Читать книгу Encyclopedia of Chart Patterns - Thomas N. Bulkowski - Страница 91
Statistics
ОглавлениеTable 7.2 shows general statistics for the big W pattern.
Number found. Most of the patterns I found manually over the years but automated finding them to highlight big Ws I might have missed (like the one shown in Figure 7.3). I checked each one to be sure it was a proper pattern. I uncovered almost 2,700 patterns in 765 stocks using data from July 1991 to September 2019. Not all stocks covered the entire period, and some no longer trade.
Table 7.2 General Statistics
| Description | Bull Market | Bear Market |
|---|---|---|
| Number found | 2,172 | 514 |
| Reversal (R), continuation (C) occurrence | 100% R, 0% C | 100% R, 0% C |
| Average rise | 46% | 30% |
| Standard & Poor's 500 change | 13% | 1% |
| Days to ultimate high | 213 | 83 |
| How many change trend? | 62% | 47% |
Reversal (R), continuation (C) occurrence. The big W is a reversal pattern. Price enters the pattern from the top (going down) and exits out the top (going up), reversing the inbound downtrend by definition.
Average rise. The average rise is decent in bull markets, with gains averaging 46%. That's if you trade it perfectly, over 2,000 times. Your trades may do better or worse than what's shown in the table. As you would expect, bear markets with upward breakouts see price climb less far than in bull markets.
Standard & Poor's 500 change. I measured performance of the S&P 500 index from the date of the big Ws breakout to the ultimate high. The table shows the result, which is inferior to the big W's performance. I'm not sure if the comparison is fair. It might be like comparing the times of two marathon runners, one who is top ranked and the other who is an amateur. The top‐ranked runner will have a better time than one who walks the 26 miles.
Days to ultimate high. To gain 46%, it takes time, about 7 months in bull markets. Bear markets see gains taking less time (about 3 months), but that's because the average rise is less, 30%. However, a check of the numbers says the velocity in bear markets is higher than in bull markets, even when a rising price trend is heading against the current in bear markets. The bear market velocity is almost twice as fast as in bull markets. It might be because of the lack of a posted speed limit.
How many change trend? I counted how many big Ws saw price rise more than 20% after the breakout, and I must say, the bull market posts a terrific number (ranking sixth and eighth for bull and bear markets, respectively).
Big Ms in bear markets show good performance, too, but not as good as in bull markets. That's expected, of course (a rising tide lifts all boats in bull markets, but bear markets depress performance after upward breakouts).
Table 7.3 shows cumulative failure rates for big Ws. The breakeven failure rate is 9%. That means 9% of the big Ws fail to see price rise more than 5% after the breakout.
Look at how the failure rates climb. In bull markets, 20% fail to see price rise more than 10%.
Table 7.3 Cumulative Failure Rates
| Maximum Price Rise (%) | Bull Market | Bear Market |
|---|---|---|
| 5 (breakeven) | 201 or 9% | 48 or 9% |
| 10 | 229 or 20% | 85 or 26% |
| 15 | 206 or 29% | 82 or 42% |
| 20 | 185 or 38% | 57 or 53% |
| 25 | 177 or 46% | 45 or 62% |
| 30 | 141 or 52% | 36 or 69% |
| 35 | 144 or 59% | 32 or 75% |
| 50 | 296 or 73% | 45 or 84% |
| 75 | 248 or 84% | 44 or 92% |
| Over 75 | 346 or 100% | 40 or 100% |
Table 7.4 Breakout and Post‐Breakout Statistics
| Description | Bull Market | Bear Market |
|---|---|---|
| Breakout direction | 100% up | 100% up |
| Performance of breakouts occurring near the 12‐month low (L), middle (M), or high (H) | L48%, M46%, H42% | L31%, M29%, H28% |
| Throwbacks occurrence | 64% | 66% |
| Average time to throwback peaks | 8% in 6 days | 11% in 6 days |
| Average time to throwback ends | 12 days | 12 days |
| Average rise for patterns with throwbacks | 45% | 27% |
| Average rise for patterns without throwbacks | 48% | 34% |
| Percentage price resumes trend | 76% | 60% |
| Performance with breakout day gap | 47% | 39% |
| Performance without breakout day gap | 46% | 28% |
| Average gap size | $0.38 | $0.47 |
Want to make 50% on your trades? You'll lose 73% of the time in bull markets (that's one winner and three losers, and that's if you trade it perfectly). In bear markets, the results are even worse with 84% failing to climb more than 50%.
Set realistic goals and realize that you'll have to make more on winning trades to compensate for losing ones. Trading perfectly also helps… (that's a joke, but focusing on improving technique often leads to making more profit).
Table 7.4 shows breakout and post‐breakout statistics.
Breakout direction. All big Ws break out upward. If they don't, then they are not big Ws.
Yearly position, performance. I sorted patterns according to the breakout price and checked performance of the three ranges.
The best performance for both bull and bear markets comes from big Ws near the yearly low. The worst performance comes from those near the yearly high. It suggests that bottom fishing (buy low, sell high) works better for big Ws than momentum trading (buy high, sell higher).
Throwbacks. A throwback happens after the breakout when price rises but quickly returns to the breakout price (within 30 days). Often price resumes the rise thereafter. In Figure 7.1, for example, the throwback peaks at E before returning to the breakout price. Figure 7.2 shows a throwback after point E.
Throwbacks occur about twice every three trades. Price peaks in 6 days after rising between 8% and 11%, on average, before completing the trip back to the breakout price (or nearly so) in 12 days (that's the roundtrip total).
Notice that price does better if a throwback is absent. We've seen this behavior in other chart patterns, too. How can you tell if a throwback will happen? Look for nearby overhead resistance, such as prior peaks, round numbers (10, 20, 30, and so on), or sideways price movement between the breakout price and about 10% higher. If you see some, then expect a throwback. In my trading, I always expect a throwback will happen, and I party when they don't.
The vast majority of the time, between 60% (bear market) and 76% (bull market), price will resume the upward move after a throwback completes. Few things are more aggravating than placing a buy stop a penny above the breakout price, getting into the trade perfectly, only to see a throwback take price lower. Sometimes, the stock continues down below the bottom of the pattern and you're stopped out. Then you sit on the sidelines as the stock recovers and soars like an eagle. Fortunately, this type of behavior is rare for big Ws (Table 7.9 says it happens 20% or less).
You can always wait for a throwback to complete before making a trade. However, if you wait, you'll miss investment opportunities where a throwback doesn't occur (that is, you'll miss the best performers).
Gaps. Breakout day gaps push price higher, meaning performance is better if a gap appears. However, in bull markets, the extra push probably won't be enough to wake you from a sound sleep. In bear markets, it's more of a jolt.
How can you tell if price will gap higher? I've no idea. However, I measured performance from the opening price the day after the gap to the ultimate high. So if you see a gap, you can buy into the situation and maybe score extra performance points for doing so.
Table 7.5 shows how the sizes of big Ws perform.
Height. Tall big Ws perform better than do short ones. To use this finding, measure the height of the big W from the peak between the two bottoms to the lower of the two bottoms. Divide the result by the breakout price (the price of the peak between the two bottoms). If the result is bigger than that shown in the table, then you have a tall pattern.
Table 7.5 Size Statistics
| Description | Bull Market | Bear Market |
|---|---|---|
| Tall pattern performance | 48% | 31% |
| Short pattern performance | 44% | 29% |
| Median height as a percentage of breakout price | 11.9% | 16.2% |
| Narrow pattern performance | 45% | 30% |
| Wide pattern performance | 47% | 30% |
| Median width | 23 days | 21 days |
| Short and narrow performance | 44% | 30% |
| Short and wide performance | 44% | 26% |
| Tall and wide performance | 49% | 32% |
| Tall and narrow performance | 48% | 29% |
Table 7.6 Volume Statistics
| Description | Bull Market | Bear Market |
|---|---|---|
| Volume trend | 69% down | 74% down |
| Rising volume trend performance | 45% | 25% |
| Falling volume trend performance | 47% | 31% |
| Heavy breakout volume performance | 46% | 33% |
| Light breakout volume performance | 46% | 24% |
Width. Bear markets don't see a performance difference, but in bull markets, wide patterns perform slightly better than narrow ones. The median width between narrow and wide is about 3 weeks (see the table). The width is the time between the two bottoms.
Height and width combinations. Patterns that are tall and wide outperform the other combinations of height and width. You might want to avoid short patterns (either narrow or wide). They underperform in bull markets and don't do that well in bear markets.
Table 7.6 shows volume statistics for the big W pattern.
Volume trend. Volume trends downward most of the time, as measured using linear regression between (and including) the two bottoms of the big W. As I mentioned, don't discard a big W because it has an unusual volume trend. Remember, the trend is your friend.
Rising/Falling volume. Patterns with falling volume, as measured from bottom to bottom in the big W, perform best in both bull and bear markets. The bear market shows the biggest difference between the two values (31% versus 25%). The 25% number comes from 136 samples, so additional samples will likely narrow the gap.
Breakout day volume. Bull markets don't see a performance difference, but bear markets do with heavy breakout volume helping performance. The difference seems unusually wide, though, so additional samples will likely narrow the spread.
Table 7.7 is one of my favorite tables because this kind of trading information can pay for the price of this book.
How often are stops hit? I placed a stop‐loss order at the peak between the two bottoms and price touched this (after the breakout and on the way to the ultimate high) over 70% of the time. Scoot the stop a bit lower (the middle of the big W, measured from the peak to lowest valley) and the stop triggers less often, between 13% and 20% of the time. Place it at the lower of the two bottoms and it'll hardly trigger at all.
The safe place to locate a stop‐loss order is below the bottom of the big W, but the distance between the stop and the buy price may entail a large potential loss. The rumors are true: Positioning a stop is an art you need to master to become a successful trader.
Once you decide where to place the stop, convert the potential loss into a percentage of the current price. If you gasp at the result, then the stop is too far below the current price. Either adjust the stop location, abandon the trade altogether, or hope the trade succeeds.
Table 7.8 shows how big W performance has changed over three decades. Because bear markets only occurred in the 2000s, they were excluded from consideration.
Table 7.7 How Often Stops Hit
| Description | Bull Market | Bear Market |
|---|---|---|
| Pattern top | 73% | 75% |
| Middle | 20% | 13% |
| Pattern bottom | 2% | 1% |
Table 7.8 Performance and Failures Over Time for Bull Markets
| Description | Bull Market |
|---|---|
| 1990s | 42% |
| 2000s | 51% |
| 2010s | 45% |
| Performance (above), Failures (below) | |
| 1990s | 8% |
| 2000s | 8% |
| 2010s | 11% |
Performance over time. The 2000s were the outstanding decade for performance with gains averaging 51%. The worst performance came in the 1990s with the 2010s nestled comfortably between those two.
Failures over time. I would expect to see low failure rates in the 2000s because that was the best performing decade, and yet the table shows it's tied at 8% with the 1990s. The 2010s showed a slight uptick in failures. The failures I'm reporting on, by the way, are 5% failures. They count how many big Ws fail to see price rise more than 5% after the breakout.
Table 7.9 shows statistics related to busted patterns.
Busted patterns count. Big Ws rarely bust, as the table shows (compared to other patterns, which see busts in the 40% range). What does this mean? Up to 20% of the time, on average, price will fail to rise more than 10% after the breakout before reversing and closing below the bottom of the big W. So if you want to make a lot of money, then one in five trades won't exceed 10% profit, on average, and that's if you trade it perfectly.
Busted occurrence. For those patterns that busted, I sorted them into one of three bins: single, double, and three or more busts (triple+). The table shows single busts happening most often followed by double and triple busts. In some other pattern types, we see triple busts coming in second place, so big Ws behave themselves.
Busted and non‐busted performance. Compare the performance for all busted patterns (drops of 15% in bull markets) with the performance of big Ms (not Ws). Big Ms see price drop an average of 17% after a downward breakout, and they act as the proxy for a non‐busted big W.
Busted pattern performance for big Ws is not as good as trading a big M chart pattern. However, if you are lucky enough to trade a single busted big W, then price drops an average of 23% (in bull markets). That's exceptional for a bearish pattern; of course, your mileage may vary.
How can you tell if the big W will single bust (as opposed to double or triple bust)? I'll leave that as an exercise for the reader (translation: I have no idea, but focus on nearby overhead resistance that might turn price down or the timing of an earnings announcement).
Table 7.9 Busted Patterns
| Description | Bull Market | Bear Market |
|---|---|---|
| Busted patterns count | 389 or 18% | 105 or 20% |
| Single bust count | 220 or 57% | 78 or 74% |
| Double bust count | 123 or 32% | 19 or 18% |
| Triple+ bust count | 46 or 12% | 8 or 8% |
| Performance for all busted patterns | –15% | –18% |
| Single busted performance | –23% | –22% |
| Non‐busted performance (big Ms) | –17% | –22% |
