In the test I ran this fall, the system looked to give decent annual returns. However, we can improve the system even more! One way to do so is to set stop losses. Many people will use the range of the stock’s price to determine the stop. However, if you are using the same system to trade many stocks on a percentage risk basis, then you can apply some methods across them all. This analysis below focuses on the adverse or average negative excursion the systems’ trades take in order to be successful. We already know that our adverse excursions are in line; we wouldn’t be working on honing an already profitable system here if the excursions were not healthy.
Using the worst of the two: Average Maximum Adverse Excursion (AVG MAE) or the average loss per losing trade, figure out a general area to start backtesting. Determine whether using a system-wide stop percentage basis is an improvement to results using Profit Factor. The idea is your profit factor rises when making more money without losing as much money, and this is always something you want to improve.
Here is the raw data and the data filtered via the “Adverse Excursion Filter”:
The way the filter works is you take all the trades that at any point traded less than the worst of (avg loss or avg MAE), in this case -3.01% (avg loss), and change the PL on these trades to -3.01%. We do this whether they were winners or losers before the filter. What we find out is whether cutting losses by exiting trades saves more loss than it limits winners from making more profit.
It will always cause the winning percentage to drop, so expect that. You can then try different values. For example, I could use the AVG-MFE at -2.5% instead of the -3.01%. The resulting differences are negligible, and I have found my method here finds the sweet spot well. Optimization should be quick and in a “yes it works, no it doesn’t” fashion, don’t get into setting the perfect stop value as this risks a trader disease known as data curve fitting.
RESULTS: We don’t have to look very far to see if we have indeed improved the system. PROFIT FACTOR measures this for us, and the filter gives us +0.26 basis points in the results shown. This translates to worst-case roughly +3.0% annualized. Not a huge boost to the account, but as you will see in upcoming posts, adding a couple more good tactical filters will add up to a good increase in profitability.
It is important to note that the “Value in the Trend” system didn’t benefit from the filter having mixed results. This is because the idea in the system is to stay put in trades. Having a stop take you out of 11 good long winners that took 20 days to develop isn’t good for that system. Also, any of the metrics that involve time will be slightly off in the second picture as I didn’t change the trades’ lengths. For this analysis, I only care about profit and loss.
Overall, for a swing system this is a filter that I see work to improve backtesting, and walk-forward testing, as well as my live trading. 🙂