Читать книгу Encyclopedia of Chart Patterns - Thomas N. Bulkowski - Страница 149
Statistics
ОглавлениеTable 11.2 shows general statistics for this chart pattern.
Number found. I uncovered 2,424 broadening tops in data from July 1991 to November 2019 in 873 stocks. Not all stocks covered the entire time, and some of them no longer trade.
Reversal (R), continuation (C) occurrence. By definition, since we are dealing with a top, an upward breakout is a continuation and a downward breakout means a reversal of the prevailing upward trend.
Average rise or decline. I compared the average rise or decline with the average for all chart pattern types and found that the broadening top underperforms in all market conditions (bull/bear) and breakout directions, despite what you see in the table.
Standard & Poor's 500 change. Notice how the general market helped or hindered price. The numbers, when compared to the average rise or decline, suggest trading with the market trend (bull market, upward breakouts and bear market, downward breakouts) for the best performance. This makes intuitive sense.
Days to ultimate high or low. The time it takes price to rise to the ultimate high or sink to the ultimate low varies from 253 days (about 8.5 months) to 37 (5 weeks).
I compared the average speed of price for upward versus downward breakouts in bull markets. The decline should have taken 79 days, but the average is 56 as shown in the table. Thus, price drops faster than it rises. In this case, it's not quite twice as fast (which is what I've seen in other chart pattern types).
Comparing the two bear market columns, we see price rise 25% in 91 days and drop 22% in 37 days. If price dropped as fast as it climbed, it should have taken 80 days to drop 22%. In this case, price drops more than twice as fast in bear markets as it rises.
How many change trend? This is a measure of how many broadening tops see price move more than 20% away from the breakout price. I like to see values above 50%, but only one of the columns qualifies.
Table 11.3 shows failure rates for broadening tops. I found that 18% of broadening tops in bull markets after upward breakouts fail to see price rise more than 5%. That's how you read the table.
Here's another example. Just over half (52%) of patterns in bull markets with downward breakouts fail to see price drop more than 10%. Notice how quickly failure rates increase.
Bear markets with downward breakouts have the lowest failure rate initially, 9%, but the edge over bull markets with up breakouts does not last long. For moves of 15% and higher, bull markets with upward breakouts maintain a lower failure rate.
How is this information useful? Imagine that the measure rule predicts price will climb from 40 to 50, a 10‐point rise. That's a move of 25%. How many broadening tops in bull markets, on average, will see price rise more than 25%? Answer: 46%. To put it another way, 54% of them will fail to see price rise more than 25%.
Table 11.4 shows breakout‐related statistics for broadening tops.
Breakout direction. In bull markets, price breaks out upward most often. In bear markets, the breakout direction is random.
Table 11.3 Cumulative Failure Rates
| Maximum Price Rise or Decline (%) | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| 5 (breakeven) | 219 or 18% | 35 or 18% | 219 or 27% | 19 or 9% |
| 10 | 160 or 31% | 31 or 33% | 196 or 52% | 30 or 24% |
| 15 | 119 or 41% | 24 or 45% | 125 or 67% | 39 or 43% |
| 20 | 81 or 48% | 28 or 59% | 83 or 77% | 25 or 55% |
| 25 | 78 or 54% | 22 or 70% | 60 or 85% | 27 or 68% |
| 30 | 62 or 59% | 10 or 75% | 40 or 90% | 16 or 76% |
| 35 | 62 or 64% | 9 or 80% | 26 or 93% | 13 or 82% |
| 50 | 121 or 74% | 19 or 89% | 45 or 99% | 28 or 96% |
| 75 | 122 or 84% | 10 or 94% | 9 or 100% | 5 or 99% |
| Over 75 | 191 or 100% | 12 or 100% | 1 or 100% | 3 or 100% |
Yearly position, performance. I sorted the breakout price into one of three bins. Upward breakouts show the best performance when the breakout price appears in the middle third of the yearly high–low price range. Downward breakouts favor the lowest third of the range.
Throwbacks and pullbacks. A throwback or pullback occurs about two‐thirds of the time, and it takes 11 or 12 days to complete the trip back to the breakout price (on average).
In all cases, when a throwback or pullback occurs, performance suffers. To avoid a throwback or pullback, look for nearby support or resistance that is strong enough to turn price back. Good luck with that. I've not been able to predict whether a throwback or pullback will happen. I just assume they will and go from there. If they don't appear, I'm as happy as a kid with a bag full of candy on Halloween.
You might try looking for overhead resistance or underlying support within 5% to 11% away from the pattern (those percentages are averages of how far price travels in 6 days, according to the table).
Gaps. Sometimes a breakout day gap helps and sometimes not. Downward breakouts have results that are one percentage point apart, so they may not be statistically significant. Upward breakouts see better performance if a gap is absent. That's counter to normal trading wisdom. Other chart patterns see performance improve if a gap occurs.
If you want to participate in a gap trade, know that I measured performance using the opening price the day after the gap, so you can buy a stock after it shows a gap and maybe you can participate in the better performance.
Table 11.4 Breakout and Post‐Breakout Statistics
| Description | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| Breakout direction | 60% up | 49% up | 40% down | 51% down |
| Performance of breakouts occurring near the 12‐month low (L), middle (M), or high (H) | L 34%, M 46%, H 41% | L 23%, M 26%, H 25% | L –16%, M –14%, H –12% | L –23%, M –22%, H –17% |
| Throwbacks/pullbacks occurrence | 67% | 64% | 67% | 67% |
| Average time to throwback/pullback peaks | 5% in 7 days | 7% in 6 days | –6% in 6 days | –11% in 7 days |
| Average time to throwback/pullback ends | 12 days | 11 days | 11 days | 12 days |
| Average rise/decline for patterns with throwbacks/pullbacks | 38% | 24% | –13% | –19% |
| Average rise/decline for patterns without throwbacks/pullbacks | 48% | 26% | –14% | –26% |
| Percentage price resumes trend | 75% | 69% | 54% | 49% |
| Performance with breakout day gap | 38% | 22% | –14% | –21% |
| Performance without breakout day gap | 42% | 26% | –13% | –22% |
| Average gap size | $0.84 | $0.35 | $0.92 | $0.63 |
Table 11.5 shows statistics related to size.
Height. In all cases, tall patterns see better post‐breakout performance than short ones.
To use this finding, measure the pattern's height from the highest high to the lowest low. Divide the difference by the breakout price (for upward breakouts, it is the highest high in the pattern; for downward breakouts, use the lowest low). Then compare the result with the median shown in the table. A result higher than the median means you have a tall pattern; lower means a shorter pattern. For best results, select tall broadening tops and avoid short ones.
Table 11.5 Size Statistics
| Description | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| Tall pattern performance | 46% | 26% | –16% | –22% |
| Short pattern performance | 36% | 24% | –12% | –21% |
| Median height as a percentage of breakout price | 10.6% | 15.2% | 10.9% | 18.5% |
| Narrow pattern performance | 39% | 26% | –13% | –21% |
| Wide pattern performance | 44% | 24% | –14% | –22% |
| Median width | 45 days | 45 days | 38 days | 41 days |
| Short and narrow performance | 38% | 28% | –11% | –20% |
| Short and wide performance | 35% | 13% | –11% | –23% |
| Tall and wide performance | 47% | 29% | –16% | –22% |
| Tall and narrow performance | 43% | 20% | –15% | –22% |
Width. In three of four contests (columns), wide patterns see better post‐breakout performance than do narrow ones. The one exception happens in bear markets after upward breakouts.
I found it odd that bear market, up breakout percentages for height and width are the same. I checked my spreadsheet, and the information is correct.
To use width as an indicator, compute the time from the end of the pattern to the start (in calendar days, not price bars). If the result is more than the median listed in the table, then you have a wide pattern.
Height and width combinations. Table 11.5 shows the performance results after combining the characteristics of height and width. Most of the time, tall and wide patterns show the best performance. In bear markets, after downward breakouts, short and wide patterns outperform. That's probably a statistical fluke, and the percentages are close, anyway.
Table 11.6 shows volume‐related statistics.
Table 11.6 Volume Statistics
| Description | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| Volume trend | 67% up | 73% up | 67% up | 66% up |
| Rising volume trend performance | 42% | 27% | –14% | –23% |
| Falling volume trend performance | 41% | 19% | –13% | –20% |
| Heavy breakout volume performance | 41% | 25% | –14% | –21% |
| Light breakout volume performance | 43% | 24% | –13% | –23% |
Volume trend. Volume trends upward from 66% to 73% of the time on average. However, don't throw away a trade because the pattern has a downward volume trend. Try to recycle.
Rising/Falling volume. Half the time, the volume trend doesn't really matter to performance. In bear markets, we see a wider separation of results. For example, after upward breakouts in bear markets, price averages a climb of 27% compared to just 19% for patterns with falling volume. The sample counts (146 versus 54, respectively) are not as robust as I like to see, so the results may change.
Breakout day volume. The broadening top is a rebel: It doesn't conform to common wisdom about heavy breakout volume (which is, heavy breakout volume helps performance). As the table shows, sometimes it helps, sometimes it hurts, and half the time it can't make up its mind. However, as a general rule, you should see improved performance (probably meager) if breakout day volume is above the prior 30‐day average.
I added Table 11.7 to this edition to highlight where price might stop on the way to the ultimate high or low. For example, I found that 79% of the time, price will return to the top of the pattern after an upward breakout. You'll likely be stopped out a lot if you choose to place a stop‐loss order there.
Place a stop at the bottom of the pattern and a stop‐loss order will be hit less than 5% of the time, on average. However, the size of the loss, after an upward breakout, could be large, so do consider the size of the potential loss before deciding where to place a stop.
Table 11.8 shows how the chart pattern performed over three decades. This table only shows bull market numbers because bear markets only happened in the 2000s.
Table 11.7 How Often Stops Hit
| Description | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| Pattern top | 79% | 79% | 3% | 2% |
| Middle | 24% | 20% | 18% | 8% |
| Pattern bottom | 4% | 3% | 75% | 67% |
Table 11.8 Performance and Failures Over Time for Bull Markets
| Description | Up Breakout | Down Breakout |
|---|---|---|
| 1990s | 40% | –16% |
| 2000s | 49% | –12% |
| 2010s | 36% | –13% |
| Performance (above), Failures (below) | ||
| 1990s | 14% | 17% |
| 2000s | 15% | 29% |
| 2010s | 23% | 33% |
Performance over time. Upward breakouts saw the best performance and downward breakouts saw the worst performance in the 2000s. That makes sense. If the market is rising after an upward breakout, those patterns with downward breakouts will struggle to see price drop. It's like trying to swim against the current.
Failures over time. Failures have increased over the three decades for both upward and downward breakout directions. That's not a good omen for future performance.
Table 11.9 shows busted pattern performance. See the Glossary (“Busted pattern”) for details on what a busted pattern looks like and how to spot one. Don't forget your binoculars.
Busted patterns count. Almost half (47%) of broadening tops will bust in bull markets after downward breakouts. Notice that the fewest busted patterns happen after downward breakouts in bear markets. There you have the market current (downward) carrying along price as it drops after a downward breakout. The sample count, at 30, is tiny compared to the some of the others.
Busted occurrence. I counted the types of busts (single, double, or more than two) and found that single busts happen most often. In second place for downward breakouts is triple (or more) busts.
Busted and non‐busted performance. I compared the performance of all busts (one, two, and triple+) with single busted and patterns that don't bust (non‐busted). Busted patterns outperform when the busted breakout direction matches the market trend (upward breakout in bull markets and downward breakout in bear markets).
Table 11.9 Busted Patterns
| Description | Bull Market, Up Breakout | Bear Market, Up Breakout | Bull Market, Down Breakout | Bear Market, Down Breakout |
|---|---|---|---|---|
| Busted patterns count | 351 or 29% | 57 or 29% | 379 or 47% | 30 or 15% |
| Single bust count | 176 or 50% | 42 or 74% | 252 or 66% | 21 or 70% |
| Double bust count | 115 or 33% | 11 or 19% | 12 or 3% | 4 or 13% |
| Triple+ bust count | 60 or 17% | 4 or 7% | 115 or 30% | 5 or 17% |
| Performance for all busted patterns | –14% | –20% | 38% | 38% |
| Single busted performance | –22% | –25% | 55% | 53% |
| Non‐busted performance | –13% | –22% | 42% | 25% |
Recall that after upward breakouts in bull markets, the stock will bust by dropping. So the stock is heading down (and showing better performance, too!), even as the general market is rising. The same applies to busted downward breakouts (price rises) in bear markets (many stocks drop).
Why does this happen? My only guess is that stockholders know a good or bad situation when they see it and trade with enthusiasm, in spite of what's happening in the general market.
