Last Updated on October 25, 2022
In chapter 5 of Larry Connors‘ book High Probability Trading, published in 2009, there is a trading strategy called %b strategy. The book contains many strategies which we test in chronological order. 12 years have passed since the book was published and it could be interesting to see if the strategies are still performing well. Larry Connors tested 20 ETFs from their inception up until the end of 2008, while we test from inception until December 2020.
You can find more strategies from Larry Connors on this page:
What is the %b?
Before we start we need to explain the rather cryptic name “%b”. The “%” is a percentage and the “b” is an abbreviation for Bollinger Bands. Connors describes the indicator like this in his book:
The higher the %b reading, the more likely that the market has moved higher. The lower the %b reading, the more likely the market’s trend has been lower. Traders ideally want o buy a low %b reading and sell higher %b readings.
The BBandBot and BBandTop are the lower and upper Bollinger Bands.
The indicator usually fluctuates between 0 and 1. However, when the ETF is very overbought and above the upper Bollinger Band, the value is above 1. Opposite, when it’s oversold and below the lower Bollinger Band the value is below zero. The strategy is a mean-revertive one.
The rules of the %b strategy:
- The close must be above the 200-day moving average.
- The %b must be below 0.2 for the last three (consecutive) days.
- If 1 and 2 are true, buy on the close.
- Exit when the %b closes above 0.8.
Connors didn’t indicate the length of the lookback period in the Bollinger Band formula, neither did he mention the size of the standard deviation. We use a 5-day lookback period and two standard deviations.
If you like to know the code of Connors’ strategy plus the code for all the other free strategies on this website, click here:
For more on why we use Amibroker, you might want to read this article:
The results of %b:
We were not able to replicate the results of Connors because we lack his parameters. However, the results are very good in QQQ and SPY, but very few fills. This is the equity curve for QQQ:
The table below summarizes all the ETFs:
|Result by Connors||The average gain since||Profit|
|The average gain in %||inception to Nov.2020||factor||Difference:|
The last five ETFs were not part of Connors’ test.
How does the %b perform as a portfolio on all ETFs?
Let’s test the strategy as a portfolio of all the above 25 ETFs. The portfolio is simulated by having max five open positions at the same time (20% of the equity for each position – compounded) from the year 2000 until December 2020. The equity curve looks like this:
The number of trades is 677, the win ratio is 75%, the average gain per trade is 0.76%, and the profit factor is 1.9. The CAGR is 4.84% and the max drawdown is 16%. The low CAGR is mainly due to the low exposure (time in the market) which is only 17%.
If we change the parameters and use a 10-day lookback period we get a much better result: CAGR is 8.2%, but the drawdown increases to 24%.
What if we only trade SPY and QQQ?
The conditions are like this:
- Only positions in either SPY or QQQ
- 5-day lookback period
- 2 standard deviations
- Max one position at a time to avoid overlap
With these conditions the equity curve looks like this:
There are only 56 trades over the 20 year period. Still, the CAGR is 5.1%, the time in the market is almost 6%, and the max drawdown is 11%.
Other Larry Connors’ trading strategies
We have backtested many other strategies from Larry Connors earlier:
- Larry Connors’ Double Seven strategy (Double 7 trading strategy)
- Larry Connors’ multiple days up and multiple days down strategy
- Larry Connors’ R3 strategy
- Larry Connors’ RSI 25 & RSI 75 strategy
- Larry Connors’ 3-day high/low strategy
% B – Conclusion:
The %b strategy is a mean reversion strategy and all such strategies have performed very well over the last 25 years. The strategy has produced some good results, but it remains to be seen how the strategy performs combined with the other strategies of Mr. Connors. Moreover, we have not done any optimization tests and this could yield further improvements on the strategy.