Encyclopedia of Chart Patterns. Thomas N. Bulkowski

at B and it'll resemble a bat (or butterfly or birds or lots of other flying creatures, I think). But this pattern is called a bullish bat, so that's the story I'm sticking with.

      Without the lines, the pattern looks like a big M chart pattern or even a double top with tall sides. The difference between the bullish bat and those other patterns is the use of Fibonacci ratios to determine the turning points, so let's discuss that next.

AB/AX retrace. Let me give an example of how to find the various points. Points X and A are arbitrary minor low and high points, respectively. Minor low B is an important one. The ratio of leg AB to AX should be either .382 or .5. Because the pattern is so complicated, the rules determining valid ones limit the numbers you will find in the real world. So I used the high–low price range of one of the points in the ratio. If a Fibonacci number is within the high–low range, then it's allowed as a valid turning point.

Table 5.1 Identification Guidelines

Characteristic	Discussion
Appearance	Looks like a big M with turns located by Fibonacci ratios.
AB/AX retrace	The ratio of AB/AX is either .382 or .5
CB/AB retrace	The ratio of CB/AB is one of .382, .5, .618, .707, .786, or .886
CD/CB extension	The extension of leg CD to CB is one of the Fibonacci numbers: 1.618, 2, 2.24, or 2.618.
AD/AX retrace	The ratio of AD to AX is .886.
Volume	Volume is downward the majority of the time, but this is an observation, not a requirement.
Duration	I limited patterns to 6 months, but this is an arbitrary limit I use for many chart patterns.

Figure 5.2 This bullish bat rises after point D, but breaks out downward at E.

      For example, the low at point X is 9.76. The high at A is 12.90, and the high–low range at B is 11.81 – 11.40. The AB/AX ratio using the high at B would be (12.90 – 11.81)/(12.90 – 9.76) or .35. Using the low price at B would give (12.90 – 11.40)/(12.90 – 9.76) or .48. The .382 number nestles comfortably between .35 and .48, so XAB is a valid turn.

      CB/AB retrace. In a similar manner, I use the high at A, low at B, and high–low range at C (12.57 to 11.98) to qualify the CB/AB retrace. Plugging in the numbers we get a range of .39 to .78. That range must encompass one of the numbers listed in the table, and it does. In fact, several of the numbers listed fall within that range. Although such a large range might seem like the recognition rules are lax, only a few patterns qualify. Let's continue with the CD extension.

      CD/CB extension. CD is longer than CB, so we call that an extension and not a retrace. Let's go through the math for this one. It follows the same rules as before. I use the low at B, high at C, and the high–low range at D to qualify the three points. D is at 10.81 to 10.19. Plugging them into the equation, we get (12.57 – 10.81)/(12.57 – 11.40) or 1.50 using the high at D. The low at D becomes: (12.57 – 10.19)/(12.57 – 11.40) or 2.03. Two numbers fit that range: 1.618 and 2, so the turn qualifies.

      AD/AX retrace. For the last turn of the bat, I don't use the high–low range. Rather, the number must be within .03 of the target .886 value (using the low at D only). So we get (12.90 – 10.19)/(12.90 – 9.76) or .86, which is within .03 of the .886 target. This pattern qualifies as a valid bullish bat, but your software may use a different algorithm and find a different result.

Volume. The bullish bat doesn't have a volume requirement. Rather, you'll see volume recede from point X to D (the width of the pattern) most of the time. I'll discuss volume in the Statistics section. Point F in the figure shows a downward volume trend.

      Duration. There are no rules limiting the width of the pattern, but I limited them to six months.

Focus on Failures

      Figure 5.3 shows a bullish bat that fails to be bullish. I show the five turns at XABCD. It's a valid bullish bat even though it looks weird. Price at D should see the stock turn higher, but it doesn't. It does form a minor low, but that doesn't translate into any meaningful rebound in the stock.

      Instead of turning upward at D, the stock continues lower, breaking out downward at E. Price continuing lower at D (instead of turning upward) is the first type of failure.

      The second type of failure is when price does turn upward at D, but doesn't climb far. I consider a move no more than 5% to be a failure. Figure 5.3 shows this type of failure, too, because price climbs just 3% after D.

      Five percent failures only happen about 10% of the time, so they are a concern, but not a huge one.

      Let's talk about statistics next.

Figure 5.3 This bullish bat isn't bullish at all. Price fails to turn higher at D and breaks out downward from the pattern.

Table 5.2 General Statistics

Description	Bull Market
Number found	259
Breakeven failure rate	10.2%
Average rise after D	44.3%
Volume trend	80% Downward
Performance Up/Down volume	35%U, 47%D

Statistics

Table 5.2 shows a few general statistics for the bullish bat.

      Number found. This complicated pattern is rare, but it could be the result of the way I programmed my computer to find them. I unearthed bats (in both bull and bear markets) in only 260 stocks despite searching over 1,300 stocks. Patterns found in the stocks stretched from July 1991 to April 2019. There were so few patterns that I limited the presentation to bull markets only, as I mentioned.

      Breakeven failure rate. The breakeven failure rate is a measure of how often patterns fail to rise more than 5% (as measured from the low at D). With such a high average rise (44.3%), you might expect to see a low failure rate. Indeed, that's what the pattern shows, at 10.2%.

      Average rise after D. The average rise measures from the low at point D. The 44.3% rate compares favorably to the 42.4% average rise from non‐Fibonacci‐based patterns.

      Volume trend, performance. Volume trends lower 80% of the time as measured from the start of the pattern to the end using linear regression. Performance is best if the trend is downward (with rises averaging 47%!).

      However, don't let the wide difference (35% and 47%)