Читать книгу Encyclopedia of Chart Patterns - Thomas N. Bulkowski - Страница 108
Statistics
ОглавлениеTable 8.2 shows general statistics for the broadening bottom chart pattern.
Number found. I dug up 1,238 patterns in 667 stocks starting from August 1991 to September 2019 but removed the bear market ones because they were too few when sorted by breakout direction. Not all stocks covered the entire range, and some no longer trade. Both up and down breakouts are for bull markets.
Reversal (R), continuation (C) occurrence. By definition, a bottoming pattern has price entering the pattern from the top. A pattern acting as a reversal sends price out of the pattern upward (reversing the downtrend). A continuation pattern breaks out downward (continuing the downtrend).
Average rise or decline. Price posts a 45% rise after an upward breakout, helped along by a bullish general market. Downward breakouts suffer when price tries to drop in a bullish market. That's like swimming against the current.
Table 8.2 General Statistics
| Description | Up Breakout | Down Breakout |
|---|---|---|
| Number found | 599 | 405 |
| Reversal (R), continuation (C) occurrence | 100% R | 100% C |
| Average rise or decline | 45% | –15% |
| Standard & Poor's 500 change | 14% | –2% |
| Days to ultimate high or low | 240 | 47 |
| How many change trend? | 52% | 29% |
Standard & Poor's 500 change. The performance of the index can't compare to the performance posted by the chart pattern. That's typical, but I'm not sure it's a fair comparison. We're comparing a perfect trade to the index, using the same dates as the broadening bottom trade. It does show that broadening bottoms can beat the indices, and it also shows that the general market helps performance.
Days to ultimate high or low. This is a measure of how long price takes to reach the ultimate high or low (after the breakout). For upward breakouts, it'll take about 8 months of worry to reach the ultimate high. Downward breakouts take about 6 weeks to drop 15%.
If you compare the ratio of 45% in 240 days to 15% in 80 days, we discover that price drops nearly twice as fast as it rises. That might be a hint for options traders. You might be able to reach your price target faster during a downtrend than an uptrend.
How many change trend? Over half of broadening bottoms with upward breakouts see price rise more than 20% after the breakout (which is good). Downward breakouts suffer, with only 29% dropping more than 20%. The best patterns see price forming strong and lasting trends.
Table 8.3 shows failure rates. How do you measure failure? It took a while before I was able to answer that. I measured the move from the breakout price to the ultimate high or low and sorted the results into bins. Then I counted how many entries I had in each bin. It's like sorting coins you found under the seat cushions into piles of dimes, nickels, and quarters, and then counting how many dimes you found, and how many nickels, and so on.
If the breakeven cost of trading is 5%, then we see that 16% of the patterns with upward breakouts will fail to see price rise more than 5%. Downward breakouts are worse, with 26% of them failing to drop more than 5%.
Notice how the failure rates climb. Almost half (46%) of downward breakouts see price drop no more than 10%. Ouch.
Table 8.3 Cumulative Failure Rates
| Maximum Price Rise or Decline (%) | Up Breakout | Down Breakout |
|---|---|---|
| 5 (breakeven) | 98 or 16% | 106 or 26% |
| 10 | 71 or 28% | 80 or 46% |
| 15 | 60 or 38% | 67 or 62% |
| 20 | 56 or 48% | 36 or 71% |
| 25 | 33 or 53% | 34 or 80% |
| 30 | 34 or 59% | 25 or 86% |
| 35 | 26 or 63% | 26 or 93% |
| 50 | 55 or 72% | 27 or 99% |
| 75 | 68 or 84% | 4 or 100% |
| Over 75 | 98 or 100% | 0 or 100% |
What does this mean? Trading chart patterns and expecting a huge gain is unrealistic. Large gains happen, sure, but you might want to invest for the long term (buy and hold) or stick to swing trading and nibble off what you can. Keep your expectations realistic.
Table 8.4 shows breakout‐related statistics.
Breakout direction. This just in: Broadening bottoms break out upward more often than downward!
Yearly position, performance. I sorted the breakout price into one of three bins, depending on where it was in the yearly high–low price range. The best performance came when the breakout price was near the yearly low. The worst was when it was near the yearly high. That suggests bottom fishing (buy low, sell high) works better for the broadening bottoms than momentum trading (buy high, sell higher).
Throwbacks and pullbacks. The next several rows in the table dissect throwbacks and pullbacks. If you don't know what a throwback or pullback is, ask your mom (or check the Glossary).
Throwbacks happen 69% of the time. Price breaks out upward from a broadening bottom, rises for a week by an average of 6%, and then returns to (or comes close to) the breakout price by day 12.
If you're an experienced swing trader, you might want to short a downward breakout and close the position in a week or when price drops 7%. However, the median decline is just 5%, so it might not be worth it.
Table 8.4 Breakout and Post‐Breakout Statistics
| Description | Up Breakout | Down Breakout |
|---|---|---|
| Breakout direction | 60% up | 40% down |
| Performance of breakouts occurring near the 12‐month low (L), middle (M), or high (H) | L 48%, M 45%, H 43% | L –16%, M –13%, H –11% |
| Throwbacks/pullbacks occurrence | 69% | 62% |
| Average time to throwback/pullback peaks | 6% in 7 days | –7% in 7 days |
| Average time to throwback/pullback ends | 12 days | 12 days |
| Average rise/decline for patterns with throwbacks/pullbacks | 43% | –14% |
| Average rise/decline for patterns without throwbacks/pullbacks | 48% | –16% |
| Percentage price resumes trend | 75% | 49% |
| Performance with breakout day gap | 48% | –14% |
| Performance without breakout day gap | 44% | –15% |
| Average gap size | $0.63 | $0.74 |
Anyway, I compared the performance of broadening bottoms with and without throwbacks or pullbacks and found that the pattern performs better without a throwback or pullback happening. That's not a surprise because I've seen that behavior in other chart patterns, too.
After a throwback or pullback ends, the stock resumes moving upward 75% of the time after an upward breakout and drops 49% of the time after a downward breakout. Be careful when thinking you can short after a pullback completes. Price might continue rising instead (51% do).
Gaps. Do breakout day gaps help performance? Sometimes. Gaps are not a big indicator of future performance. I checked the statistics for various types of chart patterns (double bottoms, head‐and‐shoulders, and so on), and the average performance improvement is one percentage point. Upward breakouts in bull markets saw price climb by two percentage points if they had a gap. That's for all non‐Fibonacci chart patterns. In other words, there's not a big performance difference when you average all the numbers together.
For broadening bottoms, gaps help performance by an average of one to four percentage points, depending on the breakout direction. The gap size is slightly larger after a downward breakout, and that might have something to do with how price drops faster than it rises.
Table 8.5 shows pattern size statistics. This is one of my favorite tables because height is usually the best indicator of future performance.
Height. Tall patterns outperform. What is meant by tall? Compute the height of the pattern from the top of the broadening pattern to the bottom and divide the result by the breakout price. A tall pattern will have a ratio larger than that shown in the table for the associated breakout direction.
Table 8.5 Size Statistics
| Description | Up Breakout | Down Breakout |
|---|---|---|
| Tall pattern performance | 48% | –17% |
| Short pattern performance | 42% | –12% |
| Median height as a percentage of breakout price | 12.0% | 13.2% |
| Narrow pattern performance | 46% | –14% |
| Wide pattern performance | 44% | –16% |
| Median width | 41 days | 39 days |
| Short and narrow performance | 39% | –13% |
| Short and wide performance | 49% | –11% |
| Tall and wide performance | 42% | –17% |
| Tall and narrow performance | 63% | –17% |
Warning: Just because a pattern is tall does not mean it'll outperform. The numbers are an average of hundreds of perfect trades. Plus, you might flub the trade anyway.
Width. Width is not as strong an indicator for future performance as is height. The table shows an example of this, too. After an upward breakout, narrow patterns outperform but wide ones do better after downward breakouts.
To determine width, measure the calendar days from the start of the pattern to the end and compare the result to the median width in the table. A value higher than the median means it's wide.
Height and width combinations. According to the table for upward breakouts, if tall patterns outperform and narrow patterns outperform, you'd expect the combination of tall and narrow to be the best performer. Indeed, that's what happens, but that's not always the case.
For downward breakouts, tall and wide patterns should outperform, but the results show that anything tall is best. Avoid short patterns.
Table 8.6 shows volume‐related statistics. If the height table is my favorite, then volume is at the other end. I think traders put too much emphasis on volume. Remember that for every share sold, one is bought. If an institution buys a gazillion shares, they are probably buying it from another institution that is selling a gazillion shares.
Volume trend. Volume trends higher most often in the chart pattern.
Rising/Falling volume. I sorted performance by the trend direction and found that patterns with a rising volume trend outperform.
Breakout day volume. Heavy breakout day volume only sees improved performance for broadening bottoms after upward breakouts.
Table 8.7 shows how often price reaches a stop location. You can use this information to help locate a stop‐loss order, should you decide to use one. I'm not being cute here. Investors (buy‐and‐holders) should not use a stop in my opinion. Shorter‐term traders would be wise to use a well‐placed stop or a mental stop (if you have the willpower to obey that).
I sliced the chart pattern in half and measured how often price during a trade returned to touch the top, middle, or bottom of the chart pattern. See the Glossary (“Stops”) for details on how I did this in case you're interested. (Hint: I didn't use power tools.)
Table 8.6 Volume Statistics
| Description | Up Breakout | Down Breakout |
|---|---|---|
| Volume trend | 65% up | 67% up |
| Rising volume trend performance | 46% | –15% |
| Falling volume trend performance | 43% | –14% |
| Heavy breakout volume performance | 46% | –15% |
| Light breakout volume performance | 43% | –15% |
Table 8.7 How Often Stops Hit
| Description | Up Breakout | Down Breakout |
|---|---|---|
| Pattern top | 77% | 1% |
| Middle | 23% | 15% |
| Pattern bottom | 3% | 73% |
Table 8.8 Performance and Failures Over Time for Bull Markets
| Description | Up Breakout | Down Breakout |
|---|---|---|
| 1990s | 40% | –17% |
| 2000s | 46% | –14% |
| 2010s | 46% | –14% |
| Performance (above), Failures (below) | ||
| 1990s | 15% | 13% |
| 2000s | 17% | 30% |
| 2010s | 17% | 31% |
If you place a stop at the top of the pattern, price will take out the stop 77% of the time after an upward breakout. Downward breakouts will only reach the top of the pattern 1% of the time. That makes sense, doesn't it?
Table 8.8 shows the performance over three decades. How has the pattern performed over time? Let's find out.
Performance over time. Upward breakouts in the 1990s suffered but downward breakouts did better. I can't explain why. The 2000s contained not one but two bear markets, but I excluded those results from the table.
Failures over time. The 1990s were the worst performers, but they have the best (lowest) failure rates. Again, this puzzles me. Because the failure rate is a function of performance, then I'd expected patterns that showed big moves to have lower failure rates. They don't.
Table 8.9 shows busted pattern performance. At one time, I thought that busted patterns were the way to make a bundle trading chart patterns. They can be, but it's not as easy as you might expect.
Busted patterns count. I counted the number of busted patterns and found that 42% of broadening bottoms with downward breakouts will bust. Ouch. It's less painful for upward breakouts, with a quarter of them busting.
Busted occurrence. If we sort the busted patterns into three bins, single busts, double busts, and three or most busts (triple+), we see the results in the table. Notice that most of the busts are single ones.
Busted and non‐busted performance. The last three lines in the table show the performance of busted and non‐busted ones. Notice that single busted patterns perform better than all busted patterns (single, double, triple+) and also beat the non‐busted pattern performance.
Perhaps now you understand why trading busted patterns might be the way to riches. Probably not, but we can dream. Try looking for a single busted downward breakout from a broadening bottom. Then try to carve out a portion of the 46% average rise.
