you move the stop‐loss order around, you get different sell prices. You're not testing the double bottom. You're testing how well the stop‐loss order works at different values. So we haven't answered the basic question, “how do you test chart patterns?”
Given the same entry signal (17.06), let's say you traded this stock perfectly. Where would you sell? When price slides below F, you'll be taking a loss if you have to sell, so trying to shoot for 19 on the upper right of the chart isn't optimum. The trade goes negative for a while before it shows a profit.
Ultimate High
How about selling at E, when price peaks? That would give you the most profit before the stock dropped into loss territory. That peak is what I call the ultimate high. If you traded this stock perfectly, you'd buy at 17.06, using the opening price the day after an upward breakout, and you'd sell at the high price at E.
I'm not testing how well a stop‐loss order works because I'm not using a stop or MACD or a moving average crossover system to find the exit. I'm selling at the highest possible price before things go wrong. I'm selling at the ultimate high. A perfect double bottom trade.
How do I automate this? In my software, I use two rules to find the ultimate high (on a historical price chart, not real time).
1 Find the highest high before price drops 20%, measured from the high to the close.
2 If price closes below the bottom of the chart pattern, then the search for the ultimate high stops, and we use the highest high found after entering the trade.
In this example, the stock drops to F and closes at 15.70. The low at B is 15.71, so price has closed below the bottom of the pattern at F. We use rule 2 to find the ultimate high: The highest high between the entry price (17.06) and F (the bottom of the pattern) is E. E is the ultimate high.
Let's pretend the double bottom is much lower so I can tell you how rule 1 works. Let's also assume that the close at J is at 13.
Rule 1 says to look for the highest high before price drops 20%, measured from the high to the close. The high between the buy price and J is E, at 17.80. So we look for price to close 20% below this, or 14.24. When price closes at or below 14.24 (again, assuming this happens at J, at 13), we've found the ultimate high, which is E, at 17.80. So we bought at 17.06 and sold at 17.80 and made a profit of 74 cents a share.
We didn't use a stop‐loss order. We didn't use MACD or the moving average crossover. We executed a perfect trade by buying exactly when we should have and selling at the ultimate high. We tested how well the double bottom worked if you traded it perfectly.
And that's how I measured how bullish chart patterns work.
Years ago, someone asked me if this was the same as placing a trailing stop‐loss order 20% below the high price. It's not. If you did that, you'd be stopped out at J (the fictitious one at 13) and not E.
Incidentally, the 20% value in rule 1 comes from the idea of bull and bear markets. A decline of 20% from a high in the market averages means it slid into a bear market. A rise of 20% off a market low means it entered bull market territory. I applied that idea to individual stocks when searching for the ultimate high and low.
Ultimate Low
Finding the ultimate low for bearish chart patterns works similarly. Let's take the example of the head‐and‐shoulders top at GEH. We see the left shoulder at G, the head at E, and the right shoulder at H. Price confirms the pattern when price closes below the neckline. Because the neckline slopes downward in this example (not shown), I recommend entering the trade when price closes below the right armpit, which I show as the horizontal line at I. You'd short at the open the next day.
If you can program your computer to find head‐and‐shoulders tops, that's how you enter a trade. Nothing magical here.
What about the exit signal? Again, we could use a stop‐loss order, but we already know that's not going to work from our previous example. We'd be testing various locations of the stop and not the chart pattern. So let's find the ultimate low.
Two rules:
1 Find the lowest low before price rises 20%, measured low to close.
2 If price closes above the top of the chart pattern, then the search for the ultimate low ends, and we use the lowest low found after entering the trade.
In this case, price breaks out downward from the head‐and‐shoulders top, drops to J, and then rises. The low at J is 14.60, so a 20% rise is 17.52. A close at 17.52 or higher will end the search for the ultimate low.
When price closes at K (at or above 17.52), the search for the ultimate low is over. We've found the ultimate low, which is J.
If K was a bit higher, then rule 2 would come into play. That's because price would close above the top of the head‐and‐shoulders, stopping the search for the ultimate low.
In this example, we are trading this head‐and‐shoulders perfectly. We are entering the trade at the opening price the day after a downward breakout, and we are closing it out at the ultimate low, the lowest low before price rises. A perfect trade. We are testing how well this chart pattern works. We're not testing stop‐loss order placement or MACD or a moving average crossover scheme.
The performance of most of the chart patterns in this book follows these two ideas: the ultimate high and the ultimate low. For bullish patterns, I look for the ultimate high. For bearish patterns, I look for the ultimate low.
Chart Pattern Failures
Once you understand what a perfect trade means, you can look for failures. What does it mean when a chart pattern fails? To answer that, I had to invent a new concept, which I call the breakeven failure rate or the 5% failure rate. Those two phrases are synonyms. All I did was count how many chart patterns failed to see price rise or decline more than 5%.
In a bull market, for example, I found that 18.8% of head‐and‐shoulders top patterns failed to see price drop more than 5%. If your cost of trading is 5%, then you'll know that nearly 20% of head‐and‐shoulders you trade, if you trade them perfectly and often enough, will fail to cover your costs.
Realistic?
So there you have it. Once you understand that I'm measuring performance from the breakout price to the ultimate high or low, then you'll understand most of the statistics in this book. You'll understand that the performance numbers are based on perfect trades, made hundreds or even thousands of times to arrive at the average rise or decline I show in Table x.2 (where x is the chapter number).
Are the results realistic? Not really. You likely won't be able to duplicate them in real life. You might do better or worse, depending on your skill and luck.
The double bottom example entered the trade the day after the breakout. If you placed a buy stop a penny above the top of the double bottom, you'd be entering at a better price (most of the time, based on my tests) than using the opening price the day after the breakout.
So you could do better than the numbers shown in this book. All you have to do is find the price of the ultimate high, weather the 20% drop, wait for a recovery, and sell at a price above the ultimate
