Earlier posts this week addressed the need to keep a strategy simple in order to avoid exposure to curve-fitting bias. While that is true in a general sense, there are also times when adding a level of complication to a strategy can improve its performance without significantly increasing its risk of ruin.
We saw an example of this with Shaun’s Euro Scalping Strategy. When a timing filter is applied that only trades the strategy during slower trading times, the performance metrics shot up, giving us a very interesting system. What would happen if we added a second layer to this filter?
Jeff from System Trader Success took his research on Shaun’s Euro Scalping Strategy a step further this week. He decided to experiment with introducing a volatility filter to the strategy. Once he determined the optimal parameters for the filter, he tested how it would improve the base scalping strategy as well as the time filtered version.
Optimizing the Volatility Filter
Jeff’s goal was to use daily price action to determine how Shaun’s Euro Scalping Strategy performed on days where volatility was high and compare that with how the strategy performed on days where volatility was low.
The biggest lesson here is that in order to optimize the filter, Jeff went back to the original scalping strategy. In order to get the most amount of data and limit curve-fitting, he scrapped the time filter for now.
The lesson here is that if you are going to use multiple filters, it is a good idea to build, テスト, and optimize them separately. Then you can stack them together. Stacking the filters while optimizing them can skew the data in a way that confuses which filters are actually working.
The Volatility Filter
Applying the volatility filter to the original scalping strategy slightly improved the strategy. It increased the profit factor from 1.38 宛先 1.43 while significantly lowering the maximum drawdown. The average profit per trade rose from $24.58 宛先 $25.45.
The win rate was kept pretty much the same, and the overall net profit was reduce because 128 trades were filtered out. あなたが見ることができます。, there was definitely improvement, but not nearly as much as with the time filter.
While the results of applying the volatility filter to the original strategy were not impressive, the original plan was to apply the filter to the time filtered strategy. Putting the two filters together produces the best version of the scalping strategy yet.
The combined filters improve the profit factor to 3.44 and the average profit per trade to $56.74. The maximum drawdown is reduced to less than half of the maximum drawdown of the original system.
The new combined filter strategy reduces the total trades to 164, where the time filtered strategy had 233 and the original strategy had 456. Despite how few trades the combined filtered strategy makes, it returns almost as much net profit as the original strategy.
What we have done with these combined filters is find a way to isolate the very best of the best trades from the original strategy.