small amounts of dataand then expand our analysis.

The first price behavior characteristic to analyze is the typ-ical daily range. Figure 1 (opposite page) shows a dailychart of the QQQs from March 30 to the middle of June, aperiod that encompassed up, down and sideways pricemovement.

Figure 2 (right, top) shows the daily range (high minuslow) for each bar in the period in Figure 1. The blue line isa three-period moving average (MA) of the daily rangesand the red line is a 10-period MAof the ranges. Both MAsreveal the typical daily range is approximately $2, the jumpat bar 13 notwithstanding.

The second characteristic is the difference between closingprices on a day-to-day basis. Figure 3 (right, bottom) showsthe absolute values of close-to-close price changes over the

same period. As would normally be the case, the differ-ences in closing prices are, on average, less than the dailyranges (and the same volatility spike occurs at bar 13). Asin Figure 2, the blue line is a three-period MA and the redline is the 10-period MA. Here, the averages tend to bebetween $1 and $1.50, except for the early observations inthe example.

To delve further into this aspect of the QQQs’ price

ACTIVE TRADER • October 2001 • www.activetradermag.com 14

continued on p. 15

The absolute value between closing prices indicates the closing differences tend to be between $1 and $1.50, exceptfor the earlier observations. The blue line is the three-periodmoving average of the bar-to-bar closing price differencesand the red line is the 10-period moving average.

FIGURE 3 CLOSE-TO-CLOSE CHANGES

5 . 0

4 . 5

4 . 0

3 . 5

3 . 0

2 . 5

2 . 0

1 . 5

1 . 0

. 5

0

Po i n t s

Price barsSource: Excel

A histogram of the daily ranges of the QQQs for the periodin Figure 1 shows the typical daily range is approximately$2. The blue line is a three-period moving average of thedaily ranges and the red line is a 10-period moving average.

FIGURE 2 DAILY RANGE

9

8

7

6

5

4

3

2

1

0

Po i n t s

Price bars

Source: Excel

The simplest of statisticscan provide a great deal of insight into a stock’sb e h a v i o r.

TRADING Strategies

behavior, we’ll display this data in a for-mat that makes it easier to see how thethe high, low and close relate to theopen. Figure 4 (right) shows each of thedaily open-high-low-close bars adjustedso the opening price is zero. The high ofeach bar re p resents the diff e re n c ebetween the high and the open; the lowof each bar is the difference between theopen and the low, and the dash on eachbar is the difference between the closeand the open.

The next step is to create two versionsof this chart to analyze the differencebetween bars that closed up from thep revious close and those that closeddown from the previous close. Figure 5(below) shows only bars that closed upfrom the previous day’s close. Figure 6(p. 62) shows those bars that closeddown from the previous day’s close.

F i g u re 5 reveals that, of days thatclosed higher than the previous day,a p p roximately one-third of the bars’lows were less than 50 cents below theopen, one-third were between 50 centsand $1 below the open, and one-third were between $1 and$1.50 below the open. Interestingly, a number of the bars fromthe final third closed below the open but made highs that wereapproximately 50 cents above the open for the day.

Figure 6 shows when the market closed down, the highstended to range between 50 cents and $1 above the open. Thebar with the highest high (nearly $2 above the opening price)

closed down for the day, but above the opening price — a pricepattern often referred to as a reversal day.

These charts reflect a bias for the low or the high dependingon if the market closes up or down for the day. On up-closingdays, the majority of the daily ranges are above the openingprice (with the highs often reaching $2 or more above the open-ing), while the lows are often between 50 cents and $1 below

the open. The opposite is true for down-closing days.

Now we’ll expand this kind of simplestatistical analysis to cover the periodfrom June 22, 1999, to June 13, 2001 (500days). During this period, the marketclosed up 254 days, closed down 238days and closed unchanged eight days.

Figure 7 (p. 62) sorts the daily rangesin 50-cent increments. (The 50-cent col-umn shows the number of days withranges up to and including 50 cents andso on.) There were zero days with a dailyrange of 50 cents or less; 13 days withranges of 51 cents to $1; 63 days withranges of $1.01 to $1.50; and 72 days withranges of $1.51 to $2. The daily range wasgreater than $10 just seven days. The typ-ical daily range was around $2, similar tothe results for the shorter period exam-ined in Figure 2.

Approaching the data from the oppo-site direction provides another perspec-

continued on p. 16

Only the bars from Figure 4 that closed up from the previous day’s close areshown here. Approximately a third of the bars had lows near 50 cents belowthe open, a third were between 50 cents and $1 below the open, and a thirdwere between $1 and $1.50 below the open.

FIGURE 5 CHARACTERISTICS OF UP-CLOSING DAYS

3 . 5

3 . 0

2 . 5

2 . 0

1 . 5

1 . 0

. 5

0

- . 5

- 1 . 0

- 1 . 5

- 2 . 0

- 2 . 5

- 3 . 0

- 3 . 5

Source: Excel

$7.28

In this chart the daily bars are adjusted so the opening price is zero. The high ofeach bar represents the difference between the high and the open; the low ofeach bar represents the difference between the open and the low; and the hashmark on each bar represents the difference between the close and the open.

FIGURE 4 DAILY BARS IN TERMS OF OPENING PRICE

8 . 07 . 57 . 06 . 56 . 05 . 55 . 04 . 54 . 03 . 53 . 02 . 52 . 01 . 51 . 0. 50

- . 5- 1 . 0- 1 . 5- 2 . 0- 2 . 5- 3 . 0- 3 . 5

Source: Excel

Price bars

Po i n t s

Price bars

Po i n t s


tive. For example, out of the 500 days, how many days did theQQQs have a range greater than $2? It turns out that there were352, or 70.4 percent of the total days.

Figure 8 (below, right) is a similar breakdown of the typicaldifference in closing prices, expressed as absolute values. Eightdays were unchanged from the previous close; 98 days closed

with a difference of up to +/-50 cents; 92days closed with a difference between+/-51 cents and +/-$1; 66 closed with ad i ff e rence between +/-$1.01 and +/-$1.50; and 54 closed with a differencebetween +/-$1.51 and +/-$2.

Only 182, or 36.4 percent, of the 500days the QQQs closed by more than +/-$2 from the previous close. The dataindicates that although 70 percent of thetime the QQQs have a daily rangegreater than $2, only 36 percent of thetime will they close more than $2 (high-er or lower) from the previous day’sclose.

As noted, 254 of the 500 days closedhigher than the previous day. Figure 9(p. 64) shows that for these days (usingabsolute values for the diff e re n c ebetween the open and the low) the open-ing price was equal to the low of the dayfive times; the difference between theopen and the low was 50 cents or less 77times; and the difference between the

open and the low was 51 cents to $1 73 times, which means themarket was down as much as 51 cents to $1 before reversing toclose up on the day. On only 32 days, or 12.5 percent of thetime, did the market fall by more than $2 and still close up forthe day.

Of the 238 days the market closed down, the open and the


continued on p. 17

There were zero days with a daily range of 50 cents or less;13 days with ranges greater than 50 cents up to and includ -ing $1; 63 occurrences with ranges greater than $1 up toand including $1.50; and 72 days where the daily range wasgreater than $1.50 up to and including $2. The daily rangewas greater than $2 approximately 70 percent of the time.

FIGURE 7 SORTING THE DAILY RANGES

8 0

7 0

6 0

5 0

4 0

3 0

2 0

1 0

0

Source: Excel

There were eight unchanged closing prices; 98 that closedbetween unchanged and +/-50 cents; 92 that closed greaterthan 50 cents and up to $1; 66 that closed greater than $1and up to $1.50, and 54 that closed greater than $1.50 andup to $2.

FIGURE 8 SORTING THE CLOSE-TO-CLOSE CHANGES

1 2 0

1 0 0

8 0

6 0

4 0

2 0

0

80 0 0 01

7855

1 92 02 2

3 83 6

5 75 5

6 97 2

6 3

37

13

9 89 2

6 6

5 4 5 2

3 22 4 2 2

1 51 0

5 5 7 4 1 2 0 1 1 1

Source: Excel

On days the market closed lower than the previous close, the highs tended torange between 50 cents and $1 above the open. The bar with the highest high(nearly $2 above the opening price) closed above the opening price despiteclosing lower than the previous close.

FIGURE 6 CHARACTERISTICS OF DOWN-CLOSING DAYS

3 . 5

3 . 0

2 . 5

2 . 0

1 . 5

1 . 0

. 5

0

- . 5

- 1 . 0

- 1 . 5

- 2 . 0

- 2 . 5

- 3 . 0

- 3 . 5

Source: Excel

Price bars

Po i n t s

Price bars Price bars

R a n g e R a n g e

high were the same only 10 times (see Figure 10, below,right). The market reached a high of between one cent to 50cents before turning down 84 times, and was up between 51cents to $1 65 times before reversing.

Figures 9 and 10 reveal there were some volatile days inthis period, in that five times the QQQs were down morethan $4.50 and still closed up, and three times were up morethan $4.50 and then closed down.

These price behavior characteristics provide practical trad-ing guidelines for systematic and discretionary traders alike— regardless of other analytical approaches they use.

For example, if you trade intraday bars and came into thetrading day with a trade that was up $2 for the day, youshould consider taking some partial profits becausealthough the QQQs exceed a $2 range 70 percent of the time,they close +/-$2 from the previous close only 36 percent ofthe time. As a result, the statistics suggest the market is like-ly to turn against this position by the close.

Or, perhaps your analysis indicates the QQQs have agood chance to rally, but the market starts the day with anearly sell-off, trading down from the opening price. The sta-tistics from Figure 9 (which showed the QQQs were downas much as 51 cents to $1 before reversing to close up on theday 73 times) suggest opportunity is there when the market

is down $1.If you are developing a short-term systematic approach

based upon indicators, you can substitute a hard-moneystop based on typical behavior of the market instead ofwaiting for an indicator to give a signal at the close.

For example, when reviewing your test results you mightuncover that if your system was long and the indicator sig-nals a reversal point that causes losses greater than $3 at theclose, you could substitute a hard-money exit rule based onthe statistics that show the market rarely recovers after ithas moved a certain amount intraday. In addition, knowingthe market’s typical behavior can help you better decidehow much capital to risk on any one trade.

In his book Trading in the Zone: Master the Market withConfidence, Discipline and a Winning Attitude , trader and authorMark Douglas writes, “The best traders treat trading like anumbers game, similar to the way in which casinos and pro-fessional gamblers approach gambling.”

To understand any market, you must measure and classify itsbehavior in terms of probabilities. You can combine these sta-tistics with other analytical techniques and build sound tradingmethods based on concrete price behavior characteristics.

This kind of analysis reveals each market has typical behav-ior patterns, as well as the potential to — on any given day —trade outside of its typical pattern. That is a fact of life in trad-ing, but by using simple statistics, you’ll be better equipped totrade effectively.Ý


Analyzing the absolute values of the differences betweenthe open and the low for days the market closed higherreveals that only five times was the opening price equal tothe low, and 77 times the difference between the open andthe low was 50 cents or less. The market recovered frombeing down between 51 cents and $1 before reversing toclose up for the day 73 times.

FIGURE 9 DAYS THE MARKET CLOSED UP

90

80

70

60

50

40

30

20

10

0

5

7 77 3

5 0

1 71 3

1 02 120 0

3 1

Source: Excel

On days the market closed down, the open was equal to thehigh only 10 times. The market turned down after reachinga high between one cent to 50 cents 84 times. The marketwas up between 51 cents and $1 65 times before reversing.

FIGURE 10 DAYS THE MARKET CLOSED DOWN

100

90

80

70

60

50

40

30

20

10

0

1 0

8 4

6 5

3 8

1 89

4 3 131 0 0 01 1

Source: Excel

To understand any market,you must measure and classify its behavior in terms of probabilities.

Price bars

R a n g e

Price bars

R a n g e

BY THOM HARTLE

W ith more than 10,000stocks to choose from,traders need to ensurethey are in the ones

that are moving. No one can afford to sitin markets that are going nowhere.

Money goes where opportunities lie,and traders and investors often defineopportunity and commit capital basedon either technical conditions, such ashigh price momentum, or fundamentalcharacteristics, such as strong earningsgrowth.

H o w e v e r, there’s no reason funda-mentals and technicals cannot comple-ment each other. In fact, you cani m p rove your chances of success byfocusing on stocks with both good tech-nical and fundamental characteristics.After all, a company’s ability to deliverpersistent positive business performanceshould translate into a chart with anu p w a rd, zigzagging price pattern —perfect for the nimble trader.

One easy place to get a list of stockswith strong technical and fundamental

characteristics is Investor’s Business Daily(IBD), a newspaper that publishes aweekly list of stocks with top fundamen-tals. While IBD does not provide the pre-cise criteria of the screening process usedto identify these stocks, they rank highin IBD’s SmartSelect Ratings.

Get SmartSelectAll IBD screens and ratings are basedupon a proprietary stock database. Forexample, its relative strength ratings com-p a re an individual stock’s price perform-ance to the IBD stock database, not to theS & P 500 or a particular sector, as is thecase with other services. Here is a briefexplanation of the IBD rankings, knownas the SmartSelect Ratings:

SmartSelect Composite Rating: Com-bines all five IBD SmartSelect Ratings.

Of the five, the earnings per share ratingand the relative strength rating receivethe most weight.

Earnings Per Share (EPS) Rating:Compares the last two quarters and thelast three to five years of earningsgrowth and stability to that of all othercompanies. A 90 rating indicates that acompany’s earnings growth outper-formed 90 percent of all other companiesin the database.

Relative Price Strength (RS) Rating: Astock’s relative price change in the last12 months vs. all other stocks.

Industry Group Relative PriceS t rength Rating: Compares a stock’sindustry price performance in the lastsix months to 196 other industries. Topindustries rate A+, the worst rate E.

Sales + Profit Margins + ROE Rating:A


TRADING Strategies

F u n d a m e n t a l l yT E C H N I C A L T R A D I N GTechnical and fundamental

analysis don’t have to be

mutually exclusive.

This trading approach

finds stocks with top

fundamentals and buys

them in uptrends when

they pull back.


company’s recent sales growth, profitmargin and return on equity (ROE) arecombined and graded from A to E.

Accumulation-Distribution Rating: Aprice and volume formula determines ifa stock is under accumulation (buying)or distribution (selling) in the last 13weeks. A grade of A indicates heavybuying and a grade of E reflects heavyselling.

Each week IBD singles out approxi-mately 20 to 30 stocks with top funda-mentals. These stocks are highlighted inyellow in the NYSE and Nasdaq MainTables. They provide a starting point foridentifying solid trading opportunities.The next step is to use specific technicalcriteria to narrow the list further andultimately signal trades.

The strategy that will be outlined inthe following sections will focus onstocks with strong fundamentals, andwill operate exclusively from the longside of the market.

Blending in the trend: Chaikin Money FlowIBD performs its screening of top funda-mental stocks on a weekly basis. Thismeans it is possible for a stock to switchf rom uptrend to downtrend based onsome event that won’t be picked up bythis screening process. (Later, we’ll lookat an example of a company that stayedon the top fundamental stocks list for twoweeks despite losing almost 30 percent ofits value.) A c c o rd i n g l y, it is necessary toconsult a technical trend indicator toidentify those stocks that are in uptrends.

Chaikin’s Money Flow (CMF) will beused to determine the trend for the trad-ing approach we will outline. The CMFis a ratio of weighted volume relative tototal volume over a 21-day period.

The numerator is the 21-day sum of

each day’s volume weighted by the clos-ing price relative to the day’s range. Thedenominator is the total volume over the21-day period. The CMF formula (inMetaStock language) is:

CMF = sum((((C-L)-(H-L)) / (H-L)) *V, 21) / sum(V, 21)

whereH = daily highL = daily lowC = daily closeV = volume

(Traders using eSignal 7.0 can find thecustom formula for the CMF atwww.activetradermag.com/code.htm.)

Positive CMF values indicate thestock is closing near the upper end of itsdaily range on higher volume, a sign ofpersistent buying. Negative CMF read-ings reflect the opposite.

As long as the CMF is positive, themarket is considered to be in an uptrend.The advantage of this indicator is itsfocus on volume relative to how themarket closes each day.A stock could bedown on the day because of selling pres-sure throughout the broader market, butstrong stocks should nevertheless tendto close near their highs despite overallmarket weakness.

At this point we have created a list offundamentally strong stocks and chosena tool that defines when they are inuptrends. Now we must decide the bestway to enter trades.

Buying low and selling highToday’s market requires a trader to buylow and sell high, which is very differentfrom a few years ago when a tradercould buy high and sell higher. As a

result, traders need an indicator thatidentifies relative low and high points ina market. Bollinger Bands fill this role inthe strategy.

Bollinger Bands are price envelopesconsisting of lines plotted two standarddeviations above and below a movingaverage (MA), typically a 20-period MA.Most of a market’s fluctuations will fallbetween the bands, so when price reach-es or exceeds one of the bands, the mar-ket, by definition, is at a relatively highor low level. (For more information onBollinger Bands, see the Active TraderInterview with John Bollinger in theJanuary 2001 issue.)

For example, when a market is at thelower Bollinger Band, it is nearing anextreme downside level relative to itsrecent price history. If price is at theupper Bollinger Band, it is reaching anextreme upside level based on recentprice history. An important caveat is thata simple tag of the upper or lower banddoes not guarantee price will reverse. Astrongly uptrending stock will persist-ently tag the upper band, while a stockin a strong downtrend will repeatedlyhit the lower band.

In this case, the Bollinger Bands willbe set to a lookback period of 10 daysinstead of the conventional 20 days.

Trade rulesThe strategy is designed to go long onshort-term countertrend moves thatpunctuate the longer-term uptrend. Therules for the strategy are:

1. Identify those IBD top fundamental stocks that also have positive CMF readings. These stocks also must be trading above $20.

2. Wait for price to tag the lower Bollinger Band.

3. Go long if the stock trades 5 cents above the high of the daily bar that hit the lower Bollinger band. If price does not trade 5 cents above the high of the tag bar, then go long 5 cents above the high of the next bar.

4. Place a stop loss 5 cents below the low of the entry bar.

5. Take profits if the stock moves one dollar in your favor.

Money goes where opportunities lie, andtraders and investors often define opportunityand commit capital based on either technicalconditions, such as high price momentum, or fundamental characteristics, such as strong earnings growth.

continued on p. 20

If the CMF is not positive on the barthat tagged the lower Bollinger Band butturns positive on the next bar, use thehigh of the second bar. (If the CMF is

negative on the first bar that tags thelower Bollinger Band, we cannot use thehigh of that bar; if the CMF is negative onthe bar following the tag bar, we still can-

not go long.) Occasionally the CMF willmomentarily edge into negative territo-ry when the price is at a lower extre m e .Do not enter the market after the secondb a r. Stocks also must be trading above$20 because the one-dollar movere q u i red for the profit target is less like-ly to occur in lower-priced stocks.

Trade examplesThe following trades are taken fromsome of the 17 top fundamental Nasdaqstocks highlighted by IBD on Feb. 2,2002. When trading this strategy, youcan use the list for trading during themonth, checking each week to makesure your stocks remain on the list.

F i g u re 1 (top left) shows A l l i a n c eGaming Corp (ALLY), one of the topfundamental stocks as of Feb. 1, 2002.The entry criteria were met on Feb. 8:The previous bar had tagged the lowerBollinger Band, the CMF was positive,and the stock traded above the high ofthe tag bar ($29.85). On Feb. 11, the stockreached a high of $31.44, exceeding theone-dollar profit target.

Then on Feb. 20, the stock touched thelower Bollinger Band a second time andthe CMF was positive. The next day wewent long 5 cents above the previousday’s high of $29.67. The stock rallied to$30.95 the same day.

Figure 2 (bottom left) shows NvidiaCorp. (NVDA). In early February, thestock tagged the lower band but theCMF was negative at the time. The CMFturned positive on the next bar, but thestock did not trade above the high of thisbar; no trade was entered. On Feb. 8 thestock tagged the lower band once more,but the CMF had turned negative again;no trade. The next day the CMF turnedpositive and we went long on Feb. 12when the stock pushed above the Feb. 11high of 61.00 by 5 cents. The stockreached 63.35 the same day, fulfilling theprice target. Later in the month, the CMFwas negative when the stock tagged thelower band (the stock was in a visibledowntrend at this point) and no tradewas entered.

In Figure 3 (opposite page, top), pricetagged the lower band on Feb. 7, but didnot trade above the high of the tag baron Feb. 8. As a result, the Feb. 8 high(57.75), plus 5 cents, became the refer-ence point to enter a trade on Feb. 11. Wewent long on Feb. 11, and the stock hit59.96 that day (more than a $2 move),


On Feb. 5 the stock hit the lower band, but no trade was entered because theCMF was negative. The next day the CMF turned positive, but price did nottrade above the previous high, again denying entry. On Feb. 8 the stocktagged the lower band but the CMF was negative. However, the next day (Feb.12) the CMF turned positive and a long trade was triggered when the stockexceeded the Feb. 11 high by 5 cents.

FIGURE 2 DELAYED ENTRY

Source: eSignal 7.0

70.00

65.00

60.00

55.00

50.00

0

Buy

No entry

Sell

Chaikin Money Flow

Price Nvidia Corp (NVDA), daily

28 4 11 19 25 4February March

The stock tagged the lower Bollinger Band on Feb. 7 while the CMF was positive. A long trade was entered on Feb. 8 when the stock moved 5 centsabove the Feb. 7 high. The price target was hit the next day. Later in themonth another setup formed and the entry and exit occurred on the same day.

FIGURE 1 CAPTURING QUICK PROFITS

Source: eSignal 7.0

35.00

32.50

30.00

27.50

0.25

0

Buy Buy

Sell Sell

Chaikin Money Flow

Price Alliance Gaming Corp (ALLY), daily


fulfilling the profit target. Also noticethat in this case, the low was equal to theprevious day’s low, so the position wasnot stopped out.

On Feb. 20 price tagged the lowerband but the market fell for two moredays. Another tag occurred on Feb. 22,and the high of the day was 52.50. OnFeb. 25 the stock rallied above the triggerlevel, and on Feb. 26 it made a high of53.57. (If you experienced slippage onyour entry, you may have had to wait forthe full one-dollar profit.)

This example shows why you shouldtake the profit at the target level, despiteyour entry price: The same day the stockdropped below the entry bar’s low of51.50, which would have resulted in aloss of more than $1.

The rest of the storyEarlier we mentioned that just because astock is on the top fundamentals listdoesn’t mean it’s in an uptrend at thetime. This final example is from January2002, and it shows how market partici-pants can turn on a stock with avengeance.

On Jan. 2, one company on the topfundamental list was Dynacq Interna-tional (DYII). Figure 4 (bottom right)shows the stock steadily climbed from alow of 11.00 in September 2001 to 29.25on Jan. 7; its PE pushed above 35. Inearly January, IBD ratings for DYII were:

SmartSelect Composite Rating: 98Earnings Per Share Rating: 99Relative Price Strength Rating: 95Industry Group Relative Price Strength

Rating: B+Sales + Profit Margins + ROE Rating: AAccumulation/Distribution Rating: B

However, two days after announcingearnings of 17 cents per share (vs. ana-lysts’ expectations of 14 cents per share)on Monday, Jan. 14, an article was pub-lished questioning the company’s ac-counting practices. The stock closeddown nearly 7.00 — a 28-percent loss.However, Dynacq was still listed as a topfundamental stock as of Jan. 25 becausethe IBD screenings do not take intoaccount sudden changes in investor per-ception.

Fundamental screening, technical tradingThe stocks highlighted as having topfundamentals by IBD offer many oppor-

tunities for nimble traders. You simplyneed a solid set of procedures for defin-ing entry, risk and profit-taking.

The set described here that incorpo-rates the CMF and Bollinger Bands can be

a departure point for further testing, andcan be modified to suit your own person-al risk tolerance and profit goals.Ý


A stock can get crushed despite having top fundamentals. Two days after a positive earnings announcement, a published article raised concerns regarding accounting practices at the company, and the stock closed downnearly 28 percent.

FIGURE 4 FUNDAMENTALS DON’T TELL THE WHOLE STORY

Source: eSignal 7.0

25.00

20.00

15.00

10.00

5.00

0.25

0

-0.25

Negative article

Chaikin Money Flow

Price Dynacq International (DYII), daily

28 1 8 15 22 29 5 12 19 26 3 10 17 24 31 7 14 22 28 4October November December January 2002 February

The trade on Feb. 25 illustrates that, regardless of your entry price, it’s wiseto exit at the $1 profit target. On the same day the trade was entered, priceeventually dropped below the bar’s low of $51.50, which would have resultedin a loss larger than $1.

FIGURE 3 TAKE PROFITS WHEN YOU CAN

Source: eSignal 7.0

65.00

60.00

55.00

50.00

0.2

0.1

0

Buy

Buy

Sell

Sell

Chaikin Money Flow

Price Panera Bread Co. (PNRA), daily


22 www.activetradermag.com • September 2002 • ACTIVE TRADER

TRADING Strategies

here will always be a select group of traders who know just

when the market is hitting its low for the day, enter a long posi-

tion and get out near the high. These traders share a keen, intu-

itive sense of price direction.

Is such talent mandatory for trading success? Not at all. Those of us without

an innate feel for the markets can perform simple analysis that allows us to

understand the typical price behavior of our favorite stocks and futures. We can

then incorporate this valuable information into our strategies.

We’ll look at some unique ways to observe and quantify market behavior,

and illustrate how to use this information in actual trading. The instrument

we will analyze is the S&P500 index tracking stock (SPY), but you can per-

form the same analysis on any market.

The goal in examining the price data is to find patterns in the daily

ranges, close-to-close ranges, and the relationship between the open and the

low (for days the market closes up) or the open and the high (for days the mar-

ket closes down). Analyzing simple price patterns makes it possible to deter-

mine the odds that different kinds of price moves will occur.

As a result, the characteristics revealed in this analysis can be used to tailor

trading strategies to the realities of the markets you trade, resulting in more dis-

ciplined and probability-based trading — in essence, a mechanical way of

achieving a measure of the “intuition” enjoyed by that select group of traders.

ACTIVE TRADER • September 2002 • www.activetradermag.com 23

continued on p. 24

Getting to know a market means

understanding the odds associated

with the price moves it typically makes.

A handful of simple calculations can put

these probabilities at your fingertips.

BY THOM HARTLE

Daily rangeThe first aspect of price behavior to con-sider is the average daily range, which

functions as a basic measure of volatility— the larger the range, the higher thev o l a t i l i t y. Figure 1 (above) shows 60

trading days of data (lateFebruary to mid-May 2002)in SPY. The stock switchesfrom uptrend to downtrendand back again.

F i g u re 2 (left, bottom)shows the daily range (highminus low) of each bar fromFigure 1. The blue line is athree-bar moving average ofthe daily ranges and the redline is a 10-bar moving aver-age. Comparing this chart toFigure 1 reveals the highestvolatility occurred whenSPY was bottoming out inMay. Although there is notenough data to draw a firmconclusion, this does sup-port the trading adage thatmarket turning points —especially bottoms — aremore volatile than trendingperiods in terms of dailytrading ranges.

Therefore, if some of yourother indicators were hint-ing at a top or bottom (suchas a divergence betweenprice and an oscillator), you

could refer to these simple moving aver-ages of the daily range as one more con-firmation of a pending reversal.

In addition, the dailyrange information is vitalfor intraday traders — thosewho close out all open posi-tions by the end of eachtrading session. By knowingthe current average rangeyou can set targets based onhow much the market youtrade typically moves in aday. Also, intraday tradersmust be able to identifymarkets with enough intra-day volatility to make themworthwhile to trade.

Close-to-close rangeThe second measure we’llanalyze is the diff e re n c ebetween closing prices on aday-to-day basis. We will useabsolute values (i.e., all posi-tive numbers) of the close-to-close price moves to makethe analysis comparable for


Arranging the ranges (high-low) of the days in the test period highlights the market’svolatility characteristics. The high-volatility area corresponds to the May bottom.

FIGURE 2 DAILY RANGE

S&P 500 index tracking stock (SPY), daily118.00

117.00

116.00

115.00

114.00

113.00

112.00

111.00

110.00

109.00

108.00

107.00

106.00

105.00

3.0

2.5

2.0

1.5

1.0

0.5

0

March April May25 1 11 18 25 1 8 15 22 29 1 6 13 20

The initial test period for SPY contained both uptrending and downtrending conditions.

FIGURE 1 SPOTLIGHT ON SPY

Source: CQGNet

Source: Excel, data by eSignal

Day

Three-bar moving average

10-bar moving average

High-volatility area

up and down markets.Figure 3 (right) shows the

d i ff e rence between closesfor the bars in Figure 1, witht h ree- and 10-bar movingaverages. Note that there isan outlier (an extreme read-ing) at data point 48 — aclosing difference of $3.91.This reading is much largerthan the same bar fro mFigure 2 because of the lowclose on May 7 and a gap-up opening and high closeon May 8. Although a largerprice gap occurred a fewbars later, the close-to-closemove was not as exaggerat-ed because the bar beforethe gap closed near its high.

Some traders may wantto use average “true range”(the greatest absolute differ-ence between today’s highand today’s low, today’s high and yes-terday’s close, or today’s low and yester-day’s close) for this calculation.However, by choosing close-to-close andhigh-minus-low ranges, you have statis-tical references to know when marketactivity is significant relative to whatmany other people in the market aretalking and thinking about. After all, the

financial press never says the marketclosed up today twice the average truerange, but they do comment on closingdifferences — e.g., today was the highestclose in three weeks, or at one point thestock was down $6 from yesterday’sclose, and so on. With historical num-bers in hand you can put price action inperspective and separate facts fro mhyperbole.

A different perspectiveA different way of viewing this priceinformation is shown in Figure 4 (left).The price data is adjusted so that eachbar’s opening price is zero, the high isthe net gain from the open, the low is thenet decline from the open, and the closeis the difference from the open. This per-spective highlights how much the mar-

ket moved above or belowthe open on a given day.

The next step is to segre-gate the bars in Figure 4 interms of days the marketclosed up from the previousclose (Figure 5, p. 26) anddays the market closeddown from the pre v i o u sclose (Figure 6, p. 26). Thisanalysis will reveal whetherthe market has any observ-able tendencies on up-clos-ing or down-closing days.

During this 60-day peri-od, the market closed up 28bars and down 32 bars.(Keep in mind that somebars closed below the openbut above the previous closeor above the open but belowthe previous close.) As youmight expect, the bulk of theintraday price movement is


continued on p. 26

Measuring the distance between closing prices provides a second perspective on marketvolatility. Close-to-close moves reflect price gaps between bars.

FIGURE 3 CLOSE-TO-CLOSE RANGE

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0

This graph shows the daily ranges plotted with the opening price at zero. This highlightshow far above or below the open the market moved and where it closed on different days.

FIGURE 4 ADJUSTING THE DATA

3.0

2.5

2.0

1.5

1.0

0.5

0

-0.5

-1.0

-1.5

-2.0

-2.5

-3.0



Day

Day

above the zero axis (i.e.,higher than the open) on up-closing days and below thezero axis (i.e., lower than theopen) for down-closingdays.

There were four instanceswhere price was up morethan $1 above the open andclosed down (that is, closedbelow the previous close),and five instances where theprice was down $1 or morefrom the open and closedup. In addition, there were13 instances (46 percent) inFigure 5 where the price was50 cents or more below theopen, and still rallied toclose up for the day.

On the other hand, SPYmoved more than 50 centsabove the open and closed

down for the day only nine times in 32observations (28 percent). Also note thatSPYmade no unusual surges or declinesduring this period. However, there areonly 60 bars of data, so a longer-termview would be required before readingtoo much into these statistics.

Getting more in depthTo get a more reliable picture of theprice behavior in this market, we will

expand the data period to500 daily price bars (May25, 2000, to May 24, 2002)and perform more sophisti-cated analysis. During thisperiod, the market closedabove the previous close245 times and below it 255times. The average dailyrange was $1.68, which is inline with the data fro mFigure 2.

Figure 7 (opposite page)analyzes the data by group-ing the daily ranges by sizein 50-cent increments (i.e., apenny to 50 cents, 51 centsto $1, $1.01 to $1.50, etc.)and counting the number ofdays that fell into eachgroup. The daily range was$1 or less on 17 days, $1.01to $1.50 on 118 days, and$1.51 to $2 on 135 days (the


Those of us without an innate feel for the markets can perform simpleanalysis that allows us to understand the typical price behavior of the instruments we trade.

These are the days from Figure 4 that closed up from the previous close. Isolating thesedays allows you to identify any characteristics specific to up-closing days.

FIGURE 5 UP-DAY ANALYSIS

3.53.02.52.01.51.00.5

0-0.5-1.0-1.5-2.0-2.5-3.0-3.5

Isolating the down-closing days helps highlight the price behavior of these bars.

FIGURE 6 DOWN-DAY ANALYSIS


3.53.02.52.01.51.00.5

0-0.5-1.0-1.5-2.0-2.5-3.0-3.5


Day

Day


highest total). Reading the chart from right to left,

the SPY traded in a range greater than$2.50 on 128 days (only 25.6 percent ofthe time). The most extreme reading fora daily range, between $8 and $8.50,occurred when the Fed cut interest rateson Jan. 3, 2001.

Figure 8 (p. 28) sorts the close-to-closeprice moves. The average closing changewas $1.27. The largest number of close-to-close ranges was less than 50 cents.The highest closing diff e rence wasbetween $6.01 and $6.50. The SPYclosedwith a change greater than $2.50 only 12percent of the time.

Highs and lows analysisFigures 9 and 10 break down the datainto days the close was up from the pre-vious close and days the close was downf rom the previous close, re s p e c t i v e l y.The data is divided into 25-cent incre-ments.

F i g u re 9 (p. 28) compares the(absolute) difference between the openand the low on days the market closedup from its previous close. On eightoccasions, the open was the low. Fifty-t h ree times the market traded some-

where between a penny and 25 centsbelow the open and closed up for theday, while the market traded between 51cents and $1 below the low and stillclosed up for the day 66 times. The mar-ket traded more than $1.50 below theopen and still closed up for the day bare-ly more than 5 percent of the time. (Keepin mind the open can be up or down forthe day, and the close is compared to the

previous close. In other words, the mar-ket could open up $1, trade $2 downfrom the open and finish 50 cents upfrom the previous close for the day.)

This data tells us a few things abouttrading a long position. First, unlessthere is incredibly positive news, there’sno need to go long on the open, becausethe market closed up only eight times (3percent) on days the open was also the

low of the day.Also, you can use this

information to decidewhere to place a stop if youstarted the day with anopen long position. In thiscase, if you are holdingpositions overnight withthe goal of catching a one-to three-day price swing,you cannot place your stop25 cents or less below theopen if you are long. Doingso is inviting the market tostop you out, because SPYtraded less than 25 centsaway from the open only 25percent of the time. In otherwords, the odds you will bestopped out are 75 percent.

Here’s another scenario:If you are an intraday trad-er and your system is flash-ing a buy signal when priceis down $1.50 from theopen (remember just slight-ly more than 5 percent ofthe time the SPY t r a d e d

higher after being down $1.50 from theopen) you may want to pass on thetrade.

Now let’s look at the statistics fordown-closing days. Figure 10 (p. 29)examines the diff e rence between theopen and the high on days the marketclosed down from its previous close. Theopen and the high were the same 16days, and there were 67 days when the

market opened, traded up between apenny and 25 cents, and then declined.

This analysis can help determinewhere to place a buy stop if short fromthe day before. First, it appears thatbecause of the bear market, stops can beplaced close to the open: The SPYtradedless than 25 cents above the open whilestill closing down 32 percent of the time.

Breaking down the daily ranges into 50-cent increments is the first step toward determin -ing the probabilities of different kinds of price moves. The daily range for SPY most oftenfell between 1.5 and 2.99.

FIGURE 7 DAILY RANGES

Analyzing basic price relationships makesit possible to determine the odds that different price moves will occur.

140

120

100

80

60

40

20

00 0 0 101 13 2

8 917

118

135

103

57

30

17

Daily range ($)


continued on p. 28

(Keep in mind, though,that’s not enough ro o mbecause there’s a 68-percentchance of being stoppedout. Nonetheless, thisshows the effect of the bearmarket: The market tradedup more than 75 cents andstill closed down only 46percent of the time.

If you are an intradaytrader and your system isflashing a sell signal whenprice is up $1 from the open,remember that the SPYtraded $1 higher and stillclosed down only 32 per-cent of the time over thepast year. However, as is thecase with all these statistics,32 percent is simply a num-ber, and it should not beused as the sole basis of atrade; it should complementsome other form of techni-cal analysis and preferably aback-tested strategy.

This type of analysis canbe incorporated with other technical sig-nals to set profit targets and improve

other aspects of money-management.For example, you could plan to take par-

tial profits once the market hits the aver-age daily range.

Finally, Figure 10 showsthere were two occurrencesof the high being morethan $5 above the open ondown-closing days. Thesetwo occurrences were inSeptember 2001 when theSPY opened sharply lower,re c o v e red from thee x t remely weak openingprice, but still closed downfor the day (more than$5.75 once and $1.43 theother time).

Putting the numbers to workCompiling the informationf rom this analysis givesyou a comprehensive anddetailed picture of a mar-ket. The characteristicssuch analysis reveals abouta market can be of greathelp in determining logicalstop levels and profit tar-gets, especially for short-term and intraday tradersas mentioned earlier. In


continued on p.29

The majority of close-to-close moves were between 50 cents and $1.99 (the first threebars). This information, along with that from Figure 7, gives you an indication of thesize price move you can expect over different periods.

FIGURE 8 CLOSE-TO-CLOSE BREAKDOWN

140

120

100

80

60

40

20

00 011

5 2 2

158

113

134

95

4653

25

Measuring how far the market drops below the low and then closes above the previousclose indicates the odds of a higher close fall significantly when the down move is morethan 1.99 points.

FIGURE 9 OPEN-TO-LOW MOVES ON UP-CLOSING DAYS

70

60

50

40

30

20

10

0

8

120

2 20

42

66

53

3533

2017

Close-to-close move ($)

Open-to-low move ($)




addition, some of the num-bers highlight the impactof the bear market, and onenumber is somewhat sur-prising.

Despite the bear marketof the last two years, thenumber of up closes todown closes is not that dif-f e rent (245 vs. 255).Obviously, if we measuredthe net difference in clos-ing prices, the re s u l t swould be greater for downdays.

Another bear marketphenomenon is there wereonly eight up-closing daysfor which the open was thelow price; for down-clos-ing days the open was thehigh price 16 times. Inaddition, a hump is evi-dent at the 26- to 50-centpoint in Figure 9, while

Figure 10 shows a steadily decliningseries of lower highs. In other words,even though the market did close up forthe day, there was enough selling pres-sure to push prices well below the open,which places the peak in Figure 9 to theright compared to Figure 10. Both thesephenomena could be attributed to thebear market.

Analyzing a bull market period ofsimilar length might make the differ-ences between bull and bear marketsm o re evident, and would possiblyenable you to identify a bull market or abear market earlier than some othertechnique, such as the rising or falling200-day moving average. For example, ashift in the peaks for the high-low analy-sis in Figures 9 and 10 could be subtleevidence of a change in the underlyingtrend. Another idea is to track the num-ber of occurrences when the open iseither the low or the high of the day.

From subjective to objectiveHigh and low are usually relative termsin trading, because markets almostalways trade higher or lower than where

they are now. Gauging the typical pricemovement gives you a set of practicalreferences for what is high and what islow in the markets you trade, so youdon’t have to guess. And anytime youcan bring descriptive analysis thatmatches real-world experience, youimprove your chances for success.

By incorporating typical price behav-ior analysis into your decision-makingprocess, you move away from tradingdecisions based upon greed and fearand move toward objective, probability-based trading.

You will also develop a big-pictureview of your trading, because you willknow the characteristics of the marketsyou trade and how and why yourapproach works, and that any one tradecan occur outside the typical behavior ofthe market.

Both steps — reducing emotion andlooking at trades as a group of decisionsrather than isolated events — lead tomore effective trading.Ý

Similar to Figure 9, this graph organizes open-to-high moves on days the market closes below the previous close.

FIGURE 10 OPEN-TO-HIGH MOVE

70

60

50

40

30

20

10

0

16

0 0 00 111 2 1 13 2

54

67

37

32

23

14

The data tells usthat unless there’sincredibly positivenews, there’s noneed to go long SPYon the open,because the marketclosed up only eighttimes (3 percent)on days the openwas also the low ofthe day.



BY THOM HARTLE

T h e re’s no doubt tradersunwilling to sell short are at adisadvantage in a bear mar-ket. But even traders who

a ren’t biased against short-selling some-times operate under the misconceptionthat bear markets and down moves aresimply inversions of bull markets and upmoves. As a result, they think tradingf rom the short side is just a matter ofreversing the rules of a long-side strategy.

This is only true in the broadest sense,however, because bull and bear marketshave notable differences. One oft-repeat-ed example is that sell-offs tend to besharper and quicker than rallies (moreon this later). But there’s still more toprofiting on the short side than just tak-ing profits more quickly than you wouldon a long trade. As noted in “Losingyour shorts” (Active Trader, September2002, p. 56), even when the prevailingtrend is down, it’s often more difficult tomake consistent money selling shortthan it would be going long if the condi-tions were reversed.

To improve your short-selling acu-men, it’s a good idea to perform in-

depth analysis of the markets you trade,compare their bull and bear characteris-tics, and design and implement yourtrading strategies accordingly.

Research refresher“Familiarity breeds profitability” (ActiveTrader, September 2002, p. 38) discussedthe typical price behavior of the S&P 500depository receipts (SPY) using dailyopen-high-low-close bars, with the goalof identifying characteristics that couldserve as the basis for trading strategies.

Analysis of the two years from May25, 2000, to May 24, 2002 (which beganwith a closing price of $140.25 and fin-ished with a closing price of $108.69),suggested some of the test results were

skewed because of the bear market con-ditions that prevailed during this period.

To see if the results were different dur-ing a bullish phase, the same analysiswas performed on daily SPY price datafrom June 1, 1998, to May 23, 2000. At thebeginning of this bull period the SPYclosed at $109.53; the last close was$138.00.

We’ll compare the results of the twoperiods, which we’ll refer to as “bear”and “bull” market results, re s p e c t i v e l y.F i g u re 1 (opposite page, top) shows thep revailing downtrend of the bear marketperiod and Figure 2 (opposite page, bot-tom) shows the uptrending conditionsthat characterized the bull market period.

30 www.activetradermag.com • November 2002 • ACTIVE TRADER

B U L L v s . B E A R : The details matterMarket pundits often quip bull markets are different than bear markets.

But other than one goes up and the other goes down, what does the distinction

really mean? This comparison of bull- and bear-market characteristics provides

concrete statistics upon which to base upside and downside trading strategies.

continued on p. 32

TRADING Strategies

ACTIVE TRADER • November 2002 • www.activetradermag.com 31

From May 25, 2000, to May 24, 2002, the S&P tracking stock fell 22.5 percent. Even so, this period was punctuated bysubstantial counter-rallies.

FIGURE 1 BEAR MARKET

150.00

145.00

140.00

135.00

130.00

125.00

120.00

115.00

110.00

105.00

100.00

95.00

90.00July Aug. Sept. Oct. Nov. Dec. 2001 Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. 2002 Feb. Mar. Apr. May

Source: eSignal

S&P Index Trust (SPY), daily

The S&P tracking stock gained 26 percent from June 1, 1998, to May 23, 2000. Comparing the typical price behavior ofthis period to that of the subsequent bear period, it’s possible to determine if and how trading strategies should beadjusted for up- and downtrending conditions.

FIGURE 2 BULL MARKET

S&P Index Trust (SPY), daily

Source: eSignal

155.00

150.00

145.00

140.00

135.00

130.00

125.00

120.00

115.00

110.00

105.00

100.00

95.00

July Aug. Sept. Oct. Nov. Dec. 1999 Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. 2000 Feb. Mar. Apr. May

We’ll begin by compar-ing the differences in dailyclosing prices for the twoperiods.

Differences from c l o s e - t o - c l o s eF i g u re 3 (left) shows thedistribution of absolute val-ues for the day-to-daychanges in closing pricesduring the bear marketperiod. Figure 4 (bottomleft) provides the sameinformation from the bullmarket period.

The bear period close-to-close changes peak in the$0.01 to $0.50 range, whilethe bull market peak is the$0.51 to $1.00 range. Theaverage change during thebear market period was$1.27, vs. $1.39 during thebull period.

Apparently, when thereis greed involved, tradersare willing to continue topay up until the close.Market lore holds that abear market falls faster thana bull market rises, but inthis case, it appears peopledon’t like to carry shortpositions overnight, sothere may be a tendency tocover short positions goinginto the close despite a pre-vailing bear market.

F i g u res 5 (opposite page,top) and 6 (opposite page,bottom) compare the dailyranges (high minus low) ofthe two periods. The similar-ities between the two chartsa re somewhat surprising.The one diff e rence is the bullmarket did have one daywith a range between $0.01and $0.50. But despite thel a rge upswing and down-swing, the typical dailyranges did not vary.

Figures 7 and 8 (p. 34) arethe open-to-low moves fordays the SPY closed up.


Interestingly, the close-to-close price moves in the bull period are similar to those of thebear period, except that in this case the most moves were in the $0.51 to $1 range.

FIGURE 4 BULL MARKET CLOSE-TO-CLOSE BREAKDOWN

140

120

100

80

60

40

20

0

5

113

1 110 14 27

1

44

88

117

62

29

16

8



The majority of close-to-close moves were between 1 cent and $1.50 (the first threebars). This information, along with that from Figure 5, gives an indication of the sizeprice move you can expect over different periods of time.

FIGURE 3 BEAR MARKET CLOSE-TO-CLOSE BREAKDOWN

140

120

100

80

60

40

20

00 011

5 2 2

158

113

134

95

4653

25



H e re, the diff e re n c ebetween the bear marketand the bull market is moreapparent. First, there wereonly eight days during thebear market for which theopen was the low for theday, whereas the bull mar-ket had 24 days.

Next, the diff e re n c ebetween the open and thelow during the bear marketpeaked in the $0.26 to $0.50range; during the bull mar-ket it peaked between $0.01and $0.25. This indicatestraders aggressively buyright after the open in a bullmarket, thereby supportingprices, while during bearmarkets they engage inmore selling that drives thelow further down from theopening before prices rallyto close up on the day.

Figures 9 and 10 (p. 35)are the open-to-high movesfor days the SPY c l o s e ddown. Again, the differencebetween the two periods isvisible. There were 16 daysduring the bear period forwhich the open was thehigh for the day, comparedto just six for the bull peri-od. In other words, duringa bull market price stilltends to advance after theopen, even on days thateventually close down.

The bear market open-to-high difference peak was 67days for the range between$0.01 and $0.25. The bullmarket numbers, on theother hand, were shifted tothe left, peaking at 53occurrences for the $0.51 to$0.75 range. Again, it is notsurprising that during abull market buyers wouldtry to push prices highereven when the market ulti-mately closed down on anyone day.


continued on p. 34

Breaking down the daily ranges into 50-cent increments indicates, along with the close-to-close moves, how much the market is likely to move from day to day. The daily range forSPY usually fell between $1.01 and $2.50 in the bear market period.

FIGURE 5 BEAR MARKET DAILY RANGES

140

120

100

80

60

40

20

00 0 0 101 13 2

8 917

118

135

103

57

30

17

Daily range ($)


There is virtually no difference in the daily ranges for the bear and bull markets, whichis somewhat surprising.

FIGURE 6 BULL MARKET DAILY RANGES

160

140

120

100

80

60

40

20

00 11 003 2 0

168

118

1

16

135

103

9

58

36

Daily range ($)


What’s it all mean?It is clear that entry, riskmanagement and pro f i t -taking tactics need to takeinto account the markettrend, bear or bull. A tradercannot use the same buysetup and risk points thatwere effective during thebull market run becausethe behavior of the marketchanges during bear phas-es. Consequently, it’s pru-dent to spend time devel-oping a sound trend indi-c a t o r, and then designtrade setups based on thetrend.

Many new traders getinto trouble by ignoringthese realities. Here’s howthe typical trader learningcurve might progress: Firstyou learn you have to cutyour losses and let yourprofits run. If you devel-oped your trading tech-nique during the 1990s bullmarket, you could havedesigned a viable systemthat used tight stopsbecause, as Figure 8 shows,in the bull period SPY typ-ically traded down in theneighborhood of only $0.25to $0.75 from the openingprice, and still closed up.

However, if you failed torecognize the market’s per-sonality change during adowntrend, you may havebought after the open(when the market wasshowing initial weakness),and felt confident if themarket subsequentlymoved into positive terri-tory. But during a down-t rend (characterized bypersistent down closes), itis unlikely the market willclimb well into positive ter-ritory, as Figure 9 shows:


The bull market had most of its open-low moves for up days in the $0.01 to $0.25 range,as opposed to $0.26 to $0.50 for the bear market. This indicates traders are more likelyto support prices in a bull market.

FIGURE 8 BULL MARKET OPEN-TO-LOW MOVES ON UP-CLOSING DAYS

55

50

45

40

35

30

25

20

15

10

5

0

24

11 20 0 0 0 0 0

5 53

50

42 42

3633

18

8



This summary of how far the market drops below the low and then closes above the previous close indicates the odds of a higher close fall noticeably when the down move ismore than 50 cents.

FIGURE 7 BEAR MARKET OPEN-TO-LOW MOVES ON UP-CLOSING DAYS

70

60

50

40

30

20

10

0

8

120

2 20

42

66

53

3533

2017



There were only 67 peakopportunities, and the highwas less than $0.25 abovethe open. If you consistent-ly tried to buy post-openingweakness in this kind ofe n v i ronment, you wouldslowly bleed all of yourcapital by taking small lossafter small loss.

The trend is your friendif you know how to takeadvantage of it by usingthese kinds of marketobservations. If your indi-cator signals the trend isdown, you can use the typ-ical bear-market behavioras a starting point for deter-mining entry and exit rules— e.g., going short the SPYif the market is up $0.25 to$0.50 from the open, withyour stop a little higherthan $1 above the open.Develop separate rules foruptrend market phases.

No getting around the workYou may be thinking:Rules, rules, rules. Isn‘tthere an art to trading? Yesand no. Intuitively orientedtraders probably have akind of mental spreadsheet,and the profits to prove it.

If your profits are notmeeting your expectations,take the time to quantifythe markets, as well as yourtrading approach. You willquickly see the relationshipbetween your actions andthe market’s, the results ofthis coming in the form of aseries of losing or profitabletrades.Ý


This graph organizes open-to-high moves on days the market closed below the previous close. In the bear market period, the most notable concentration of days had open-to-high moves of less than 50 cents.

FIGURE 9 BEAR MARKET OPEN-TO-HIGH MOVES FOR DOWN CLOSES

70

60

50

40

30

20

10

0

16

0 0 00 111 2 1 13 2

54

67

37

32

23

14

Open-to-high move ($)


When SPY was down for the day, the open was the high only six times, and the marketoften rallied as much as a dollar before closing lower. This shows, in a bull market,price still tends to advance on down days.

FIGURE 10 BULL MARKET OPEN-TO-HIGH MOVES FOR DOWN CLOSES

60

50

40

30

20

10

0

6

0 0 0 0 1 120 0

2 1

8 8

4544

53

31

14

9

Open-to-high move ($)


BY THOM HARTLE

F or traders not blessed with exceptionalmarket intuition, the best chance of sus-tained profitability is to trade a system — aset of fixed procedures to ensure they act

on market opportunities consistently and effectively.In addition to entry and exit rules, though, a tradingsystem also must have a risk-management componentthat guards against losing a large sum of money onany one trade.

The mental challenge of following a system is giv-ing up control of your trading. On the one hand,y o u ’ re faced with the possibility of going bro k ebecause your system suffers a drawdown that is longer anddeeper than its hypothetical drawdowns in historical testing orpaper trading. On the other hand, there is the risk of missingtrade opportunities if you do not follow your system to the let-ter, even when it is in an extreme drawdown.

When should you continue to trade a system, and whenshould you stop taking its signals? The following ideas aretechniques based on analyzing past system returns thatattempt to balance the fear of losing money with the fear ofmissing trading opportunities.

Watching the equity lineOne way to manage the trade-or-not-trade dilemma is to cal-culate a moving average of your system’s equity line (the run-ning total of its dollar gains and losses) to determine when toact on its trade signals. When the equity line is above its mov-ing average, profitability is in an “uptrend,” the system is mak-ing money and you continue to take each trade (see Figure 1,above).

If the system begins to struggle, the equity line will dropbelow its moving average, at which point you simply log

36 www.activetradermag.com • February 2003 • ACTIVE TRADER

Equity curve D R AWDOWN MANAGEMENTYour system has given you four losing trades in a row.

Should you take the next signal, or move to the sidelines?

Here are some thoughts on how to determine when to

stay out of the market and when to jump back in.

Equity line

A

B

C

10-period moving average

$ 4 0 , 0 0 0

$ 3 5 , 0 0 0

$ 3 0 , 0 0 0

$ 2 5 , 0 0 0

$ 2 0 , 0 0 0

$ 1 5 , 0 0 0

A, B and C mark those periods when the equity line droppedbelow its moving average, an indication that profitability is in a downtrend and the system’s signals should not be executed.

FIGURE 1 EQUITY LINE CROSSOVERS

Source: Excel

RISK Control and MONEY Management

Trade number

trades on paper — do not execute them in the market. You con-tinue to track the logged signals and their impact on the equi-ty line and its moving average. When the trading systembegins to generate profits again, the equity line will cross backabove the moving average, at which point you resume execut-ing trades in the market.

Window on profitabilityA second approach is to track the percentage of prof-itable trades within a certain moving window — say,the most recent 10 trades. As long as the percentage ofprofitable trades in this window is above a certainthreshold (e.g., 50 percent), continue to trade. If thep e rcentage of profitable trades drops below thethreshold, you would log the trades but not executethem. Then, track the system and resume tradingwhen it returns to profitability by moving back abovethe percentage threshold.

How they stack upWe’ll apply these two approaches to a two-year hypo-thetical track record for a 10-year T-note futures (TY)trading system. The system is momentum-based anduses a preset profit target of 10⁄32 for two contracts and atrailing stop for a third contract. (The target and stoplevels were determined using maximum favorableexcursion and maximum adverse excursion analysis.For more information on these techniques, see “Ontarget trading,” Active Trader, July 2001, p. 44, and“Taking the guesswork out of stop orders,” ActiveTrader, October 2001, p. 94. All the necessary analysiscan be done using an Excel spreadsheet.)

Figure 1 shows the system’s equity line and its 10-day moving average during 1999. (The equity valuewas adjusted to reflect a starting point of $20,000, andthe track record begins in 1998 so the 10-period mov-ing average is current at the start of the year.) Usingthe rule of not taking signals when the equity linedrops below the moving average results in three peri-ods when you would have stopped trading, labeled A,B and C. (The first losing trade that causes the equityline to drop below the moving average is a real trade,but the next trade would be a paper trade only.)

Figure 2 (top) shows the original equity line alongwith the new equity line representing the performanceof those trades executed when the original equity lineis above its 10-period moving average. The new,adjusted equity line has flat periods that occur whenthe modified system is not being traded.

The moving average technique reduced the sys-tem’s profit by about $3,500, but it also decreased theoriginal equity line’s worst drawdown (refer to B inFigure 1) from $2,653 to $1,720. This is a more-than-

acceptable compromise for traders who are willing to foregosome profit in return for the ability to limit risk to a level theyare comfortable with.

F i g u re 3 (bottom) uses a 10-trade moving window and a 50-p e rcent profitability threshold, which means as long as five or

ACTIVE TRADER • February 2003 • www.activetradermag.com 37

continued on p. 38

Equity line

A B

Adjusted equity line

10-trade profitability window

$ 4 0 , 0 0 0

$ 3 5 , 0 0 0

$ 3 0 , 0 0 0

$ 2 5 , 0 0 0

$ 2 0 , 0 0 0

$ 1 5 , 0 0 0

100%

50%

0

The moving window approach (taking trades only when 50 percent or more of the past 10 trades have been winners)decreased profitability, but it did not decrease the drawdown, as the moving average technique did.

FIGURE 3 WINDOW OF PROFITABILITY

Source: Excel

Equity line

A

B

C


$ 4 0 , 0 0 0

$ 3 5 , 0 0 0

$ 3 0 , 0 0 0

$ 2 5 , 0 0 0

$ 2 0 , 0 0 0

$ 1 5 , 0 0 0

The moving average approach (represented by the adjusted equity line) gave up $3,500 in profits, but it also decreased theoriginal equity line’s worst drawdown from $2,653 to $1,720.

FIGURE 2 BEFORE AND AFTER

Source: Excel

Trade number

Trade number

m o re of the past 10 trades are profitable, the system’ssignals should be executed. If four or fewer of the past10 trades are profitable, the trades are merely trackedon paper.

There are only two periods, A and B, during whichthe percentage of profitable trades dropped below 50percent. Like the moving average approach, the mov-ing window technique sacrificed profits, and for thesame reason — lag. However, the moving window didnot offset this reduction by shrinking the drawdownas well. That’s not the end of the story, though.

Moving forward one year, Figure 4 (top) shows thehypothetical returns for the year 2000. At the begin-ning of the chart, the original equity line is above the10-period moving average, which means the originaland adjusted equity lines are identical. At point A, theoriginal equity line drops below the 10-period movingaverage, so the next trade (point B) is only logged onthe books. Unfortunately (in terms of trading the mod-ified system), it is a very profitable signal, and it push-es the original equity line back above the 10-periodmoving average.

However, the next trade is a loss, so the systemagain moves to the sidelines. The difference in prof-itability at the end of the year was less than $2,000, butthe maximum drawdown for the original equity linewas $4,592, while that of the adjusted equity line wasonly $2,692. This was certainly an improvement overthe 1999 example, but the modified system still had alengthy drawdown: It took another 30 trades beforethe adjusted equity line was able to move above itsprevious equity peak at trade 14.

Figure 5 (bottom) shows the year 2000 results ofusing the 10-period moving window with the 50-per-cent threshold. The original system had more than 50-percent profitability for the most recent 10 trades untilpoint A. After that, only paper trading was permitted.Real trading resumed at point B when the percent ofprofitable trades got back to the 50-percent threshold.

This time, the moving-window technique dramati-cally reduced the system’s drawdown — only $1,756vs. $4,592. By avoiding this drawdown, the moneymanagement technique actually outperformed theoriginal system by $500 by the end of the year.

Drawdowns and market conditionsWhat was happening in the market that resulted insuch a severe series of losing trades? Figure 6 (oppo-site page) is a daily chart of the continuous 10-year T-note futures contract. The system struggled during themiddle of the year when the T-note was in a nearly two-monthtrading range before it embarked on a trend. The profit targetsthat are typically met during a trending market were not beingreached during the trading range, so the system was takinglosses.

One lesson from this experience is the importance of logical-

ly tying a trading system to the price action of a market. Inother words, you should know what kind of price behavioryour system is relying upon to capture profits. (In this way, thesystem’s profitability itself becomes an indicator of the type ofmarket environment you are in.)

38 www.activetradermag.com • February 2003 • ACTIVE TRADER

Equity line

A

B


10-period moving average

$ 4 0 , 0 0 0

$ 3 5 , 0 0 0

$ 3 0 , 0 0 0

$ 2 5 , 0 0 0

$ 2 0 , 0 0 0

$ 1 5 , 0 0 0

Using the moving average filter was more successful in 2000 than it was in 1999, even though the adjusted equity linesuffered through a prolonged drawdown.

FIGURE 4 MOVING AVERAGE: YEAR 2000

Source: Excel

Equity line

AB


10-trade profitability window

$40,000

$35,000

$30,000

$25,000

$20,000

$15,000

100%

50%

0

The moving window technique that underperformed in 1999 outperformed in 2000, reducing the system’s drawdown andincreasing profitability by the end of the year.

FIGURE 5 MOVING WINDOW: YEAR 2000

Source: Excel

Trade number

Trade number

Stick to the rulesHaving a rule that tells you when yoursystem is failing to perform permits youto move to the sidelines based on logicand discipline, rather than fear. Them o re you can remove emotion fro mtrading, the higher your chances oflong-term success.

For the techniques outlined here, thechoices of a 10-period moving averageand a 10-trade moving window with a5 0 - p e rcent threshold were arbitrary.Other parameters may work better orworse, and should be explored fullybefore trading.

F i n a l l y, this study highlights thevalue of developing a trading methodand documenting all of its trades —both historically and in real time. Doingso allows you to perform the kind ofanalysis described here to manage yourmoney in a more scientific fashion.Ý

ACTIVE TRADER • February 2003 • www.activetradermag.com 39

10-year T-note continuous futures (TY), daily

April May June July August Sept.

3 10 24 1 8 15 22 1 12 19 26 3 17 24 1 14 21 1

93,000

92,000

91,000

90,000

89,000

88,000

Because of the congested market conditions that prevailed for a good portionof the test period, the system’s profit targets were not met as often as theywould be in a trend environment. This underscores the importance of under -standing the price behavior your system relies on to generate profits.

FIGURE 6 MARKET CONDITIONS MATTER

Source: CQGNet

BY THOM HARTLE

I n the late 1980s, Tobey Crabelpublished a series of articlesdetailing various short-term pricepatterns for futures trading. The

patterns were designed to detect short-term volatility changes — that is, a mar-ket would move into a short-term con-solidation, and then break out of it.

Two of Crabel’s patterns will be cov-ered here. The first is an inside day thatis smaller than the previous four days(IDnr4) and the second is a day whoserange is less than that of the previous sixdays (NR7). An NR7 day does not haveto be an inside day.

F i g u re 1 (right) contains examples ofthese price patterns (in a stock rather thana futures contract). Bar Ais an IDnr4: It isan inside day and its range is smaller thanthe ranges of the previous four days. BarB is an NR7 day. In this case, the day’srange is smaller than the previous sixdays’ ranges. Finally, bar C qualifies asboth an IDnr4 and an NR7 day.

Analyzing IBM daily bar data fro mOct. 1, 2001, to Nov. 15, 2002, revealed that15 IDnr4 and 44 NR7 days occurred overthis 286-trading day period. This meansthese patterns are somewhat rare .H o w e v e r, many trading opportunitiesp resent themselves if you look for thesepatterns across a number of stocks.

Trading the patternsCrabel recommended trading these pat-terns the next day using an entry strate-gy based on the opening price and a pre-set value called the “stretch.” He hadtwo versions of the strategy — onecalled the opening range breakout (ORB)

40 www.activetradermag.com • April 2003 • ACTIVE TRADER

IBM (IBM), daily

A

B

C

29 1 5 12 19 26 August

82.50

80.00

77.50

75.00

72.50

70.00

67.50

The bars at A, B and C are all relatively narrow-range bars that representshort-term consolidations. Applying specific entry rules allows you to takeadvantage of moves out of the consolidation when volatility increases.

FIGURE 1 IDENTIFYING SHORT-TERM VOLATILITY CHANGES

Source: CQGNet

Opening s h o t s

Narrow-range bars and inside

bars represent short-term

volatility lows out of which

price can move sharply. This

strategy uses a simple volatility

measurement to determine

where to enter trades to

capitalize on this behavior.

TRADING Strategies

and the second the opening range break-out preference (ORBP).

The stretch value is the 10-day aver-age of the difference between the openand the low or high, whichever is small-er. For example, if the open was 79.50,the high was 80.33 and the low was77.98, you would use the diff e re n c ebetween the open and the high (0.83) vs.the open and the low (1.52). Figure 2(top) shows the stretch range for IBMover the last year: from an extreme ofjust under $1 to just over $0.20, with theaverage just over $0.50.

The entry strategy using ORB is toplace a buy stop just above the openingprice plus the stretch amount, and placea sell stop just below the opening priceminus the stretch amount. The first stopt r i g g e red is the trade, and the other auto-matically becomes the protective stop.

If other technical indicators provideevidence of a strong trend, then use theORBP: Place the entry stop in anticipa-tion of the trend continuing. For exam-ple, if the stock is in an uptrend, enteronly a buy stop above the opening priceplus the stretch amount. If filled, placethe protective stop below the stre t c hsubtracted from the open.

Time is of the essenceAccording to Crabel, the earlier in thetrading session the entry stop is hit themore likely the trade will be profitable atthe close. A trend triggered quickly inthe session could rack up a substantialprofit by the close and should be held fora possible two- to three-day run.

Once filled (and if the market isexhibiting a strong trend early), thenmove the protective stop to breakeven. Ifyou are not filled early in the session,reduce the size of the position as timepasses through the day. Any positionsfilled near the close are suspect. In thiscase, you may have only a small unreal-ized profit, and probably should nothold the trade overnight.

Rules in actionFigure 3 (bottom) is Figure 1 with the

ACTIVE TRADER • April 2003 • www.activetradermag.com 41

Day

$1.20

$1.00

$0.80

$0.60

$0.40

$0.20

$0.00

The “stretch” is an amount added to a bar’s opening price to determine theentry point and stop level, respectively.

FIGURE 2 TAKING A STRETCH

IBM (IBM), daily

A

B

C

29 1 5 12 19 26 August

30-day EMA

82.50

80.00

77.50

75.00

72.50

70.00

67.50

Whether a market is in a trend (defined here by a 30-day EMA) dictates thetrade entry technique that is used.

FIGURE 3 FACTORING IN THE TREND

Source: CQGNetcontinued on p. 42

addition of a 30-day exponential movingaverage (EMA). Bar A is the IDnr4 day.Considering the market moved abovethe previous resistance level at 72.50 inthe act of confirming a double bottompattern, and the 30-day EMA was risingand provided a support level, it’s logicalto conclude the stock is in an uptrend.Accordingly, we want to use the ORBPguidelines.

At this time, the stretch was 0.49. Thenext day IBM opened at 71.55, so the

entry stop price was 72.04 (71.55 + 0.49)and the stop loss was 71.06 (71.55 – 0.49).That day IBM climbed as high as 73.78,fell as low as 71.19, and closed at 71.90.In this situation, when there is no profitat the close, you should take the loss (inthis case, 14 cents plus any slippage).

Bar B is the NR7 bar. The market isclearly in an uptrend, and the prudentthing to do is trade with the trend, againfollowing the ORBPguidelines. After theNR7 day, the stock opened at 76.50; thestretch was 0.48. In this case, the stockmade a low of 75.98 (0.52 below theopen) in the first few minutes, whichwould have triggered a short sale if wedid not use the ORBP rule to trade withthe trend.

The long-entry stop was 76.98, whichwas hit later in the morning. The stockclosed at 79.35 for a profit of 2.37 thatday, not including slippage. This openprofit warranted holding the positionovernight.

The next day, the stretch is used as atrailing stop, based on the opening price(79.35). The stretch had increased to 0.53,making the stop level 78.82 (79.35-.53),which was not hit. The next day, thestock opened at 81.56. Subtracting the

stretch (0.53) sets a new stop at 81.03, theapproximate level at which the tradewould have been stopped out.

Bar C is the combination IDnr4 andNR7 bar. Still trading from the long side,the market opened the next day at 81.90.Adding the current stretch of 0.53 creat-ed an entry stop of 82.43. But because thehigh was 82.00 on this day, no long tradewas executed.

As an example of what would happentaking the trade against the trend, sub-tracting 0.53 from the opening of 81.90would have put you short at 81.37. Thestock closed at 81.00, which would haveresulted in a small profit if you exited atthe close. If you stayed short, however,expecting some follow-through pro f i t

taking, you would have been stoppedout at 81.48 (using the stretch the follow-ing day to determine the stop-loss).

Profit targetsOne way to take partial profits is to use apercentage of the 10-day moving aver-age of the daily ranges as a target. Figure4 (left) compares the 10-day average ofthe daily ranges compared to the stretchamounts. The average daily range is2.55, but it ranges from just above 1.50 tojust over 3.50. A reasonable target is 66-percent of the current 10-day movingaverage, just to be sure to put away somepartial profits when you have thechance.

For example, at the bar after A inF i g u re 3, the current 10-day movingaverage of the range was 2.78. Assumingthat once you are stopped into the longposition you are confident the low at71.19 is in place for the day, add 66 per-cent of 2.78 (1.85) to 71.19 for a target of73.04. This would have given you partialprofits of 0.98 on this day before slip-page, which offsets the loss taken hadyou held the position at the close.

In addition, had you traded againstthe trend after the bar C trigger, youwould have been short at 81.37. The 10-day moving average of the daily rangeshad increased to 3.06, so the targ e twould have been the high of 82.33 – 2.04= 80.31 (3.06 * 0.666 = 2.04), which wouldhave been reached.

Finally, once in a trade, consider mov-ing your stop to breakeven if the markethas reached a point that represents 50percent of the 10-day average of thedaily range.

Be a specialistAvery important aspect of applying thiskind of pattern-based technique is yourability to execute trades with as littleslippage as possible. Identifying thesetypes of price patterns, and combiningthem with rules for entering, managingrisk and taking profits in various mar-kets amounts to specializing. And spe-cialization is as popular and pro f i t a b l ean endeavor in trading as it is in otherd i s c i p l i n e s .Ý


10-day MA of the daily ra n g e s

S t r e t c h

Day

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0

To increase the odds of profitability, you can take partial profits when pricemoves in your direction by 66 percent of the 10-day average daily range.

FIGURE 4 DETERMINING LOGICAL TARGETS

BY THOM HARTLE W hen analyzing theb road market, manytraders consult vol-ume to confirm price

moves. Total volume is most often thesubject of this analysis, but comparisonsof up volume (total volume in stocksthat are up on the day) and down vol-

ume (total volume in stocks that aredown for the day) can provide addition-al insight into market behavior.

Typically, if up volume is greater thandown volume, the market will be up onthe day, and vice versa. Figure 1 (left)shows an example of the relationshipbetween daily up and down volume.

The top half of the chart is the S&P500, and the bottom is the differencebetween up volume and down vol-ume. When the S&P closed up, thedifference between up volume anddown volume was positive; on daysthe market closed down, the differ-ence was negative.

General volume characteristicsThe battle between up volume anddown volume is similar to a car racein which one car is running at topspeed and the other is sputteringalong. When traders and investors arevery bullish, they are buying more —or more aggressively — than they areselling. Up volume is greater re l a t i v eto down volume, and the marketadvances. If traders and investors arebearish, the opposite is true. Duringperiods when buyers and sellers arebalanced, the up and down volumestatistics will be closely matched anda trading range will unfold.

An intraday view of two days —one where the bulls dominate andone where the bears are in charge —offers additional perspective on theinterplay of up and down volume.

43 www.activetradermag.com • May 2003 • ACTIVE TRADER

S&P 500 (SPX), five-minute

Up volume - down volume

7 14 21 28 1 4 11 18 25 2Nov. Dec.

9 5 0 . 0 0

9 2 5 . 0 0

9 0 0 . 0 0

8 7 5 . 0 0

8 5 0 . 0 0

8 2 5 . 0 0

8 0 0 . 0 0

7 7 5 . 0 0

1 5 , 0 0 0

1 0 , 0 0 0

5 , 0 0 0

0

- 5 , 0 0 0

- 1 0 , 0 0 0

Up volume reflects how much trading is occurring in stocks that are trading up for the day; down volume is the opposite. This bottom part of the chart plots up volumeminus down volume. Up volume was usually higher on days the market closed up, and vice versa.

FIGURE 1 UP AND DOWN VOLUME

Source: CQGNet

What goes U P must come D O W NThe balance of up volume and down volume can tell you

a lot about the market’s bias during the trading day.

Here we detail a unique volume-based

indicator and ideas for using it in the markets.

TRADING Strategies

(All of the following charts featurefive-minute bars.)

F i g u re 2 (top) shows three charts forN o v. 1, 2002. The upper left is up vol-ume, the upper right is down volumeand the S&P 500 is on the bottom.Points Aand B re p resent the pre v i o u s(Oct. 31) close, a day on which downvolume was approximately 200 mil-lion more than up volume, and theS & P closed down just less than fivep o i n t s .

On the Nov. 1 close (points C andD), however, up volume was morethan 1 billion shares and down vol-ume was less than 400 million shares.It’s no surprise, then, the S&P 500was up more than 15 points on theday. The steady rise and steeper angleof ascent of the up-volume line rela-tive to the down-volume line is anindication of solid buying power.

Figure 3 (bottom) shows the differ-ence between up volume and downvolume compared to the S&P 500 onNov. 1. When the opening bell rang,the S&P 500 was down from the pre-vious close and the diff e re n c ebetween up volume and down vol-ume was negative. At point A, thedifference bottomed and climbed intopositive territory just as the S&P 500moved into positive territory.

Also, notice that while the volumed i ff e rence made a new high at point B,the S&Pdid not make a new high untila short time later. Not all stocksincluded in the calculation of the vol-ume statistics are in the S&P500. Here ,up volume was continuing to climb ata faster pace than down volume, indi-cating the broad market was still ris-ing. The S&P 500 soon followed.

F i g u re 4 (p. 32) is more intere s t i n g .The top half of the chart shows the ratio(rather than the diff e rence) of up vol-ume to down volume. There appears tobe a greater correlation between thiscalculation and the index, as the peaksand troughs of the two series are close-ly aligned. The ratio of up-to-down vol-ume runs nearly 2-to-1 for most of the day.

T h e re are two components to the ratio— up volume and down volume. The nextstep is to analyze the individual compo-nents to see if an indicator can be cre a t e d .

Designing an indicatorThe primary challenge of developing anindicator is to have up volume anddown volume start each day at zero. Toaccomplish this, the indicator must bequick to drop values from the previous

day, like a simple moving average. Also,because both up volume and down vol-ume climb all day, it may be advanta-geous to calculate the rate of change ofthese values; acceleration or decelerationin the rate of change might forewarn, or

ACTIVE TRADER • May 2003 • www.activetradermag.com 44

continued on p. 45

Up volume - down volume

A

B

C


1-8:30 9:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 1-14:30Nov.

6 0 0 M

4 0 0 M

2 0 0 M

0

- 2 0 0 M

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

As the S&P gradually moved up, so did the difference between up volume and downvolume.

Source: CQGNet

N Y S E advancing volume, f i v e - m i n u t e

NYSE down volume, five-minute

A

BC

D


1-8:30 9:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00 1-14:30

1-8:30 10:30 12:30 1-14:30

Nov.

Nov. Nov.

1 . 0 B

8 0 0 M

6 0 0 M

4 0 0 M

2 0 0 M

0

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

1 . 0 B

8 0 0 M

6 0 0 M

4 0 0 M

2 0 0 M

0

On an bullish day for the S&P 500 index, up volume (left) increased at a much greaterrate than down volume (right).

FIGURE 2 UP VOLUME LEADS DOWN VOLUME

Source: CQGNet

1-8:30 10:30 12:30 1-14:30

FIGURE 3 MOVIN’ ON UP

at least confirm, the current price trend.The most common way to measure

rate of change is to compare the currentvalue to the value x prices ago — forexample, five bars, 10 bars, etc.

However, this isn’t necessarily the bestm e a s u rement, because it ignores theprices between the current and the old-est in the lookback period. A better cal-culation would incorporate all the price

action in the lookback period. Onesuch measurement can be derivedfrom linear regression.

A linear re g ression line is a “best-fit” straight line plotted through datapoints. If the data points are zig-zag-ging upwards, a best-fit line drawnt h rough the points would be rising orhave a positive slope. If the datapoints are zig-zagging downward s ,the best-fit line would have a negativeslope. If the data points were zig-zag-ging sideways, the best-fit line wouldbe horizontal — that is, its slopewould be zero. (For more informationon re g ression lines, see “The next-barf o recast”.)

The top of Figure 5 (bottom) showsup volume, and below it is an indica-tor that tracks the slope of a best-fitlinear re g ression line of the up volumeusing a five-period lookback. (Thefirst five readings are ignored becausethey include the previous day’s val-ues.) The market began on a negativenote, but up volume was climbingf rom the early part of the day, and theslope indicator started just above zero .A little after 9:30 a.m. CT, the slopeindicator shot above +250, indicating al a rge increase in buying.

At points A and B the slope linedropped near zero and below zero atpoint C. These low readings coincid-ed very closely with the S&P’s pull-backs within the day’s uptrend. Moreimportantly, the slope line was abovezero almost the entire day and consis-tently had readings at or above +250.

Figure 6 (opposite page, top) is likeFigure 5 except it shows down vol-ume. At point A, the slope was justover +150, much higher than the upvolume reading for the same periodin Figure 5. With down volume dom-inating the early market, the S&P 500was down in the first hour of trading.The peak at A coincided with the bot-tom in the S&P, and when the marketbegan to advance, the slope linedropped to nearly -100 (point B). Thisis where the up volume advancedabove +250 in Figure 5.

This change represented a real shifttoward up volume, and the fact the

up-volume slope line had a higher intra-day peak than the down-volume slopeline suggested an uptrend was in thewings. For most of the day, the down-volume slope line stayed between -100


Up volume/down volume ra t i o


1 - 8 : 3 09 : 0 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 1 - 1 4 : 3 0Nov.

3

2

1

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

Instead of subtracting down volume from up volume, the top part of this chart shows upvolume divided by down volume. Like Figure 3, the line steadily climbs throughout thed a y. However, this calculation tracks the S&P more faithfully than the subtraction formula.

FIGURE 4 THE UP-VOLUME/DOWN-VOLUME RATIO

Source: CQGNet

Up volume

Up-volume slope

CBA


1 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 1 - 1 4 : 3 0Nov.

1 . 0 B

7 5 0 M

5 0 0 M

2 0 0 M

0

2 5 0

0

- 2 5 0

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

Below the up-volume line (top) is a line reflecting the slope of a five-day linearregression line at each bar. A linear regression (or “best-fit”) line incorporates all data points of a lookback period. The slope sunk to (or below) the zero line at thesame time price made intraday pullbacks.

FIGURE 5 THE UP-VOLUME SLOPE LINE

Source: CQGNet

Ignore these readings

to +100, except for one excursion to+200 at point D. This coincided withthe worst sell-off of the day.

Figure 7 (bottom) shows how theup volume, down volume battleplays out on a down day. The top partof the chart plots the up volumedivided by the down volume. Thetrading day started with the up-vol-ume/down-volume ratio above 40percent, a number that fell to 20 per-cent by the end of the session.

The top portion of Figure 8 (p. 47)contains the up volume, with itsslope line. The market opened downbut the slope line bounced back tojust below +75 at point A. For the bet-ter part of the day the slope stayed atrelatively low levels as the marketworked its way down. This con-firmed the lack of buying. The onlydecent rally came at point B, whenthe slope moved back to +75.

F i g u re 9 (p. 34) shows the downvolume with its slope line for the samed a y. The slope line quickly movedabove +100 and hovered near thislevel for most of the day until surg i n ghigher toward the end of the day.

Although these are only two exam-ples of trending days, in both situa-tions the slope line provided a goodindication of the dominant marketforce. Strong uptrending days haveup-volume/down-volume ratios of2-to-1 (200) or higher. Downtrendingdays are accompanied by up-vol-ume/down-volume ratios of lessthan 1-to-2 (50). On non-tre n d i n gdays, the ratio tends to be much clos-er to 1-to-1 (100).

Market contextWhat sets the stage for an uptrendingor downtrending day? Most often,some news event will push people’spsyches in one direction or the other.The catalyst can be expected news,such as economic releases, orunplanned, such as a group of ana-lysts simultaneously making com-ments early in the morning.

Therefore, a quick check of the morn-ing volume statistics can often give anindication of what kind of day to expect.If the news was good and the up-vol-ume/down-volume ratio is 200 or high-er, expect a market that will wind its wayupward.

On the other hand, if up-

volume/down-volume is 50 or less afterthe first hour of trading, the sellers willlikely be in control all day, barring someother news that reverses the crowd’smentality.

Sometimes the market action may bemore technical in nature — for example,the result of profit-taking after an

extended run, or a breakout of supportand resistance. On days such as these,check the slope indicator early in the dayand see whether up or down volume ismore dominant. If the up slope line hits+200 and down slope line only rises to+50, higher highs are likely. On the other


continued on p. 47

Down volume

Down-volume slopeC

D

B

A


1 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 1 - 1 4 : 3 0Nov.

7 5 0 M

5 0 0 M

2 5 0 M

0

2 0 01 0 00- 1 0 0

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

The slope indicator for down volume peaked at the same points the up-volume slopeline made temporary bottoms in Figure 5.

FIGURE 6 THE DOWN-VOLUME SLOPE LINE

Ignore these readings

Up volume/down volume ra t i o


9 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 9 - 1 4 : 3 0

. 5 0 0

. 4 0 0

. 3 0 0

. 2 0 0

9 1 0 . 0 0

9 0 5 . 0 0

9 0 0 . 0 0

8 9 5 . 0 0

The up-volume/down-volume ratio (top) on a day the S&P declined essentially mimicsthe downtrend in the index.

FIGURE 7 ON A DOWN DAY

Source: CQGNet

Source: CQGNet

hand, if slope for the down volumeclimbs to +150, and the up volume dropsto –60, the bears are in control.

Taking another stepDuring obvious trend days, another cal-culation involving the volume slope

indicators can help decipher intradayprice action: the difference between theup-volume slope line and the down-vol-ume slope line. An uptrending day pro-vides the first example.

The concept here is the short-termmarket trend peaks as the up-volume

level peaks, and the up-volume slopeline peaks shortly after. At the sametime, the down-volume will troughand begin to rise, and the down-vol-ume slope line will turn up. Watchingthe diff e rences between the twoslopes can highlight volume changes,and in turn, market shifts.

F i g u re 10 (opposite page, top) isanother chart of Nov. 1. The top part ofthe chart plots the diff e rence betweenthe up-volume slope line and thedown-volume slope line. (Rememberthe first five readings are ignoredbecause they incorporate the pre v i o u sday’s readings.) A horizontal line isplaced at +50, a level chosen (instead ofz e ro) because the diff e rence betweenthe two slopes may not cross zero .

Recall that just after 9:30 CT on thisd a y, the up-volume level cro s s e dabove the down volume level (seeFigure 3), and the up-volume/down-volume ratio was close to 2-to-1 (seeF i g u re 4). Both developments aresigns of an emerging uptrend.

Let’s develop some hypotheticaltrading rules:

1. A buy signal occurs when theslope-difference line turns up after ithas fallen below 50.

2. The risk point is just below thelow of the entry bar.

3. Take a profit when the slope-dif-ference line crosses above 50, andthen back below 50.

Each letter in Figure 10 marks anentry; the exits are marked on thea p p ropriate price bar. There are anumber of good entry and exit points,with some losing trades occurringduring congestion (see points C andD). Trade E was stopped out becausethe market traded below the low ofthe entry bar. However, because thes l o p e - d i ff e rence line was below the+50 line, the trade was re - e n t e re d .Another idea is to experiment with ahigher exit threshold, such as a riseabove and a drop below 200.

F i g u re 11 (opposite page, bottom)shows a downtrending day. In this

instance, we placed a horizontal line at the-50 level of the chart showing the diff e r-ence between the two slopes. For a down-trending market (i.e., the differencebetween the up volume and down volumeis negative, and the up-volume/down-volume ratio is 50 percent or less), go short


Up volume

Up-volume slope

A

B

BA


9 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 9 - 1 4 : 3 0

4 0 0 M

3 0 0 M

2 0 0 M

1 0 0 M

1 0 0

5 0

0

9 1 0 . 0 0

9 0 5 . 0 0

9 0 0 . 0 0

8 9 5 . 0 0

Short-lived intraday rallies are accompanied by peaks in the up-volume slope line.

FIGURE 8 UP-VOLUME SLOPE ON A DOWN DAY

Source: CQGNet

Down volume

Down-volume slope


9 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 9 - 1 4 : 3 0

1 . 0 B

7 5 0 M

5 0 0 M

2 5 0 M

0

3 0 02 0 01 0 00

9 1 0 . 0 0

9 0 5 . 0 0

9 0 0 . 0 0

8 9 5 . 0 0

The slope of down volume remained steadily above the zero line and shot up at theend of the day when the index declined.

FIGURE 9 DOWN-VOLUME SLOPE ON A DOWN DAY

Source: CQGNet

when the slope-diff e rence line is above-50 and turns down. The risk point isjust above the high of the entry bar, sotake profits if the slope-diff e rence linec rosses back above -50.

The first trade (point A) was a loser,but the next trade was pro f i t a b l edespite a lower entry point than thep revious trade. Trade C was caught insome sideways market movement.You would not re-enter after this tradebecause the slope-diff e rence line wasbelow -50, not above. Trade D had anice open profit, but the majority wasgiven back, and trade E was a goodrun down until the close.

These signals could be improved bycalculating the typical price move fol-lowing an entry, using maximumfavorable excursion (MFE) analysis toset targ e t s .

The ups and downs of the marketIntraday up- and down-volume sta-tistics offer insight into the dynamicsof buying and selling pressure. Whenthe up- and down-volume numbersare nearly the same throughout theday, the market tends to trade backand forth — the case on most tradingdays. However, there are days whenthe market makes significant moves,and the intraday volume numberscan help identify trade opportunities.

R e m e m b e r, there is usually someevent that causes people to favor oneside of the market on a given day. Inthe first hour or two, you can deter-mine if up or down volume is domi-nating by looking at the raw diff e re n c ebetween the two, the ratio or by usingslope analysis. However, tre n d i n gconditions do not develop every day,which is why intraday traders havemostly good days and bad days and,only occasionally, great days.

The concept of tracking the intra-day volume statistics presented hereis the groundwork for furtherresearch. Traders could use the slope-difference line of the up volume anddown volume as the basis for a sys-tem, or elements for either entry andexit strategies within other systems.

Finally, on those days with a bullishbias, watch for the down volume indica-tor to show signs that its ascent is wan-ing. The market will typically rally fromthat point. Likewise, on bearish days, the

bears will take charge when the up vol-ume slows. This may seem counter-intu-itive, but it will have you thinking aboutbuying in an uptrend (after a pullbackdriven by an advance in the number of

down-volume stocks) or selling a rally(after a short-term rise in the number ofup-volume stocks).Ý


Slope difference

AB

B

C

C

D

D

E

FE

F

A


1 - 8 : 3 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 1 - 1 4 : 3 0Nov.

6 0 0

4 0 0

2 0 0

0

- 2 0 0

9 0 0 . 0 0

8 9 5 . 0 0

8 9 0 . 0 0

8 8 5 . 0 0

8 8 0 . 0 0

The difference between the slope of the up-volume line and the slope of the down-volume line captures many of the market’s intraday turns. Trades that wereentered are marked by letters, with the exits noted.

FIGURE 10 SLOPE-DIFFERENCE LINE

Source: CQGNet

ExitExitExit

Exit

Exit

Exit

ExitExit

Exit

Exit Re-enter

Slope difference

A

B

B

C

C

D

DE

EA


9 - 8 : 3 0 9 : 0 0 9 : 3 0 1 0 : 0 0 1 0 : 3 0 1 1 : 0 0 1 1 : 3 0 1 2 : 0 0 1 2 : 3 0 1 3 : 0 0 1 3 : 3 0 1 4 : 0 0 9 - 1 4 : 3 0

2 0 0

1 0 0

0

- 1 0 0

- 2 0 0

- 3 0 0

- 4 0 0

9 0 7 . 5 0

9 0 5 . 0 0

9 0 2 . 5 0

9 0 0 . 0 0

8 9 7 . 5 0

8 9 5 . 0 0

8 9 2 . 5 0

When the trend is down, the slope differential can be used to signal trade opportunities. Moves above and below the -50 line are the initial triggers.

FIGURE 11 TRADING ON A DOWN DAY

Source: CQGNet

ExitExit

Exit Exit

Exit

Exit

V olatility, although often as-sociated negatively withchoppy, uncertain marketconditions, is essential for

traders: The market has to go somewhereto make a position profitable. Whattraders wish for is not low-volatility con-ditions; on the contrary, they prefer highvolatility occurring in the direction oftheir positions.

The degree of volatility is especiallycrucial for intraday traders because theytry to capture short-term price move-ment; they do not have the luxury ordesire to hold a position overnight. Forsuch traders, knowing when the mostvolatile periods occur during the trad-ing day would be a valuable piece ofinformation — if there were a consistentpattern to that volatility. Traders couldthen focus on those periods with themost promise of price movement andavoid those less likely to offer tradablevolatility.

Gathering the evidenceWe will determine if different times of


Do markets have intraday

price characteristics

short-term traders can

use to improve their

strategies?

We crunch some numbers

to find out when the

market is moving and

when it’s snoozing.

BY THOM HARTLE

the trading day exhibit noticeablevolatility characteristics by analyzing10-minute price bars in the S&P 500 E-Mini futures (ES) between 8:30 a.m. and3:10 p.m. CT, from Jan. 2 to March 31,2003. This daily trading period coincideswith normal market hours at the NewYork Stock Exchange (NYSE) plus 10minutes (we will see if anything intere s t-ing occurs in the futures after the NYSEcloses). The total historical analysis peri-od encompassed uptrends, downtre n d sand sideways price movement.

The total number of price bars overthese 61 trading days is 2,440 — forty10-minute bars per day. The narrowestprice-bar range was .50 points and thelargest was 15.00 points. (We are notconcerned with direction, just the rangesof the 10-minute bars throughout thetrading day.)

Intraday time patterns Figure 1 (opposite page) shows the aver-age ranges for each 10-minute tradingperiod in the trading day, beginningwith the bar that opens at 8:30 a.m. andthe final bar that opens at 3 p.m. Themost volatile time of the day is the firsthour of trading, when the 10-minuteaverage ranges noticeably exceed thoseduring the rest of the day as traders dis-count overnight events.

Table 1 (top) lists the average valuesfor the first hour, and shows the largestaverage 10-minute range (4.46 points) isthe period beginning at 9 a.m., which islikely because so many economicreports are released at that time.Referring back to Figure 1, the averageranges taper down to less than 2 pointsat 12:10 p.m., climb again as the marketnears the close, and finally drop off inthe final few bars. It is safe to assume thecombination of a lack of news and thelunch hour plays a role in the decliningvolatility during the middle of the trad-ing day in Chicago.

This initial analysis suggests the besttrade opportunities are early in the trad-ing day. After all, markets are all about

continued on p. 51

10.00

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

0.00

Here the 2:20 period (plotted as a line) is compared to the 12:20 period(plotted as a histogram). The 12:20 range was larger than the 2:20 range 15times (25 percent of all occurrences) the same day.

FIGURE 3 2:20 PERIOD VS. THE 12:20 PERIOD

Source: Excel; data from CQGNet

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

0.00

The 9:20 periods are plotted as a line and the 12:20 periods are plotted as a histogram. There were only six times (less than 10 percent of all occurrences) that the 12:20 period had a larger range than the 9:20 period the same day.

FIGURE 2 9:20 PERIOD VS. THE 12:20 PERIOD



The largest average 10-minute range (4.46 points) was the 9 a.m. period,which corresponds to when many economic reports are released.

TABLE 1 AVERAGE RANGES: FIRST HOUR OF TRADING

Time 8:30 8:40 8:50 9:00 9:10 9:20

Average range 3.62 3.68 3.60 4.46 3.54 3.32

Day

Day

price discovery, and the process is in fullforce as traders adjust to the release ofeconomic numbers and individual sto-ries driving particular companies andmarket groups. As the trading passesthe mid-point of the session, traderswho may have held onto losing posi-tions likely will be forced out whenvolatility increases as the market nearsthe close, while traders on the right sideof the day’s trend may see better profitswith a little patience.

For a more direct comparison of high-volatility and low-volatility periods,Figure 2 (p. 50) shows the 10-minuteranges for all the 9:20 a.m. periods (theblue line) and the 10-minute ranges forall the 12:20 p.m. periods (the his-togram) over the 61-day analysis period.There were only six times (just less than10 percent) that a 12:20 p.m. 10-minuterange was larger than the 9:20 a.m. peri-od 10-minute range for the same day.The average range for the 12:20 p.m.period was 1.99, and the number oftimes the range was 2 points or largerwas 32 (52 percent). The 9:20 a.m. periodwas 2 points or larger 55 times (90 per-cent).

Figure 3 (p. 50) compares the 2:20p.m. periods (the blue line) to the 12:20p.m. periods (the histogram). The 12:20p.m. range was larger than the 2:20 p.m.range on only 15 days, or 25 percent ofthe time. However, the range of the 2:20bar was 2 points or greater 48 times (79percent).

Figure 3 also shows that one 2.20 p.m.bar had a range of 9 points. Because ofits abnormally high range, this repre-sents an “outlier” — a statistical anom-a l y. To counter the dispro p o r t i o n a t eeffect an outlier can have when calculat-ing an average, some statisticians usethe median value instead. The median isthe center point, or middle, observationin a data set, with the data arrangedfrom the lowest to the highest value.Because there are 61 observations (trad-ing days) in the analysis period, themedian will be the value 30 placesbelow the maximum range and 30places above the minimum range forany of the 10-minute periods.

16.00

14.00

12.00

10.00

8.00

6.00

4.00

2.00

0.00

This chart shows both time and size of 10-minute ranges that were 6.25points or more. Twenty-three of the 32 occurrences, or 72 percent, occurredbefore 10 a.m., with nine occurring during the 9 a.m. bar.

FIGURE 6 LARGEST RANGES


0.60

0.50

0.40

0.30

0.20

0.10

0.00

-0.10

The difference between the average and median ranges shows almost all of thevalues are positive, indicating there is a positive skew to the data.

FIGURE 4 AVERAGE RANGES MINUS MEDIAN RANGES

350

300

250

200

150

100

50

0

The ranges span 0.50 to 14.75. The most common range value is 2.00, with 313 occurrences. The number of occurrences of 2.00 or larger is 1,702, or 69.8p e r c e n t .

FIGURE 5 RANKING THE RANGES

Range

Time

Time




Figure 4 (opposite page, top) chartsthe difference between the median andaverage values for all the 10-minuteperiods. Almost all the values are posi-tive, indicating there is a positive skewto the data — which means the rangestend to be larger than the median range.Interestingly, the 12:20 p.m. and 12:30p.m. periods have negative values, againhighlighting how the market slowsdown during the noon hour. (The 2:40p.m. hour has a negative difference, aswell.)

How common is a 9-point range for a10-minute bar? Figure 5 (opposite page,middle) sorts the ranges by size, fromthe smallest (.50) to the largest (14.75)and the number of occurrences. Themost common range is 2 points (313occurrences). The number of ranges 2points or larger is 1,702 (69.8 percent).There were only two 10-minute rangesof 9 points.

Figure 6 (opposite page, bottom) plotsthe time and size of 10-minute rangesthat were 6.25 points or larger (therewere 32 such occurrences). Twenty-three(72 percent) occurred before 10 a.m.,with the most (nine) occurring duringthe 9 a.m. bar.

Analyzing a second market: QQQLet’s review the same period for the

Nasdaq 100 index-tracking stock (QQQ).The narrowest range was 3 cents, andthe largest was 63 cents. Figure 7 (top)shows the average ranges for each 10-minute period throughout the tradingday.

As was the case with the S&P E-Minifutures, the most volatile time of the dayis the first hour of trading, during which

the average ranges exceeded those of theremainder of the day. Table 2 (middle)lists the average values for each 10-minute period in the first hour.

F i g u re 8 (bottom) plots the diff e re n c ebetween the average ranges and themedian ranges for each 10-minute period.Again, almost all the diff e rences are posi-


continued on p. 53

0.25

0.20

0.15

0.10

0.05

0.00

Here are the average ranges for each 10-minute period, beginning with thebar that opens at 8:30 a.m. and the final bar that opens at 3 p.m. Again, thebest trade opportunities are in the morning when volatility is the highest.

FIGURE 7 AVERAGE 10-MINUTE RANGES, QQQ


0.0300

0.0250

0.0200

0.0150

0.0100

0.0050

0.0000

-0.0050

-0.0100

Most of the differences between the averages and the medians are positive indicating a tendency for the ranges to be larger than the mid-point of the data.

FIGURE 8 AVERAGE MINUS THE MEDIAN, QQQ


In QQQ, the opening 10-minute bar and the 9 a.m. bar had average ranges of19 cents.

TABLE 2 FIRST HOUR OF TRADING, QQQ

Time 8:30 8:40 8:50 9:00 9:10 9:20

Average range 0.19 0.18 0.17 0.19 0.17 0.15

Comparing the

averages to the

medians, almost all

the values are

positive, indicating

there is a positive

skew to the data.

Time

Time

tive, suggesting the ranges tend to be, onaverage, larger than the median range.

Figure 9 (top) shows the individual10-minute ranges from all the 9:20 a.m.periods (the line) and all the 12:20 p.m.periods (the histogram). In this case, the12:20 p.m. period had a larger rangethan the 9:10 a.m. period on only threedays. The QQQ had one 10-minute barfrom this period with a range of 50 centson one day. Now, let’s look at how thesizes of the ranges rank over the totalanalysis period.

Figure 10 (middle) shows the size ofthe 10-minute bar ranges and the num-ber of times each occurred. The mostcommon range was 9 cents (271 occur-rences). In fact, more than 72 percent ofthe 10-minute bars had a range of 9cents or higher. Using only the rangesgreater than 25 cents (there were 49),Figure 11 (bottom) plots the ranges vs.the time of the day. Again, most of theoccurrences are in the morning, with 40of the 49 (82 percent) outliers occurringbefore 10 a.m., and 10 of those in the 9a.m. bar.

Another piece, not the whole puzzleThis study says nothing about the waythe ranges unfold. The fact that theaverage for the 9:20 a.m. bar is slightlylarger than the 9:40 a.m. bar says noth-ing about the relationship betweenthese two bars. Having said that, thebest volatility opportunities (for traderswith the proper skills) occur primarilyduring the early part of the trading day;they taper off toward noon, and thenrise again. In addition, there is a slighttendency for the occasional large movewithin the individual periods, whichalso offers the trader a better-than-aver-age opportunity.

If you analyzed an intraday tradingapproach without incorporating a refer-ence to time patterns, you might devel-op target and risks points that are not aslogical or useful because of unforeseenchanges in volatility. Those traders notfollowing equities or their counterpartfutures contracts should perform thesame analysis on their favorite mar-kets.Ý


.6

.5

.4

.3

.2

.1

0

The majority of the exceptionally large occurrences are in the morning with 40 ofthe 49 (82 percent) occurring before 10 a.m. Ten of those are in the 9 a.m. bar.

FIGURE 11 LARGEST RANGES, QQQ


300

250

200

150

100

50

0

The most common range was 9 cents, with 271 occurrences. Over 72 percentof the 10-minute bars had ranges of 9 cents or higher.

FIGURE 10 RANGE RANK, QQQ


0.60

0.50

0.40

0.30

0.20

0.10

0.00

The individual 10-minute ranges for the 9:20 period are plotted as a line andthe 12:20 periods are plotted as a histogram. There are only three occurrenceswhen the 12:20 period had a larger range than the 9:20 period.

FIGURE 9 THE 9:20 PERIOD COMPARED TO THE 12:20 PERIOD


Day

Range

Time

T he best trade opportunitiesarise when a market istrending in a certain direc-tion, but markets alternate

between these desirable trending phasesand hard-to-trade congestion phases.

Given this reality, a tool that high-lights congestion would make it easierto position oneself to take advantage ofthe next trend phase. An “internal” indi-cator that helps in this regard is the vol-ume of unchanged stocks trading on theNew York Stock Exchange (NYSE).

The NYSE breaks down the total vol-ume into up volume (the volume ofstocks trading above the previous day’sclose), down volume (the volume ofstocks trading below the previous day’sclose) and unchanged volume (the vol-ume of stocks unchanged from the pre-vious day’s close).

There are two issues to explore here:First, can a high unchanged volumereading be attributed to a high numberof stocks in a directionless mode?Second, what happens the followingday respective to a high unchanged vol-ume day’s high or low? Does the markettrend after a high unchanged volumereading?

We investigate these issues re g a rd i n gNYSE unchanged volume statistics viathe S&P 500 because of the popularityof its index-tracking stock (SPY) and theE-Mini S&P f u t u res contract (ES).

The volume messageThe insight volume offers is that itd i rectly reflects the degree to whichtraders and investors are committingmoney to the market or liquidating posi-tions.

Here are a few unchanged volumestatistics from recent market history.From June 1, 2002, to June 2, 2003 (252days), the lowest unchanged volumereading (just over 1 million share s )occurred on July 5, 2002. The highestreading was 58.3 million shares onSeptember 20, 2002. Sorting the data


Trends, consolidations andUNCHANGED VOLUME

Traders often monitor total volume, or up volume

and down volume. This analysis studies

the price action that follows days with high

levels of unchanged volume.

BY THOM HARTLE

S&P 500 index (SPX), daily

Unchanged volume

A

BC

D

12 19 27 2 9June

High unchanged volume

1,000.00

975.00

950.00

925.00

40 mil.

30 mil.

20 mil.

10 mil.

The highest unchanged volume reading corresponded to a market pause thatpreceded a substantial run up.

FIGURE 1 CALM BEFORE THE STORM

Source: eSignal

TRADING Strategies

from low to high values and looking atthe top 25 percent of the unchanged vol-ume results in 63 observations, with thelowest unchanged volume reading ofthis group just over 24.1 million sharesand the highest 58.3 million. Now let’slook at the market action following ahigh unchanged volume reading.

Figure 1 (opposite page) shows theS&P 500 cash index and the unchangedvolume level at each day’s close. May23, 2003, (bar A) was an inside day andthe unchanged volume was just under39 million shares. This reading was the12th highest reading during the obser-vation period. The following day, theS&P 500 index traded below the low ofbar A by 0.09 points, but the market thenrallied and the index made a high 17.56points above the May 23 high.

The day after bar B’s unchanged vol-ume reading of 25.6 million, the markettraded 2.69 points higher than bar B’shigh, then reversed and traded below itslow by 3.89 points. Bar C had anunchanged volume reading of 31.8 mil-lion, and the index traded 13.73 pointsabove the previous day’s high beforereversing and closing near the open.Finally, on bar D the unchanged volumewas 32.5 million, and the next bar vio-lated the bar D low by just 0.84 pointsbefore turning up.

Two of the four examples had sizablemoves, one traded both above andbelow the previous day range, and twomade slight down moves before revers-ing to the up side. Were there any notice-able patterns here?

Of the 63 observations withunchanged volume of 24.1 millionshares or more, there were only threetimes the following day was an insideday. Of the 60 remaining examples, onlyfive took out both the previous day’shigh and low, forming outside bars.

Figure 2 (top, right) shows how muchthe following day’s low or high exceed-ed the previous day’s low or high when

continued on p. 56

The high is exceeded

The low is exceeded

35

30

25

20

15

10

5

0

-5

-10

-15

-20

This table shows how much the S&P 500 index exceeded the previous bar’s highor low after a day with unchanged volume of 24.1 million shares or more.

FIGURE 2 UP AND DOWN MOVES


Lower Unchanged volume Higher

30

25

20

15

10

5

0

For the most part, higher unchanged volume levels were followed by biggermoves above the previous high.

FIGURE 3 DEGREE OF MOVEMENT — UP MOVES

The low is exceeded

Lower Unchanged volume Higher

20

18

16

14

12

10

8

6

4

2

0

Unlike the higher unchanged volume/up move relationship shown in Figure 3,this table shows little correlation between unchanged volume levels and downmoves the next day.

FIGURE 4 DEGREE OF MOVEMENT — DOWN MOVES

ACTIVE TRADER • October 2003 • www.activetradermag.com 55

the previous day’s unchanged volumeexceeded 24.1 million shares. Of thehigher highs, 12 were 10 or more S&Ppoints higher, with six exceeding 20points. The average difference was 9.14

points. For days when the pre v i o u sunchanged volume was greater than24.1 million shares and the low wasexceeded, the market fell more than 10points 11 times. The average difference

was -10.13 points. Another aspect of this data to consid-

er is the relationship between theunchanged volume level and the degreeof the following day’s price movement.Figure 3 (p. 55) shows how the previousday’s high was exceeded according tothe level of unchanged volume, from thelowest to the highest. There is a slightbias in that the lower unchanged volumelevels appear to have slightly lowerextremes, with the exception of the sec-ond example from the left, which fol-lowed the last day of the year, a day typ-ically with low volume because of theholidays.

Figure 4 (p. 55) is similar to Figure 3,except that it displays the amount bywhich the previous low was bro k e n .There is no correlation between the sizeof the penetration and the unchangedvolume level.

F i n a l l y, is there any correlation to thet rend of the market? Analyzing some ofthe largest moves suggests the pre v a i l i n gt rend does not exert much influence. Forexample, Figure 5 (top) shows the S&Prose more than 27 points after the June 14,2002, unchanged reading of 26.7 millions h a res while the market was in a down-t rend. On March 28, 2003, the unchangedvolume reading was 24.8 million share sand the following day the S&P d ro p p e d17 points below the previous day’s low(see Figure 6, left). However, anotherlook at Figure 1 shows two up days fol-lowing high unchanged volume re a d i n g sin an uptrending market.

Potential applicationsThe patterns following high unchangedvolume readings can help clue you in tod i ff e rent opportunities or risks byrevealing the typical behavior followingcertain price bars.

For example, this analysis suggests iftoday’s high is approximately 15 ormore points above the high of yester-day’s high unchanged volume day, thelikelihood of higher prices becomesm o re remote. Similarly, a low thatexceeds the high unchanged volumeday’s low by more than 10 points on thedownside is a low-probability situationfor a short trade. Ý



3 10 17 24June

26.7 million shares unchanged

1,075.00

1,050.00

1,025.00

1,000.00

975.00

950.00

The day after this high-unchanged volume day was an up day, one of twodays that interrupted a downtrend.

FIGURE 5 UP MOVE IN A DOWNTREND

Source: eSignal


10 17 24 31April

24.8 million shares unchanged

900.00

875.00

850.00

825.00

800.00

The day after this high-unchanged volume day moved lower, but the marketturned back up the next day.

FIGURE 6 DOWN MOVE IN UPTREND

Source: eSignal

BY THOM HARTLE

T rend analysis of the stock market can take severalforms: measuring an index’s percentage changeover a period of time, comparing the index to amoving average and so on. Furthermore, some

traders refer to tools such as the number of advancing stocksrelative to declining stocks and volume to determine a particu-

lar trend’s strength or weakness.Spread analysis, or measuring the

price difference (or ratio) between twostock indices, is another way to gaugethe robustness of a move in the stockmarket. By identifying the typical rela-tionship between two indices and recog-nizing when that relationship deviatesfrom its pattern, you can determine thetrend and identify potential reversals.Also, although this approach is general-ly longer term, you can track theNasdaq-S&P spread on an intraday basisto insure you are on the right side ofintraday trends.

This study analyzes the relationshipbetween two of the most popular stockindices, the S&P 500 (SPX) and theNasdaq 100 (NDX), which also underliethe two most popular equity indexfutures contracts, the S&P 500 E-Mini(ES) and the Nasdaq 100 E-Mini (NQ),traded at the Chicago Mercantile


TRADING Strategies

The TELLTALE spread

Analyzing the relationship between

the E-Mini Nasdaq 100 and the E-Mini

S&P 500 can indicate when the broad

market is making a genuine move

or when it’s faking people out.

The S&P 500 consists of the 500 largest publicly traded companies measuredby market cap. It is designed to reflect the broader market.

TABLE 1 S&P 500 COMPOSITION

Top stocks Top groups

General Electric 3.19% Financials 20.50%

Microsoft Corp. 3.06% Information technology 16.20%

Pfizer, Inc. 3.00% Health care 14.80%

Exxon Mobil Corp. 2.67% Consumer staples 11.70%

Wal-Mart Stores 2.62% Consumer discretionary 11.10%

Citigroup Inc. 2.45% Industrials 10.40%

Johnson & Johnson 1.71% Energy 5.80%

American International Group 1.60% Telecom services 3.90%

IBM 1.59% Utilities 3.00%

Intel Corp. 1.51% Materials 2.70%

Source: Standard & Poor’s 6/30/03

Exchange (CME). We’ll use the spread between theS&P and Nasdaq 100 futures contracts in this analysis.

We’ll begin by looking at a longer-term use of theNasdaq-S&P spread before shortening the time hori-zon and examining ways to identify intraday trendchanges.

Finding the market leader: S&P safe haven vs. Nasdaq growthThe S&P 500 is a capitalization-weighted index of com-panies with market caps (stock price multiplied bynumber of shares outstanding) in excess of $3 billion.The larger a company’s market capitalization, the moreits stock price affects the index value.

The S&P 500 is designed to reflect the risk and returncharacteristics of the broader, large-cap market. Table 1(opposite page) shows the top individual holdings andthe group breakdown of the S&P 500.

The Nasdaq 100 is comprised of the 100 largest busi-nesses, excluding financial companies, traded on theNasdaq stock market exchange. The index uses a “mod-ified capitalization-weighted” approach, by whichstocks are weighted with a proprietary algorithm when-ever any stock represents more than 24 percent of theindex’s total market value, and/or the combined weightof all stocks with weightings of at least 4.5 percentexceeds 48 percent of the index’s total market value.

Table 2 (above) is a recent list of the top Nasdaq 100


continued on p. 59

The Nasdaq 100 is comprised of the 100 largest companies trading on theNasdaq (financial companies are excluded), based on market cap. The indexis heavily weighted with technology stocks.

TABLE 2 NASDAQ 100 COMPOSITION

Top stocks Top groups

1. Microsoft Corp. 10.15% 1. Computer & office equipment 28.39%

2. Intel Corp. 5.10% 2. Computer software/services 28.01%

3. Cisco Systems Inc. 4.49% 3. Telecommunications 11.69%

4. Amgen Inc. 4.28% 4. Biotechnology 11.45%

5. QUALCOMM Incorporated 3.65% 5. Retail/wholesale trade 9.86%

6. Dell Computer Corp. 3.24% 6. Health care 4.51%

7. Comcast Corporation 3.07% 7. Services 3.16%

8. Oracle Corp. 2.82% 8. Manufacturing 1.94%

9. eBay Inc. 2.65% 9. Transportation 0.99%

10. Nextel 2.48%Communications, Inc.

Source: www.nasdaq.com 6/30/03

Nasdaq E-Mini (NQ), weekly

S&P 500 E-Mini (ES), weekly

NQ-ES spread, weekly

Divergence between S&P 500 and Nasdaq 100

Rising spread is bullish

April July October 2003 April July

1,500

1,250

1,000

1,000

800

300

200

100

0

Calculating the difference between the Nasdaq 100 E-Minifutures (top) and the S&P 500 E-Mini futures (middle) results in a spread chart (bottom) that shows when one index is outperforming the other.

FIGURE 1 THE SPREAD PERSPECTIVE

Source: CQG, Inc.

individual stock holdings and a breakdown of its most heavilyrepresented groups.

The Nasdaq 100 contains a much higher percentage of tech-nology stocks than the S&P. Table 2 shows the computer and

office equipment industry group comprises 28.39 percent ofthe Nasdaq 100, followed by the computer software/servicesgroup at 28.01 percent.

The top group in the S&P 500 is financial stocks (20.50 per-

cent), followed by information technology (16.20 percent). Interms of growth stocks, the technology industry offers far moreopportunities than the financial services industry.

Microsoft (MSFT) is the most heavily weighted individualholding in the Nasdaq 100 and the second most heav-ily weighted in the S&P 500. Because it accounts for alarge enough percentage in both indices, MSFT has arelatively muted effect on the spread between the two.

Because of its large technology component, in anexpanding economy the Nasdaq 100 should lead (i.e.,rise at a faster rate than) the S&P 500 when the overallmarket is moving higher because more money man-agers and investors will be attracted to the potential ofgrowth stocks. On the other hand, in a declining eco-nomic environment, financial services companies offera safe haven for money (plus, many financial compa-nies pay dividends). That will tend to pull moneyaway from Nasdaq stocks and into S&P stocks. As aresult, the Nasdaq 100 should lead the S&P when themarket is moving lower, as well.

In other words, a bullish stock market is reflected byan uptrending Nasdaq 100-S&P 500 spread (theNasdaq 100 price minus the S&P 500 price). A bearishstock market will be characterized by a downtrendingNasdaq 100-S&P 500 spread.

Weekly perspectiveFigure 1 (p. 35) shows the Nasdaq 100-S&P 500 rela-tionship from the fourth quarter of 2002 into the sec-ond quarter of 2003. The top and middle charts showthe Nasdaq 100 E-Mini and S&P 500 E-mini futures,respectively, while the bottom panel shows the spreadbetween the two.

Both the Nasdaq 100 and the S&P 500 futures con-tracts reached new lows in October 2002. However,during the fourth quarter of 2002, the Nasdaq 100embarked on a substantial rally, bettering its July 2002high; the S&P 500, however, was unable to surpass itssummer (August) high. The Nasdaq-S&P spreadjumped sharply higher, reflecting the Nasdaq 100’smore accelerated rally. Both markets peaked inDecember 2002 and moved downward until February.

The Nasdaq-S&P spread made a slightly lower lowin January, just below its December low. As the S&P 500 movedlower, the spread started to climb again, reflecting the Nasdaq100’s outperformance relative to the S&P 500. The Nasdaq 100made the second low of a double bottom in March 2003, at


A bullish stock market is reflected by an uptrending Nasdaq 100-S&P 500 spread.

A bearish market is characterized by a downtrending spread.

Nasdaq E-Mini (NQ), daily

S&P 500 E-Mini (ES), daily

NQ-ES spread, daily

New high

A

B

C

D

E

F

G

H

21 1 12 19 27 2 9 16 23 1 14 21 28 1May June July Aug.

1,300

1,250

1,200

1,150

1,100

1,000

950

900

300

275

250

225

200

The growth-oriented Nasdaq 100 tends to lead the S&P to theupside as well as the downside. When it doesn’t, as was thecase at points C and F, this lack of leadership can result intrend weakness in the overall market.

FIGURE 2 HIGHLIGHTING LEADERSHIP

Source: CQG, Inc.

which point both markets rallied into June. The developingspread relationship signaled this period of strength in the over-all market: The spread bottomed in October and began form-ing a series of higher highs and higher lows, indicating a bull-

ish market environment based on the better performance of theNasdaq 100 relative to the S&P 500.

Divergence on the daily time frameIn addition to gauging the relative strength of the indices whenthey are moving in the same direction, divergence between theNasdaq 100 and S&P 500 on the daily time frame can signalpotential market corrections. Figure 2 (opposite page) is a dailychart of the E-Mini Nasdaq 100, E-Mini S&P 500 and the spreadbetween the two.

Line A shows the Nasdaq 100 rising to slightly higher highsin May while the S&P 500 surpassed its early May highs by a

wider margin (approximately 1.28 percent vs. .87 percent,based on closing prices on May 6 and May 15). The spread (seeline C) was flat between these two peaks, indicating theNasdaq 100 was no longer outperforming the S&P. This situa-

tion was followed by a short correction into the weekof May 19.

Moving forward, the Nasdaq 100 peaked in earlyJune while the S&P 500 peaked in mid-June. Thisdivergence, indicated by the declining spread (line F),was part of another correction that lasted until the endof the month.

In July, the Nasdaq 100 surged to new highs, whilethe S&P 500 made a lower high. This divergence pre-ceded a correction in the broader market.(Interestingly, though, the spread itself surged to newhighs, which reflects leadership on the part of theNasdaq and should be a longer-term bullish sign.)

The spread does not necessarily indicate a correc-tion is complete; it does not wave a red flag.Nonetheless, there is value in being alerted to condi-tions that signal a potential correction.

Intraday applicationsOn a very short-term basis, the Nasdaq-S&P spreadcan help keep you on the right side of intraday trends.The same basic guideline holds, in that the Nasdaq 100should lead the way, both up and down.

Although simple trend analysis, such as drawingtrendlines, can help to spot changes, intraday spreadcharts, like individual markets, are very volatile. Forexample, during the latter part of June 2003, the stockmarket was moving sideways.

On July 1, the market broke key support levels, butthe spread did not break its equivalent level. As themarket began to recover, and moved back up throughthe previous broken support level, the spread brokethe down trendline shown in Figure 3 (left). TheNasdaq 100 and the S&P 500 did not break their trend-lines until later in the session.

Majority rulesAnalyzing the Nasdaq 100-S&P 500 spread relation-ship reflects the idea that if the majority of stocks are

not rising — or, if the market-leading stocks are lagging thebroader market — the trend may lack staying power.

The spread between the Nasdaq 100 and the S&P 500 canfunction as a gauge of how healthy or weak the overall marketis. If one of the major indices is not keeping pace, the spreadwill fail to make new highs or lows. In those situations, watchfor a trend change. �

For other articles by Thom Hartle, visit the Active Trader online storeat www.activetradermag.com/purchase_articles.htm.


Nasdaq E-Mini (NQ), 30-minute

NQ-ES spread, 30-minute

Support

Support

Support

T1

T1

T1

25 26 27 30 1 2-8:30July

1,250

1,225

1,200

990

980

970

960

250

240

230

220

Although intraday price data is more volatile than daily orweekly data, the Nasdaq-S&P spread relationship reflects thesame dynamics. Here, the spread (bottom) pushed above itsdowntrendline before the Nasdaq or S&P futures did.

FIGURE 3 INTRADAY INSIGHT

Source: CQG, Inc.

S&P 500 E-Mini (ES), 30-minute

61 www.activetradermag.com • December 2003 • ACTIVE TRADER

TRADING Strategies

5.00K 10.00K 15.00K 20.00K 25.00K 30.00KVolume

1029.251028.751028.251027.751027.251026.751026.251025.751025.251024.751024.251023.751023.251022.751022.251021.751021.251020.751020.251019.751019.251018.751018.251017.75

The highest and lowest prices for a given day tend to have the lowest volume.

FIGURE 1 WHERE THE VOLUME IS

Source: eSignal

BY THOM HARTLE

T raders typically incorporate several tools and con-cepts, such as price patterns, indicators, moneymanagement, and entry and exit strategies in atrading plan. But how much consideration do

they give time frame?For example, people who trade on an intraday basis might

gravitate to five-minute bars without giving the matter toomuch thought. Five minutes just seems like a reasonably com-pact, easy-to-reference time frame. But is there really anyadvantage to five-minute bars over two-minute bars, 10-minute bars or any other time frame?

The answer depends on the time frame you use to define thetrend, which should be different from the time frame on which

you execute trades. This usually means defining trend direc-tion on a longer-term time frame and executing trades in thatsame direction using setups on a shorter-term time frame — a“multiple time frame” approach.

To illustrate this process, we will begin by identifying sup-port and resistance levels with a basic chart pattern. To thatend, we need to consider how support and resistance levelsoccur.

What makes a support or resistance level?Support is often characterized as a price level at which buyersrepeatedly comeinto the market

Choosing the properTIME FRAMEIntraday traders can use time frames ranging anywhere

from one minute to more than an hour. Is there a way

to find the most appropriate time frame given your

trading approach? Read on to find out.

1

1

2

2

3

3

Buy signal is a move above the high of bar 2 or

the close of bar 3.

Pivot high

Sell signal is amove below thelow of bar 2 or

the close of bar 3.

Pivot low

The pivot low and pivot high shownhere represent the evaporation ofbuying and selling at low and highprice levels illustrated in Figure 1.

FIGURE 2 PIVOT POINTS


ACTIVE TRADER • December 2003 • www.activetradermag.com 62

and hold up prices; one theory is largeinstitutional traders scoop up cheapsecurities or contracts when they drop toa certain level. Resistance is consideredthe opposite — a level at which sellersoffer large numbers or size into the mar-ket, checking a rally.

H o w e v e r, Figure 1 (opposite page, farleft) tells a diff e rent story, in terms of howthe high and low for a trading day aretypically set. It shows how many con-tracts traded at diff e rent price levels inthe September E-Mini S&P 500 future scontract on Sept. 5, 2003. It doesn’t showwhen the market traded at a particularprice, only the total volume that occurre dat each price level that day. Notice thehighest and lowest prices for the day arethe levels with the lowest volume.

Figure 1 suggests the market stoppedgoing up because increasingly higherprices eventually failed to attract enoughbuyers willing to pay for the contract atthose levels. Volume decreased as priceadvanced to the top of the day’s range,which means the market stopped goingup because it ran out of buyers.

Similarly, the low for the day occurs because at some pointfalling prices fail to attract additional selling — i.e., the marketstopped falling in Figure 1 because there were no more sellers.This process repeats throughout the day, with the market oftenmaking several new intraday highs or lows. The ultimate highand low prices can occur at any time during the day.

How does this behavior manifest itself on a standard pricechart? Figure 2 (opposite page) shows two examples of a three-bar pattern called a “pivot,” in which price makes its finalmove into new territory (a higher high or lower low) on bar 2and reverses direction to form bar 3 (making a lower high orhigher low). The first example in Figure 2 is a pivot low and thesecond is a pivot high.

Based on the conclusions about support and resistance fromFigure 1, the ultimate high or low reached in bar 2 and the sub-sequent retracement of bar 3 in a pivot pattern is the result ofprice exhaustion in that direction. With the completion of bar3, the pivot high or low is in place. An early warning a pivot isforming is the violation of bar 2’s high (for a pivot low) or bar2’s low (for a pivot high).

Once a pivot low or high is in place, the market has estab-lished support or resistance at that price level.

The daily pivotDepending on the intraday time frame, any number of pivotlows and highs can form throughout the trading day. However,only one pivot high can be the high for the day and only onepivot low can be the low for the day.

For example, Figure 3 (above) shows a five-minute chart ofthe September 2003 E-Mini S&P500 futures. There are 16 pivothighs and 16 pivot lows in this chart; to avoid clutter, only two

pivot lows and one pivot high have been labeled. Point A is thepivot high for the day and point B is the pivot low.

Here’s the problem of using five-minute bars: If you look ata daily chart and conclude the trend is up, and you want to golong, what are your chances of spotting the intraday pivot lowthat represents the lowest low of the day — the most favorableentry point?

Obviously, you need to reduce the number of intraday pivothighs and lows to improve your odds. Using a longer-termintraday time frame would accomplish this. Figure 4 (p. 63)shows a 30-minute chart. Now there are just three pivot lowsand two pivot highs. (In this example, bar 1 of the first pivotlow is using the low of the previous day’s final bar.) In themiddle of the chart, it appears the large spike-down bar wouldbe bar 2 of a pivot low (these bars are marked A, B and C), butthere is no overlap between bar A and bar C. The pivot conceptis based on the idea that a market is not attracting sellers nearthe pivot low area. In this case, the fact that bar B sold off sosharply and bar C did not retrace enough to overlap with barA is a sign bar C is just a correction of bar B, and not the begin-ning of a reversal. (If the first bar gaps lower, as it did here, itmeans the trading occurred in the evening Globex session, sothe previous day’s bar would be bar 1.)

Five bars a day keeps the guessing awayMoving to an even longer-term time frame, Figure 5 (p. 63)shows an 81-minute bar chart that divides each day into fiveprice bars. Now, only two pivot lows and one pivot highappear.

Because a pivot high or low pattern consists of three bars,continued on p. 63

S&P E-Mini (ES), five-minute

12

3

12

3

A

B

123

9/5/03 9/8/03

1,030.0

1,028.0

1,026.0

1,024.0

1,022.51,022.0

1,020.0

1,018.0

1,016.0

Although only three are labeled, this five-minute chart contained 32 pivothighs and lows.

FIGURE 3 TOO MANY TREES, NOT ENOUGH FOREST


63 www.activetradermag.com • December 2003 • ACTIVE TRADER

breaking the day into five 81-minutebars increases the chances of identifyingthe ultimate high or low for the day. Ifwe were using five-minute bars, thechances of identifying the ultimate highor low during the day would be remotebecause of the high number of pivotlows and highs.

The choice of 81-minute bars is mostappropriate for trading if you are fol-lowing the trend on the daily time frame(a multiple time frame approach). Forexample, if the daily trend was definedas up based on the market closing abovea rising 20-day moving average, youcould enter on pivot lows on the 81-minute bar time frame.

Multiple time frames In Figure 6 (opposite page), the dailyrange is plotted as boxes that “encapsu-late” the five 81-minute bars for eachday, along with a 20-day moving aver-age (plotted as a “step” line). In essence,the encapsulation rectangles are liketransparent daily bars that allow you tosee the intraday action on the 81-minutebars.

At point A, the daily bars form a pivotlow (labeled 1, 2 and 3); bar 2 temporar-ily penetrates the 20-day moving aver-age. The combination of the rising mov-ing average and the formation of a pivotlow on the daily time frame signals anuptrend. Trades can be based on the for-mation of pivot lows on the 81-minutetime frame, entering on the close of bar3 (or on a move above bar 2’s high) of anintraday pivot, and risking a move tothe low of the entry bar.

H o w e v e r, if this re p resents too muchrisk and you prefer to trade shorter- t e r mprice bars, you can work backwards to setup a similar multiple time frame analysison the time frame of your choosing.

Let’s use five-minute bars as an exam-ple. Using the same approach of five barsof the short-term time frame per one barof the longer-term time frame, you woulduse 25-minute bars to define the tre n dand look for pivot lows and highs on thefive-minute bars to enter trades. (Using25-minute bars does leave one unaccom-panied five-minute bar at the end of thetrading day, however. )

Figure 7 (opposite page) shows five-minute bars encapsulated on a 25-minute basis. Each red rectangle repre-

S&P E-Mini (ES), 81-minute

1

2

3

1

2

3

12

3

9/5/03 9/8/03

1,030.0

1,028.0

1,026.0

1,024.0

1,022.51,022.0

1,020.0

1,018.0

1,016.0

Using 81-minute bars divides a trading day into five bars, reducing the num -ber of pivot highs and lows to one each.

FIGURE 5 THE 81-MINUTE BAR



1

2 3

1

2

3 2

1

3A

B

C 1

23

9/5/03 9/8/03

1,030.0

1,028.0

1,026.0

1,024.0

1,022.51,022.0

1,020.0

1,018.0

1,016.0

Increasing the time frame to 30-minute bars reduces the number of pivothighs and lows.

FIGURE 4 REDUCING THE NUMBER OF PIVOTS


sents a 25-minute bar and pivot lowsand highs on the 25-minute bars will beused to define the trend. The first 25-minute bar of the day gaps down, butstill qualifies as bar 2 of a pivot low. Apivot low is established when the mar-ket trades above the high of the 25-minute pivot bar 2, which signals a lowon the longer-term time frame and a ris-ing trend.

At this point, you would look for apivot low on the five-minute bars to golong. Again, you can buy as the marketmoves above the bar-2 high of the five-minute pivot low, with a stop at or nearthe low of the entry bar.

In this example the market didadvance until it formed a 25-minute barpivot high, after which it tumbled hard,forming only three five-minute pivothighs (two of which are contained in thesame 25-minute encapsulated bar) dur-ing the decline. These three pivot highswere the better trade setups — everyonewants to catch this kind of sharp decline— but such opportunities are rare.

Another 25-minute pivot low formedafter the sharp decline, signaling thetrend was turning back up, and a num-ber of pivot lows formed on the five-minute bars.

Don’t put the cart before the horseMany traders select a time frame andstruggle to make their trading fit thattime frame, rather than first decidinghow they want to trade and then pick-ing an appropriate time frame. Whenselecting a time frame on which to trade,make your decision based on a multipletime frame perspective.

In this case, dividing a longer-term(daily) time frame into five shorter-term(81-minute) periods and using a three-bar pivot pattern to define trends andsupport and resistance provided theframework for a simple trading ap-proach. Trading in the direction of thetrend of the longer-term time frame andusing chart patterns or indicators on theshorter-term time frame for entry sig-nals should increase your success.Ý

Encapsulation is a patented trademark ofRobert Krausz.

ACTIVE TRADER • December 2003 • www.activetradermag.com 64


1

2

3

AMoving average

Encapsulation

Pivot low

Pivot low

Pivot low

8/25/03 9/2/03 9/8/03

1,030.0

1,022.5

1,020.0

1,010.0

1,000.0

990.0

980.0

Here, 81-minute bars are “encapsulated” by rectangles that function astransparent daily bars allowing one to see the price action of two distincttime frames.

FIGURE 6 LOOKING AT MULTIPLE TIME FRAMES


S&P E-Mini (ES), five-minute

1

2

3

1 23

1

2 3

1

2

3

Trend is down

Trend is up

Trend is up

Pivot low

Pivot low

Pivot high

9/5/03 9/8/03

1,030.0

1,028.0

1,026.0

1,024.0

1,022.51,022.0

1,020.0

1,018.0

1,016.0

Using the approach from Figure 6, the five-minute bars here are encapsulat -ed by the 25-minute range, which is the time frame that defines the trend.

FIGURE 7 BACK TO THE FIVE-MINUTE TIME FRAME



TRADING Strategies

BY THOM HARTLE

The method traderThe trading world is filled with truisms and generalities, but there are no

magical indicators or secret recipes when it comes to trading well. Pr o f i t a b l e

trading is grounded in a process — how you a p p r o a c h your trading approach.

W hen in comes to consistently profitable trading, there are many roadsleading to the promised land, but as the saying goes, “Many are called,but few are chosen.”

New and struggling traders are always looking for something — an elusive insight ortechnique — that will put everything into focus and allow them to excel. Like virtuallyany other profession, though, the answer is not some bit of mysterious knowledge or aperfect trading tool. Proficiency boils down to three steps: Specialization, preparationand execution.

This should not come as a surprise. For example, when we need medical attention foran out-of-the-ordinary problem, we seek a physician who specializes in the field of med-icine specific to our affliction. The physician will prescribe a set of procedures, be theymedicinal, surgical or otherwise, that have shown a high probability of success in cur-ing the disease or treating the symptoms we have presented to them.

We rightfully expect when a physician prescribes a course of treatment he or she hasthe proper knowledge, training and skills to do so; and that the treatment is notunproven or experimental, unless we agree to it. Unfortunately, we do not set the samehigh standards for ourselves when we attempt to make money in the markets.

Think about what so many traders and investors do — place a trade based on the lat-est news event, after reading a newspaper article, or seeing something on television.Compare this to a visit to your doctor’s office. What would you think if your doctor, ifasked for details about the medication he was prescribing, could not provide any infor-mation regarding its efficacy because it was something new he’d just seen on TV thenight before?

Part of the problem is trading does not have the relatively uniform education, trainingand regulation that are the norm in most other professions. But that is not the primaryissue. The reality is traders, because they are in an essentially individual and entrepre-neurial business, have to take responsibility for everything they do. And the most criti-cal responsibility is knowing the probabilities associated with every trade they make.

Many people believe the key to making money is finding the perfect indicator or pat-tern. This is not the case. The key is knowing the probabilities of what the market willdo in the aftermath of any price pattern or indicator, and having set procedures to makethe most of that knowledge.

To do that, you have to work through a great deal of material to find techniques thatfit your needs. It’s the trading equivalent of putting yourself through medical school andsettling on your specialty.

Determining your expertiseIt takes a considerable amount of research before you can make an informed decisionabout your area of specialization. It is only after having acquired a broad background


continued on p. 67

Knowledge is experience,

and the essence of experience is self-reliance.

— T.H. White, The Once and Future King

that you can narrow your focus and select techniques thatfit your risk-reward temperament. This balance is an impor-tant issue to consider as you work your way through vari-ous trading approaches.

Your tolerance for risk and goals for reward will impactthe markets you trade and the time frame on which youtrade them. There is no “correct” time frame to trade; thegoal is to find the one you are most comfortable with.

Consider the differences between day traders and posi-tion traders. Day traders most likely cannot cope withovernight risk and want to be flat at the end of every trad-ing session, starting each day fresh. They do not like to seebig profits turn into small profits — or worse, losses —because of overnight events.

Position traders are more likely to consider intraday priceaction as meaningless noise — the product of the randomnature of the market. They are more at ease holding posi-tions and attempting to capture longer-term trends and lesscomfortable taking multiple positions during a trading day.

Regardless of the time frame, as you study various trad-ing concepts, market tendencies and patterns, you willbegin to identify some that appear to have consistentlyfavorable outcomes. In other words, you will recognize cer-tain aspects of market behavior and think, “I’ve seen thishappen before, and the market always seems to rally.” Nowyou’re onto something.

For example, you might be studying a technique that com-bines indicators and price patterns. Perhaps, you’ve deter-mined if the market is in an uptrend (based on a set movingaverage value) and price tags the lower 10-day BollingerBand and forms a pivot low (a three-bar support pattern),t h e re is a tendency for the market to rally (see Figure 1, top).Your goal is to determine the precise probabilities of this ten-dency and determine whether it merits trading.

There is a subtle but important point to absorb here. Thispattern has specific attributes that can be defined and test-ed (in the preparation phase). This is not always the casewith “setups” traders use. There is no lack of vague andambiguous trading concepts and rules: “Look for a break-out of a tight consolidation,” “Buy when a very large barforms and prices closes near the low,” and so on. But with-out precise definitions of what constitutes a “tight consoli-dation,” a “very large bar” and “near the low,” there is noway to identify past examples of these patterns and deter-mine the odds of what will happen after them. And withoutthat information, how can you trade?

In Figure 2 (left) for example, an upward 1x1 “Gann line”(which is a 45-degree trendline that is supposed to representprice moving in equilibrium with time) is drawn off a pivotlow (a three-bar support pattern consisting of a low with ahigher low on either side). A trader might have a rule that ifan entire bar’s range is above the upward 1x1 Gann line(signaling a rapidly rising trend), it indicates the trend maybe vulnerable to a reversal. Then, if a pivot high (the oppo-site of a pivot low) forms, go short if the market breaks therising 1x1 line.

However, a 1x1 Gann line reflects a price-to-time ratio —


Nasdaq 100 index-trackingstock (QQQ), 30-minute

1

23

1/15/04 1/16/04

Tagged the lower Bollinger Band

Rising long-termmoving average

Lower Bollinger Band

Upper Bollinger Band

38.50

38.40

38.30

38.20

38.10

38.00

37.90

37.80

37.70

37.60

37.50

37.40

The trade setup shown here — whether or not it turns out tohave merit — has an advantage over many trading ideasbecause it has specific, quantifiable attributes that can betested. (The pivot low is labeled “1, 2, 3.”)

FIGURE 1 TESTABLE TRADE SETUP


Nasdaq 100 index-trackingstock (QQQ), 30-minute

1/15/04 1/16/04

38.50

38.40

38.30

38.20

38.10

38.00

37.90

37.80

37.70

37.60

37.50

When considering a trading technique, you should always askyourself if it can be converted into precise, testable rules.This Gann-based technique would not pass the test.

FIGURE 2 GANN LINE NO.1


Entire bar above 1x1 line

Trend reversal

1

2

3

12 3 Pivot high

e.g., one point or price movement per one time unit — and thisratio can change simply on the basis of how a chart is con-structed. Figure 3 (right) is the same price chart, but drawn ona diff e rent scale. There are subtle but critical diff e re n c e sbetween this chart and Figure 2 — most significantly, the breakof the 1x1 line occurs at different points on the two chartsbecause of their different price scales.

As a result, depending on how your software plots charts,different signals would occur using the same trading “rules.”(The solution in this case would be an additional rule thatrequires the price-to-time ratio to always be a fixed number.This way, every chart will always create the same signals.) So,while working through a technique, you must always askyourself if it is something that can be converted to precise,testable rules.

An example of a testable one-bar pattern is: Today’s high isthe highest high of the past 10 bars and is at least 1.5 percentabove yesterday’s high, today’s low is below yesterday’s low,and today’s close is in the lowest 10 percent of the bar.

The ability to convert market patterns into precise defini-tions, which in turn can be translated into trade setups andprocedures, enables you to test their outcomes and determinetheir value. This determination is the goal of the preparationphase.

Preparation: Beyond “system testing”Testing trade setups is hardly a new idea, and there arenumerous software programs that expedite the process ofdesigning and implementing trading systems. Sometimes,however, a great deal of computer power can be a bad thing.

We can miss opportunities if we rely on standard system-testing programs to do all our analysis. Many new tradersessentially take a trading idea and “ask” the computer tocheck it out. If the summary test statistics — net profit, maxi-mum drawdown, etc. — indicate poor performance, a traderwill likely toss out the idea.

However, if the testing had been a little more hands on —for example, incorporating direct observation of the outcomeof trades on charts and extending the performance analysisbeyond the confines of a typical analysis program — then,after a little tweaking, a successful trading approach mightemerge after all.

All it takes to perform valuable analysis is a spreadsheetsuch as Excel. First, print out a series of charts with the signalsthat are the basis for your potential trades. Next, import thesame data into Excel, or at least the date, time (if intraday),open, high, low and close. Then you can set up columns to logvarious trade setups, such as “Setup 1-buy,” “Setup 2-buy,”“Exit 1,” “Exit 2,” etc.

Most importantly, you can create a column for the maximumfavorable excursion (MFE) and maximum adverse excursion(MAE) for each trade. The MFE is the difference between thesetup’s entry price and the position’s largest open profit; the

MAE is the difference between the entry price and the posi-tion’s largest open loss. With this information, you can deter-mine the typical outcomes of your trade setups.

Figure 4 (p. 32) shows an Excel spreadsheet containing theinformation for a sample trading approach. The first twocolumns contain the date and time, the next four contain thesetup’s indicator values, and the next four show the open,high, low and close prices.

The next column is used to enter the trade number (in therow of the bar on which the trade occurred). You could also logthis number on the appropriate bar on a chart, either a printedor electronic version. The setup for this trade, which is basedon a set of rules that indicate a purchase on the close of the bar,is labeled L-1.

The next two columns contain the MFE and MAE for the

trade. For example, trade number 35 occurred on the close ofthe 12:00 bar. The position reached its maximum open profit of18 cents on the 14:30 bar; this value is entered on the same rowas the trade number in the MFE column. The trade reached itslargest open loss of -6 cents on the next bar, and this MAE


continued on p. 69

The key to making money is knowing the probabilities of whatthe market will do in the aftermath of any signal generated by

a price pattern or indicator.

Nasdaq 100 index-tracking stock (QQQ), 30-minute

1/15/04 1/16/04

38.60

38.40

38.20

38.00

37.80

37.60

37.40

Because the Gann line is inconsistent and can changedepending on how a chart is constructed, the same priceaction shown in Figure 2 produces different results here sim -ply because the chart is drawn using a different scale.

FIGURE 3 GANN LINE NO. 2: SAME “SETUP,” DIFFERENT RESULT


Entire bar above 1x1 line

Trend reversal

Pivot high

1

2

3

1

23

value is entered in the same row in the MAE column. Because the trailing stop was not hit and positions are not

held overnight, the trade was exited at the close. Use the sameline for key information about each trade (the highlightedrows). This will help later when you want to sort the data bytrade number. The final two columns contain the trade’s profitor loss (P/L) and the reason for the exit.

The other three trades were based on a different setup (L-2)and two of them were stopped out using a trailing stop. Thelast trade shown was exited on the close.

Anatural question is how many trade examples are necessaryto draw meaningful conclusions about a trade setup. But them o re important issue is diff e rent types of market conditions.First, you should review a setup using price data that hasu p t rends, downtrends, trend reversals and trading ranges.Avoid testing in a period dominated by one price direction. Thatsaid, gathering information on at least 100 trades is a good start.

At this point you can begin to do some interesting work. Forexample, you can copy all the trades and paste their values inanother spreadsheet page and sort the data diff e rent ways. Forexample, you could sort by trade number, then by MFE to see ift h e re is a price target value that has a high probability of being hiton a regular basis. In Figure 4, the MFE was more than 10 cents fort h ree of the trades. If you had 100 trade examples, you might findthe market was making a favorable move of 10 cents at least 60p e rcent of the time. If the corresponding MAE is low — meaning,the trade isn’t going against you more than it is going in yourfavor — then you have a potentially viable trade idea.

S i m i l a r l y, you could sort by MAE and determine at what pointyou should abandon a trade because a majority of trades that hita particular MAE level do not recover to be winners. Other ideasto consider are sorting trades by the type of setup or time of dayto determine if there is a particular time that works best for a par-ticular setup.

None of this analysis can becompleted until you have spe-cialized on a trade approachand have converted it into pre-cise entry and exit rules. Butonce you have completed thesefirst two steps — specializationand preparation — you willhave a very clear understand-ing of the nature of your strate-gy. You might discover it is notas good as it initially appeared,and avoid rushing into themarket and losing money. Also,by manually working throughthe charts and logging theinformation in a spreadsheet,you have the opportunity todiscover a slight variation onyour original idea that may bean improvement. This is lesslikely to occur if you let thecomputer do all of the work.

“Pattern probabilities” (op-posite page) illustrates anotherway to use a spreadsheet to

m e a s u re the probabilities of trade setups and determine theirv a l u e .

Finally, all this work is an excellent way to develop goodskills for trading when the market is actually open. It is similarto a basketball player spending hours in the gym shootingnothing but foul shots so he is confident and behaves auto-matically during the pressure of a real game.

Once you have tested your ideas and identified high-proba-bility targets from the entry setups and risk points for manag-ing the trade, you are ready for the execution phase.

Implementing the strategyAt this point, you have a collection of procedures to follow andcan trade objectively and consistently. You have moved awayfrom trading by the seat of your pants, jumping from one tech-nique to the next and wondering why you experience incon-sistent results using a patchwork collection of strategies.

With this framework you no longer have to concern yourselfwith forecasting the markets; simply follow your tested pro c e-


Additional researchThe MFE and MAE analysis concept is from John Sweeney’sbook, Campaign Trading: Tactics and Strategies to Exploitthe Markets (1996, John Wiley & Sons).

Trading in the Zone by Mark Douglas (2001, Prentice HallPress) is a good source of information on the value ofthinking and trading in probabilities.

“On-target trading,” Active Trader, July 2001, p. 44.

“Taking the guesswork out of stop orders,” Active Trader,October 2001, p. 58.

Spreadsheet analysis allows you to perform a great deal of hands-on analysis. This exampleshows the trade date and time, indicator values, price data, trade number, setup, MFE andMAE values, profit or loss, and reason for exit. After compiling this information for all yourtrades, you can then sort the data and perform in-depth analysis that reveals key informa -tion about profit targets, stop levels and other aspects of a strategy.

FIGURE 4 AN EXCELLENT SOLUTION

Source: Excel

continued on p. 70

Another way to use a spreadsheet is to determine theprobabilities of a price pattern before you evendesign specific rules for trading it.

Figure 5 (below) shows an excerpt from a spreadsheet thatshows several statistics — the average price move, medianprice move, the maximum price move and the minimum pricemove — in the three days following the completion of a pricepattern.

To perform this kind of analysis, you must first import theopen-high-low-close price data for the period you wish to test(the “open” column is hidden here). The period here spansMay 31, 2000, to Dec. 13, 2000; many of the rows are hiddento conserve space.

Column F contains the pattern’s conditions. In Excel, it iseasy to string together several “If” conditions that describe apattern. In this case, the pattern is a bar with three conditions:

1 . The high must be at least one percent above the previoush i g h ;

2 . The close must be below the previous close; 3 . The close must be in the bottom 10 percent of the price

b a r.

These conditions are shown in the formula bar (for cell F22) as:

1. (C22-C21)/C21>0.01 2. E22<E213. (E22-D22)/(C22-D22)<0.10

Each condition is preceded by the “IF” function, and the“1,0” at the end of the argument indicates that if all the con-ditions are true, “1” will be entered in the appropriate cell inthe F column; if even one of the conditions is false, “0” willbe entered. By dragging this formula to fill all the cells in col-umn F, each bar that fulfills the pattern criteria will beflagged with a 1. (Alternately, if you have programmed thepattern conditions into your analysis software, you will beable to automatically include this information when youoffload the price data to the spreadsheet.)

Columns G-O contain the close-to-close price moves, MAEand MFE values for the three days after each pattern occur-rence. Dragging the formulas makes calculating the numbersfor the sheet a very simple process. This analysis could be car-ried out for as many days as you wish. Also shown is the totalnumber of patterns that occurred.

For example, row 22 holds the data forthe Oct. 24, 2000, pattern and shows themarket dropped 1.89 percent the dayafter the pattern (day 1), had an MFE of.69 percent and an MAE of -2.40 percent.By the close of day 2, however, the mar-ket had risen 1.57 percent, and so on.

The summary statistics at the bottomallow you to judge the probabilities ofthe raw pattern signal. These are onlyexamples of the kinds of statistics youcould include. Others are the percentageof positive and negative returns at eachtime interval, a separate breakdown ofpositive returns and negative returns,and so on.

If a pattern’s probabilities are favor-able, you can then proceed to developingand testing trading rules to maximize thepattern’s potential.

— Active Trader Staff

Column F contains the pattern criteria. The remaining columns show theclosing price differences, and MFE and MAE values for qualifying price bars.

FIGURE 5 BASIC PATTERN TEST

Source: Excel

Pattern probabilities


d u res. This will relieve you of the psychological pre s s u re ofbeing right or wrong on any one trade. You will be trading basedon probabilities that have been proven over dozens of trades.

But that doesn’t mean the work is over.

Going forwardIt is important to continue to update your spreadsheet withreal-time trades. Markets change, and if you begin to seeresults that are inconsistent with your original analysis, youcan spot this before it becomes a real problem.

The three steps outlined here — specialization, preparationand execution — reflect the same basic process that carries peo-ple in a wide range of disciplines from the novice to profes-sional stage. Moving from broad-based approaches to special-ized techniques that are converted to procedures with proba-bility-based outcomes puts you in a position to succeed.

It takes a lot of work, but the end result is having the skillsto be consistent, regardless of the trading approach you use.Ý


N A S D A Q 1 0 0 V O L U M Eand the Q Q Q

One way to gauge how the market

might move is to analyze its

characteristics when it’s going nowhere.

TRADING Strategies

U nchanged volume is the volume of stocks withprices that are unchanged from the previousday’s close. It is one of three components of thetotal trading volume provided by the New York

Stock Exchange and the Nasdaq. The other two are up volume,which is the total volume of those stocks currently up on theday, and down volume, which is the total volume for stocksdown on the day.

“ Trend, consolidations and unchanged volume” (A c t i v eTrader, October 2003, p. 44) analyzed the NYSE unchanged vol-ume to see if it highlighted trading opportunities in the S&P500. The research showed high unchanged volume readingscorresponded to the market being in a balanced state, or whattypically takes the form of a congestion period. Seasonedtraders know trends emerge from such periods, and if the mar-ket as a whole is congested at the close of one day, it is worth-while to anticipate a trending day the next day.

Because of its popularity, the price action of the Nasdaq 100index-tracking stock (QQQ) and the unchanged volume of theNasdaq Exchange will be the focus of this analysis.

Choosing a measuring stickThe first step in the analysis is to determine what constitutes ahigh unchanged volume reading. We looked at 500 daily read-ings of the unchanged volume for the Nasdaq beginning onMarch 25, 2002, and ending on March 17, 2004. The lowest was6.2 million (Feb. 6, 2004) and the highest was 269.6 millionshares (July 24, 2002).

By sorting the values from lowest to highest it was possibleto determine the highest 25 percent of unchanged volumereadings were above 35.01 million shares. Let’s look at somehigh unchanged readings and the price action in the Nasdaq

Nasdaq 100 index-tracking stock (QQQ), daily

Nasdaq unchanged volume

A B

C

D

E

F

G

AB

C

D

EF

G

20 26 Feb. 2 9

39.00

38.50

38.00

37.50

37.00

36.50

100 mil.

75 mil.

50 mil.

25 mil.

The price of the Nasdaq 100 index-tracking stock isshown here with the level of unchanged volume below.

FIGURE 1 UNCHANGED VOLUME

Source: CQGNet Inc.

BY THOM HARTLE

100 index-tracking stock the next day tosee if there are any useful patterns.

Sizing up the marketFigure 1 shows the price for QQQ in thetop window and the closing value forNasdaq unchanged volume in the bot-tom window between Jan. 20 and Feb. 9,2004. Bars A, B and C each had consecu-tive unchanged volume re a d i n g sexceeding 35 million shares.

Bar A was a day the open, high, lowand close were all below the previousday’s respective open, high, low andclose. It would be easy to conclude fromthe price action that Bar A was a weakday relative to the previous day’s priceaction. However, the high unchangedvolume reading indicated the sellingwas not widespread for the broad mar-ket.

Bar B climbed 3 cents above bar A’shigh, then reversed to close below thep revious day, although the low wasabove the previous day’s low. However,the unchanged volume was again abovethe 35-million-share threshold. Over thenext two days the market traded up andthen down, with low unchanged vol-ume readings that indicated broad mar-ket participation. The unchanged vol-ume readings here did not portendmuch besides the days’ ranges expand-ing after the high unchanged readings.

Bar D was the second of two bigdown days, and while the market closednear its low, the high unchanged vol-ume reading indicated the broad marketre c o v e red more than the QQQ.Following a higher opening the nextday, the QQQ fell an additional 35 centsbelow the previous day’s low beforerecovering and closing higher on theday.

Bar E was an outside day (a bar witha higher high and lower low than the


continued on p. 73


Lowest HighestUnchanged volume

1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

The rising regression line reflects the increased upside price movement asthe unchanged volume level increases.

FIGURE 3 UPSIDE FOLLOW-THROUGH


The low is exceeded

2.00

1.50

1.00

0.50

0.00

-0.50

-1.00

-1.50

This diagram shows how far the QQQ rallied above the highs and fell belowthe lows of days with 35 million or more shares of unchanged volume.

FIGURE 2 PUSHING ABOVE THE HIGH, BELOW THE LOW

Sources for this and following figures: charts — Excel; data — CQGNet

previous bar), followed by bar F, aninside day (a bar with a lower high andhigher low than the previous bar). Bothdays had very high unchanged volumereadings. The next day, the QQQgapped down on the open and closed atits low.

Bar G was an inside day, and the nextday, when the market broke above barG’s high, the rally continued an addi-tional 57 cents.

What does it mean?In this example there were two clustersof high unchanged-volume days, one ofwhich (bars E and F) preceded a goodf o l l o w - t h rough day. There also weretwo, single, high unchanged-volumedays (bars D and G), both of which werefollowed by bars that exceeded the lowand high by 35 cents and 57 cents,respectively.

It appears if the high or low of a barwith high unchanged volume is violat-ed, some follow-through price actioncan occur. Let’s look at the data and seewhat the typical price action is follow-ing a high unchanged-volume reading.

Of the 500 days, using the top 25 per-cent of unchanged-volume readings asthe cutoff, there were 125 days thatexceeded 35 million shares of unchangedvolume. Of those 125 days, the next daywas an inside day 10 times and an out-side day 21 times. Figure 2 (p. 72) dis-plays the price moves for those days thehigh was exceeded and those the lowwas exceeded. (We calculated the diff e r-ence between the two highs or two lows;inside days were excluded.)

The average for the days the high wasbroken was a 32-cent move above theprevious high. If the previous day’s lowwas broken, the average move was 36cents below the previous day’s low. Ifthe outside days are removed, the aver-age move for those days that pushabove the previous day’s high climbs to36 cents. The average break below theprevious day’s low increases to 43 cents.

Figure 3 (p. 72) sorts the upside fol-l o w - t h rough moves according to theunchanged volume levels (from lowestto highest) to see if there is any correla-tion between the level of unchangedvolume and the follow-through move-ment when the high is broken. Notice


1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

After a high unchanged-volume day, the QQQ exceeded that day’s high by 7cents or more 75 percent of the time.

FIGURE 5 MOVES ABOVE THE PREVIOUS HIGH

The QQQ fell 16 cents or more 75 percent of the time after breaking belowthe low of a day with high unchanged volume.

FIGURE 6 MOVES BELOW THE PREVIOUS LOW

As was the case when the high was exceeded, the higher the unchanged vol -ume, the more the QQQ followed through on moves below the low.

FIGURE 4 DOWNSIDE FOLLOW-THROUGH

The low is exceeded

Lowest HighestUnchanged volume

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

the regression line rises slightly from leftto right, indicating there is a slight biasfor the amount of upside price action asthe unchanged volume climbs.

Figure 4 tells a similar story regardingwhat happens when the low is penetrat-ed. Again, the regression line rises (at aslightly higher rate compared to theprice action when the high is broken),indicating a correlation between highunchanged volume and follow-throughin the direction of the breakout the nextday. The fact the average break when thelow is violated is more than the averageprice move when the high is broken (andthe slightly higher rate as the unchangedvolume level climbs) reflects how mar-kets tend to fall faster than they rise.

F i g u re 5 shows the price movesbeyond the previous day’s high sortedfrom lowest to highest. For those daysthe high was violated, the high was 7cents or more (up to $1.64) above theprevious day’s high 75 percent of thetime.

F i g u re 6 sorts the drop below the pre-vious day’s low from lowest to highestfor days following bars with unchangedvolume of 35 million or more. When thelow was broken the market fell 16 centsor more (up to $1.21) 75 percent of thetime — again, more evidence of howquickly markets can dro p .

Trading implicationsOne thing that stands out is the opportu-nity for short-selling traders: A breakbelow the previous day’s low followinga day with unchanged volume of 35 mil-lion or more resulted in at least 16 centsfollow-through 75 percent of the time.

Day traders should pay special atten-tion to those days that follow highunchanged-volume days because therewere only 10 inside days out of 125.Therefore, you can expect the high orlow of a day with high unchanged vol-ume to be violated the next day.You canuse your favorite setups to take advan-tage of this if the market opens withinthe previous day’s range.

This research should be kept up todate, as market volatility can change.These numbers could be very different ayear from now.Ý

For information on the author see p. 10.


Thom Hartle - Active Trader Magazine - Trading Strategies Analysis Collection Vol1

Documents

Transcript of Thom Hartle - Active Trader Magazine - Trading Strategies Analysis Collection Vol1