Encyclopedia of Chart Patterns. Thomas N. Bulkowski

so it qualifies as a valid extension.

      DA/XA retrace. Supposedly, this ratio is critical, so I allow point D to miss the number by 3% (.03). I do not use the high–low range of D in the equation and I use 3% instead of 1% because this pattern is rare enough as it is. We have (55.98 – 47.81)/(56.93 – 56.93) or .9. The .9 value is near the .886 target, so the pattern qualifies as a valid bearish bat.

      As you can imagine, computing these ratios by hand is tedious and error prone, so I programmed my computer to find bats by first finding all peaks and valleys and then using the method described to glue them together.

      Volume. I'll discuss a few volume statistics in Table 4.2, but volume trends downward in the pattern most of the time. With few samples, it's difficult to say whether up or down trending volume leads to better or worse performance, but current statistics favor a downward volume trend for the best performance.

      Duration. I limited patterns to a length of 6 months or less. That's an arbitrary limitation.

Table 4.2 General Statistics

Description	Bull Market	Bear Market
Number found	537	128
Breakeven failure rate	17.7%	4.5%
Average decline after D	–14.3%	–20.2%
Volume trend	74% Downward	67% Downward
Performance Up/Down volume	–14%U, –15%D	–20%U, –20%D

Focus on Failures

Figure 4.3 shows a failure of a bearish bat pattern. Why is it a failure? As the name implies (bearish bat, not bullish), price should drop after point D (meaning price should trend lower after the pattern ends). However, price continues trending higher in this example. Price closes above the top of the pattern (X) at point E, posting an upward breakout.

      I checked the statistics and found that 70% of the 665 bats I looked at break out upward, regardless of the market condition (bull or bear). Don't let that number give you nightmares. As awful as the 70% number is, it doesn't tell the complete story. Swing traders will soon learn (as discussed in Table 4.2) that price turns lower at point D 86% of the time. Price may still break out upward, as Figure 4.3 shows, but maybe you can make money when price drops between D and the breakout.

      Figure 4.3 This bearish bat isn't bearish at all.

I don't show this situation in Figure 4.3, but if price does turn lower at D but drops less than or equal to 5%, then that's also a failure. If you were to trade a pattern with a 5% failure, you'd be hard pressed to make money. The 5% failure rate (also called the breakeven failure rate) is 17.7% for this pattern in bull markets (but just 4.5% in bear markets).

      Since I'm throwing around a lot of numbers, let's discuss statistics.

Statistics

      Table 4.2 shows general statistics related to the bearish bat pattern.

      Number found. Despite using the high–low price range at most of the Fibonacci turns, as I discussed in the Identification Guidelines, few bats appear in the historical price record. I found them in 491 stocks despite searching from June 1991 to August 2019. Not all stocks covered the entire period and some no longer trade.

      Breakeven failure rate. I counted the number of patterns which saw price drop no more than 5% after peaking at D (using the high price at D to the ultimate low). The table shows the failure rate. Anything above zero is too high, but that's in an ideal world. The bear market failure rate, at 4.5%, is terrific. The 17.7% failure rate in bull markets is not so terrific.

      When you consider that a declining price trend in bull markets is like swimming against the current, you would expect a high failure rate. Swim with the bear market current and the failure rate drops to one‐fourth the bull market rate.

      Average decline after D. Point D ends the pattern and price is supposed to head lower. The average decline is 14.3% in bull markets but substantially better in bear markets, over 20%.

      Volume trend, performance. I measured the volume trend using linear regression but often you can tell the trend just by looking at the chart. Figure 4.2, for example, shows volume (E) higher on the left half of the pattern than the right half, so the trend is downward. In fact, the average bearish bat will have a downward volume trend at least most of the time.

      Does the volume trend give us any hint of better or worse performance? Not that you'd notice. The widest spread between the two numbers is one percentage point in bull markets. That's probably not statistically significant.

Trading Tactics

      The real purpose of the bearish bat is for swing traders to predict the turn at D and then go short, anticipating a drop to the bottom of the pattern. Will price sink that far? Maybe, maybe not. Let's look at some number to see if we can model behavior better.

Table 4.3 Price Move after Pattern End

Description	Bull Market	Bear Market
How often does price turn at D?	86%	86%
How many drop to point A?	35%	38%
How many drop to point B?	81%	80%
How many drop to point C?	48%	46%

Table 4.3 uses numbers to show how price behaves after the pattern completes at D.

      How often does price turn at D? At the end of the pattern, price can form a minor high at D or price can continue trending upward (as in a straight‐line run up). Most of the time (86%), the stock will form a minor high at D. To find the reversal rate, I checked for a minor high appearing there (that is, I checked to see if price actually turned lower). Figure 4.3 shows an example where the stock failed to turn significantly lower at D.

      How many drop to…? This measures the drop after D but uses the pattern's turns as targets. For example, I found that in bull markets, the stock reached turn B 81% of the time. Turn B is the closest turn below D, so it should,