Counting Crows

"Strong stocks tend to rebound sharply after experiencing 3 to 5 consecutive down days...After a 3 to 5 bar drop, the master trader buys the very next time the stock trades above a prior bar's high."

Velez's successful Pristine Trading System is based on this concept.

This three-day drop is, of course, another way of defining the Japanese candlestick three-black crow formation (see sidebar "Three black crows"), but in the "ancient investment techniques of the far east," as candlestick expert Steve Nison refers to it, this formation indicates a reversal. In fact, according to various candlestick investors, it is one of the most reliable of all candlestick formations, the drawback being that it rarely occurs. From Nison's book Japanese Candlestick Charting Techniques:

"If there are three declining consecutive black candles, it is called three black crows pattern. The three black crows presage lower prices if they appear at high price levels or after a mature advance... Likened to the image of a group of crows sitting ominously in a tall dead tree, the three crows have bearish implications. The three lines should close at, or near, the lows. Ideally, each of the openings should also be within the prior session's real body."

Sidebar 1:
THREE BLACK CROWS PATTERN

So which is it? A buy or a short? Ancient Japanese wisdom or high-tech computer programming? I thought it would be easy enough to check, but I was wrong. I found it extremely difficult to find or program a good screen for three black crows. What is visually easy to find turned out to be daunting to program. I could be looking right at a three black crow formation and be unable to program MetaStock to identify it. Using code that defined large candlesticks as varyingly bigger percentages of the average trading range for the past seven to 21 days, plus an uptrend using linear regression, turned out to be too tricky for me. See sidebar "Sharon three blackcrows + linear regression."

SIDEBAR 2:
SHARON THREE BLACKCROWS + LINEAR REGRESSION

Staff Writer Dennis Peterson eventually created this code for three black crows for MetaStock. You can find the pattern in an uptrend using this code, although it only finds one per stock at a time.

body1:=C-O;
body2:=Ref(C,-1)-Ref(O,-1);
body3:=Ref(C,-2)-Ref(O,-2);
crow1:=body1<0 AND body2<0 AND body3<0;
crow2:=Abs(body3)>Abs(Ref(C,-3)-Ref(O,-3))AND
     Abs(body3)>Abs(Ref(C,-4)-Ref(O,-4))AND
     Abs(body3)>Abs(Ref(C,-5)-Ref(O,-5))AND
     Abs(body3)>Abs(Ref(C,-6)-Ref(O,-6))AND
     Abs(body3)>Abs(Ref(C,-7)-Ref(O,-7))AND
     Abs(body3)>Abs(Ref(C,-8)-Ref(O,-8))AND
     Abs(body3)>Abs(Ref(C,-9)-Ref(O,-9));
crow3:=LinRegSlope(C,10)>0;
sharon:=crow1 AND crow2 AND crow3;
sharon

Cautioning they were only preliminary, Bulkowski found that three black crows didn't indicate a reversal when the market indexes were in an uptrend but were significant in a downtrend. Since more black crow formations occurred in a downtrend, that made them indicative of a reversal.

So I had a preliminary hypothesis. Three black crows are significant, particularly so when the market is in a downtrend. I then turned to StockCharts.com, a site that screens for candlestick formations. Greg Morris, a prominent name in candlestick methodology and author of Candlestick Charting Explained and CandlePower, developed the candlestick scans available there.

On April 17, StockCharts.com found four three-black crow formations. Figure 1 shows the scan. The three under NASDAQ were penny stocks, too small for me to trade. The one under NYSE was PHH (Figure 2).

FIGURE 1: StockCharts.com candlestick scans for April 17, 2007.

FIGURE 2: PHH showing the three black crow formation from the StockCharts.com scan.

Looking for stocks in an uptrend, defining "large" candlestick bodies as at least twice the 14-day average range with three consecutive falling open/closing prices, and a 10% candlestick body overlap yielded nothing over the past year on the Standard & Poor's 500. Whereas the scan modified by StockCharts.com found one or two a day per exchange, it would be useless for a website like StockCharts.com -- although not necessarily for an individual -- to have a scan that only found one likely candidate every few months. With that in mind, I went back a year and built a spreadsheet of the three black crows from their scan.

METHODS
From February 3 to December 29, 2006, there were 599 black crow patterns, not including penny stocks. My goal was to find out if the black crow pattern was predictive when combined with market trends and stock trends. To do this, I set up a spreadsheet in Excel (Figure 3) with the following conditions:

FIGURE 3: SAMPLE SPREADSHEET. A simple record of three black crow occurrences and consequences.

FIGURE 4: Daily chart of the DJIA showing the market's trend.

2. Stock and trend. These were defined as a series of higher highs and higher lows established over a minimum period of four weeks for uptrends and the reverse for downtrends. If the stock wasn't clearly in an up- or downtrend, I rated them as "none."

3. Results. I indicated whether the stocks reversed or continued their previous trend, or consolidated two months following the black crow pattern. I gave them some leeway to establish direction. The pattern didn't have to appear at the top of a reversal.

4. Trade. I set up a 10% reward versus 5% risk trades and noted whether I would have taken the trade and whether it was long or a short. This was my version of back-testing.

5. Comments. Personally, I put in rough entry and exit points for the trades. I had no set standards for deciding entries, except on the day after the three black crow formation that they be conservative, and close to round numbers so the math would be quicker. The exits were either from the 5% trailing stop-loss or after a large gain, again arbitrary.

Figure 5 shows the occurrences of three black crows by week with the DJIA in the background for comparison. The total number of occurrences are shown in the bars and broken out by results, whether the stock was shorted (yellow), bought (red), or not traded (blue). The black crows peaked just prior to the reversal in the DJIA, the DJIA then being in a mature uptrend. On the reversal, occurrences lessened until the trend reversed and broke to the upside. Another short-lived spike formed, then occurrences slowly declined. This may have been seasonal going into the year-end holidays, as well as a characteristic of a renewed uptrend.

FIGURE 5: Three black crows results with the DJIA in the background.

Falling stocks at the beginning of a new and powerful uptrend may be sparse. Similarly, the number of occurrences increased right before the reversal. These signals are the types expected from a momentum indicator.

There were 137 total successful shorts, the majority occurring at the top of the DJIA's mature uptrend and during the downtrend, and hardly any at the beginning of the new trend. The 192 long plays showed the opposite tendencies, the highest percentage occurring at the beginning of the new trend, and the least during the downtrend. Throughout the period, the highest number of black crows, 270 out of 599, resulted in no trade, with the majority occurring in trendless stocks.

By that point I believed there was a pattern to the black crow formation that seemed intuitively correct. I wanted to find out if it was predictable. I had stocks in various trends from StockCharts.com's scan but didn't know whether to short, go long, or leave them alone. I then tabulated the occurrences by trend -- cross-referencing the DJIA trend to the stock trends and tabulate the results, whether the stock rose, consolidated or fell following the black crow formation. Figure 6 shows the results.

FIGURE 6: Black crow occurrences by market direction and stock trend and the resulting trades. Significant percentages are highlighted in blue.

There was one more play worth noting -- even though the percentages of shorting success are under 50%, the gains (not shown) are considerably larger on a short than a long, again reinforcing the tenet that the market goes down faster than it goes up. Therefore, the huge jump in shorts in week 10 may have yielded good profits, or at least signaled the impending reversal and the possibility of shorting the black crow pattern.

CONCLUSION
So what of the bearish implications of the candlestick pattern? Returning to the words of Steve Nison, who states, "The three black crows presage lower prices if they appear at high price levels or after a mature advance..."

There it is: High price levels or after a mature advance. The black crows found by StockCharts.com have to be put in their proper context before they're usable. If they appear at the opposite end of the spectrum -- low price levels prior to an advance -- they can be used as buys following Oliver Velez's trading strategy. Once they occur, buy in on the following day that the stock closes above the previous day's high. That is, wait until the stock stops going down before buying in.

As an example of its use, on May 1, 2007, there were six black crow patterns in StockCharts.com's scan. Of those, I liked Constellation Energy Group (CEG) (Figure 7) for a long trade.

FIGURE 7: Six-month daily chart of CEG.

The stock was bouncing off the 20-day moving average, and the drop from the black crow pattern presented a perfect entry point. The stop would be slightly below the 20-day moving average with the target at $95. There's an immediacy and simplicity to the pattern that makes it worthwhile to review.

There are a number of naysayers as well as studies on the effectiveness of Japanese candlestick patterns that take them at face value and plot gains made on the following day. Those studies find nothing significant in the patterns as the stocks have no preference for either up or down movement following their occurrences. But by placing them within a larger context, their value as an indicator becomes apparent. Far from being the Far Eastern indicator that transcends stock market analysis, the three black crows fit hand-in-glove with technical analysis methodology; you only have to know how to use them. Where have I heard that before?

SUGGESTED READING
Bulkowski, Thomas [2000]. Encyclopedia Of Chart Patterns, John Wiley & Sons.
Morris, Gregory L. [1995]. Candlestick Charting Explained: Timeless Techniques For Trading Stocks And Futures, McGraw-Hill.
Nison, Steve [1991]. Japanese Candlestick Charting Techniques, New York Institute of Finance/Simon & Schuster.
Velez, Oliver [2007]. Strategies For Profiting On Every Trade: Simple Lessons For Mastering The Market, Marketplace Books.