As of 11/20/2024   Indus: 43,408 +139.53 +0.3%     Trans: 17,002 -26.31 -0.2%     Utils: 1,055 +1.25 +0.1%     Nasdaq: 18,966 -21.33 -0.1%     S&P 500: 5,917 +0.13 +0.0%	YTD  +15.2%    +6.9%    +19.7%    +26.3%    +24.1%	Targets    Overview: 11/12/2024     46,000 or 43,000 by 12/01/2024   18,000 or 16,600 by 12/01/2024   1,075 or 1,000 by 12/01/2024   20,000 or 18,400 by 12/01/2024   6,100 or 5,800 by 12/01/2024
Chart Pattern Indicator: on 11/18/24

As of 11/20/2024   Indus: 43,408 +139.53 +0.3%     Trans: 17,002 -26.31 -0.2%     Utils: 1,055 +1.25 +0.1%     Nasdaq: 18,966 -21.33 -0.1%     S&P 500: 5,917 +0.13 +0.0%	YTD  +15.2%    +6.9%    +19.7%    +26.3%    +24.1%
Targets    Overview: 11/12/2024     46,000 or 43,000 by 12/01/2024   18,000 or 16,600 by 12/01/2024   1,075 or 1,000 by 12/01/2024   20,000 or 18,400 by 12/01/2024   6,100 or 5,800 by 12/01/2024
Chart Pattern Indicator: on 11/18/24

Bulkowski on Pattern Pairs Rank

Initial release: 12/29/2021. Added ascending scallops and adjusted the figures as necessary: 12/30/21.

The idea behind pattern pairs is to pick a chart pattern type (like double bottoms) to buy and another to sell (like a head-and-shoulders top). You buy the upward breakout from the double bottom, hold for a few years, and sell when a head-and-shoulders top appears and breaks out downward. Along the way, you give price a chance to rise far enough to overcome other pattern pair trades which are stopped out for a loss. This is a trend-following strategy.

This article summarizes the average performance of pattern pairs, highlighting the top ten.

Out of 4,081 different pattern pairs, I removed 997 because they had fewer than 30 trades. Statistics from those can be unreliable. That's why you'll see some gaps in the rankings. The statistics below apply to the remaining 3,084 pattern pairs. Commissions, fees, dividends, interest earned on balances waiting to be invested were not included in the totals.

You can download (for your personal use only) an Excel spreadsheet of the summary results.

• Download Results Spreadsheet (563k, .xlsx file)
• Overview
• Largest Winning Trades
• Smallest Losing Trades
• Highest Net Profit
• Highest Annualized Net Profit
• Trades with No Stop order: Net Profit
• Highest Win/Loss Ratio
• Highest Expectancy
• See Also

Pattern Pair Rank: Overview

The results shown above apply to the results summary spreadsheet (Download Results Spreadsheet (563k, .xlsx file), same as above link), not as a summary of the individual trades. The distinction is an important one.

In my tests, I researched each pattern pair and computed the average performance for the pair. I reported that average on another spreadsheet, which I call the Results Spreadsheet (the same one you can download at the above link). The above table is an average of the data in the Results Spreadsheet. In effect, it's an average of averages. It's not ideal, but I believe it has value. With that in mind, let's look at the results.

The above table shows the average and median values of the various categories for pattern pairs shown in the Results Spreadsheet. The following definitions apply to all tables.

Win: Average. This is the average percentage gain for all pattern pairs which showed a profit. For example, the average gain was 167%. The median win, which is the middle value from a sorted list, was 142%. Half of the average trades did better than this and half did worse.

Net Profit. When you combine the profits and losses from a pattern pair, the average net profit was a gain of 39%.

Annual Profit. This is the net profit annualized. If you held a pattern pair for three years and made 33% (as a fictitious example), that's 11% each year.

No Stop: Net. This is the net profit for trades which did not use a stop loss order.

Win/Loss Ratio. This is how many trades showed a profit, on average.

Expectancy. Expectancy is a way of gauging winning and losing trades and how much money you might make trading a pattern pair. The formula is: Expectancy = (% wins * Average win) - (% losses * Average loss).

Pattern Pair Rank: Largest Winning Trades

Table 2 shows the pattern pairs with the highest percentage gain. The table is sorted on the "Win: Average" column, and includes winning trades only.

The winning pair is to buy a busted rectangle top and sell an Eve & Eve double top. A busted rectangle is a rectangle with a downward breakout. Price drops no more than 10%, reverses and closes above the top of the pattern. If you have a buy stop in place a penny above the top of the rectangle, you may open the trade near the optimum price.

Sell when an Eve & Eve double top confirms when price drops at least a penny below the bottom of the chart pattern.

On the buy side, a busted rectangle top appears most often (five times!).

On the sell side, a busted Eve & Adam double bottom appears twice in the list. Six of ten rows use busted patterns, which is a strong showing.

The average win ("Win: Average") seems outsized and unbelievable. I researched this and found a trade in Amazon stock which lasted 21 years and made 72,000+%. It's possible that someone held onto the stock that long and bought it at the right moment (the breakout from the chart pattern). A review of other trades for that pattern pair found 10 trades from 8 stocks with gains exceeding 1,000%. Another 10 trades showed gains between 100% (a double) and 1,000%. In other words, large gains are not common but they do happen. I chose to leave them in because I've had such large gains in my trading, often holding stocks for many years, too.

Losses ("Loss: Avg") are reasonable but they are averages. The size of the loss will depend (mostly) on the height of the chart pattern. One of the things I've noticed with rectangles, is that the losses are small compared to other, taller, chart patterns. Recall that a stop is set a penny below the bottom of chart patterns, so the taller they are, the larger the potential loss but the less often a stop triggers.

Net profit is about double the size of annual profit, suggesting a hold time of 2 to 3 years.

The "No Stop: Net" is how much you can make, on average, if you do not use a stop loss order. If you're willing to buy-and-hold for 2 to 3 years, then the rewards can be huge.

One of the big surprises in this table and others is the win/loss ratio ("W/L Ratio") is small. I don't know if this is true or not, but I've heard that trend following strategies commonly have a win/loss ratio in the 40% range.

Expectancy: You can make almost $30 a share per trade (lines 2 and 10) or you can make $3.10 (line 5). Even though the average winning trade is three or four digits (percentage), you have to pick a big winner if only 10% of your trades win (line 5).

Pattern Pair Rank: Smallest Losing Trades

Table 3 shows the results sorted by the "Loss: Avg" column. These are the pattern pair trades which lost the smallest amount. The busted rectangle for buying seems to be a favorite (three times in the table). On the sell side, busted patterns are plentiful with a busted Adam & Adam double bottom appearing twice.

Notice that a rank of 6 appears twice. That's not a mistake. Results are a tie for those two lines. You'll also notice that the loss average column shows 5% often. If you expand the number of decimal places, the rank order is correct for what's shown.

Look at line 1, the pattern pair with the smallest average loss. The win/loss ratio is just 7% and only 2 of 28 trades were profitable and yet they made a combined 306%. That's due to big gains from those two winning trades. I'd much prefer trading line 8 with a 32% win/loss ratio.

Pattern Pair Rank: Highest Net Profit

Table 4 ranks the results by the highest net profit. Busted rectangle tops appear often as a buy pattern. There's no consistent sell pattern except to say that half are busted ones.

The number of trades has climbed dramatically, suggesting the statistics are more reliable. The win/loss ratio is higher than in other tables, too.

Pattern Pair Rank: Highest Annualized Net Profit

Table 5 sorts the results by the highest annualized net profit. This is different from the prior table because it sorts the numbers according to results per year. If you have a trade which makes 25% in half a year, that's a 50% annualized return (not counting dividends). A trade making 25% in two years drops the annual gain to 12.5%. As I mentioned before, the net profit is often between 2 and 3 times higher than the annual rate, suggesting a 2 to 3 year hold time. That makes sense. It's going to take time to make large gains.

Pattern Pair Rank: Trades with No Stop order

Table 6 sorts the table by how well pattern pairs perform when a stop loss order is not used. Busted patterns appear 7 times as buy patterns and 5 times as sell patterns. Hold times appear to be closer to 3 years than 2. The average win seems outsized probably due to large gains from a few stocks. Losses are contained except for line 10, at 25%. Ouch! Samples sizes are good. Expectancy is good. I'm feeling good, too.

Pattern Pair Rank: Highest Win/Loss Ratio

If you want to win more often, Table 7 shows the results sorted by the highest win/loss ratio.

Losses are big, over 20% but annual profit is good. Trades are few except for line 5, with 523 (323 winners and 200 losers). For that line the average win is 154% but you have to tolerate an average loss of 24%.

Pattern Pair Rank: Highest Expectancy

Table 8 shows the results sorted by the expectancy column.

Four ascending scallops and two 3 rising valleys appears in the buy column. Busted patterns are plentiful for selling (8 times in the table).

-- Thomas Bulkowski

See Also

• Pattern pairs trading. Buy a bullish pattern and later sell a bearish one.
• Summary report on the top 10 performing pattern pairs, sorted by various categories.
• Improvements. Describes what can you do to boost performance when trading pattern pairs.
• Ugly patterns: Do they outperform? Yes.
• Use pattern width to improve trading results.
• Best buy months.
• Buy low or buy high? Buy near yearly low for larger gains, less risk.
• Moving average study & stock trends. Buy in a downtrend, sell on the uptrend.