Encyclopedia of Chart Patterns. Thomas N. Bulkowski
Читать онлайн книгу.
the stock really turn when it's supposed to and how high does the stock rise? Let's find out. Table 3.3 shows statistics regarding the behavior of price after the pattern ends.
How often does price reach or exceed D? If you use the pattern to predict turn D, then price will hit the target nearly all of the time. The 100% hit rate in bull markets is a round‐off thing. I found three patterns out of 1,741 where the stock failed to drop down to the calculated D. So it does fail to reach D, but it's very rare. That's the good news (it reaches D).
Table 3.3 Price Move after Pattern End
| Description | Bull Market | Bear Market |
|---|---|---|
| How often does price reach or exceed D? | 100% | 99% |
| How often does price turn at D? | 38% | 33% |
| How often does D appear within a week of calculated time? | 50% | 47% |
| How many rise to point A? | 40% | 22% |
| How many rise to point B? | 83% | 80% |
| How many rise to point C? | 47% | 33% |
How often does price turn at D? Here's where the pattern stumbles. For those stocks dropping to D, only about a third will turn when they are supposed to. Figure 3.3 shows an example of this when price ignores D and continues lower to bottom at F.
How often does D appear within a week of calculated time? Using the first three turns to predict the time when turn D will appear works about half the time. It could be better, but sometimes having a clue when something will happen can improve your trading. Use this idea just like I explain with price in the Identification Guidelines. That is, find the ratio between the time to move from B to C compared to the time to move from A to B. Point D should appear following the formula, (C – B)/Ratio + C, where ratio is the number you just found.
I'll discuss an example of this in the Sample Trade.
How many rise to…? For those patterns seeing price drop to D and turn upward, how far does price rise? The answer to that question is vital for swing traders.
I used the pattern's turns to gauge success. Point B is closest to turn D, and we see price climb that far 83% of the time in bull markets. That's very good.
Use A, which is at the top of the pattern, and we see that price makes it up that far less than half the time (40%).
Hopefully you can use the information in these tables to help improve trading results using Fibonacci‐based patterns.
Sample Trade
Figure 3.4 shows a trade George made. His computer program found the AB=CD pattern on the way down to D. Let's go through his buy decision.
The pattern appears at turns ABCD on the chart. The high at point A is 6.65, the low at B is 5.81, for a height of 84 cents. Subtracting the height from the high at C (6.55) gives a calculated turn D of 5.71. The high–low range at D is 5.94–5.59. It's the first point where the stock drops below the target, and it accurately marks the turn. So that becomes point D.
Let's see if we can predict when the stock will reach D. Point A is 9 November 2009, B is at 8 December, and C turns at 16 December. The ratio of CB/BA is 8/29 or .28. That's the ratio between the two legs.
Figure 3.4 George took a loss buying this AB=CD.
For point D, we use the formula D = (C – B)/Ratio + C or (8)/.28 + (16 December 2009). Turn D is predicted to appear on 14 January 2010. In this example, turn D appeared on 20 January, and that's within a week of the predicted time. I consider that a good prediction.
Volume sloped downward as line G shows. Even though volume slopes upward most of the time in an AB=CD pattern, the direction is nearly random (53%). He made this trade in a bull market (the bear market ended in early 2009), and Table 3.2 says a falling volume trend penalizes performance, but only by about a percentage point (dropping from 38.4% to 37%).
He used the traditional measure rule for chart patterns to estimate how far price would climb if the pattern broke out upward. The height of the pattern is A – D (6.65 – 5.59) or 1.06, so the target would be 1.06 + 6.65 or 7.71, which is the height added to the high at point A. A downward breakout would be D – 1.06 or 4.53 (or 19% lower).
“When the stock made a strong push upward at E,” he told me, “I bought at the open the next day and received a fill at 6.02. And I just ate breakfast, too.” He grinned.
I scratched my head.
“The stock filled…ate breakfast. Get it?”
I exhaled, rolled my eyes, but said nothing. With a calculated rise to 7.71, he was looking at a potential gain of 18%. “That sounds reasonable,” he said.
For grins, he calculated a new target based on half the height (1.06/2 or .53 added to the high at A) or 7.18. If he sold at the closer target, he'd make 9%.
“Let's look at the historical chart.” He punched keys on his computer and pulled up a chart from 2007. “Overhead resistance here,” he pointed and swept his finger across the screen. Price moved horizontally for a year. “Looks like a cloudbank pattern,” before the bear market started and sucked the stock down. If the stock worked as he hoped, it would have to traverse through that resistance to reach the full height target. “Am I asking too much for price to rise that far?”
A check of Table 3.3 didn't set well with him. With just 38% of patterns showing a reversal at D, he decided to place a tight stop a penny below the low at D (5.59 – .01 or 5.58).
“The trade went bad right from the start. The day I bought, the stock closed lower, almost reversing the move up at price bar E. Four days later,” he pointed to F, “the stock hit my stop and cashed me out at 5.58 for a loss of 7%. That's one day before it bottomed.”
The day after he was stopped out, the stock began a rebound that carried the stock up to the full‐height target, 7.71, shown on the chart as point H.
Looking back at the trade, the AB=CD pattern found point D accurately and on time (within a week of the prediction), but the stock broke out downward (at F), not upward. It busted the downward breakout when it closed above the top of the pattern (above A) and continued higher to reach the full‐height target (H).
4 Bat®, Bearish
RESULTS SNAPSHOT
Appearance: Looks like a big W with turns located by Fibonacci ratios.
Downward Moves