One of my favorite quotes:
“If you torture the data long enough, it will confess to anything.” – Ronald H. Coase
Coase didn’t realize it when he said it, but it perfectly applies to systematic trading.
And it should keep you awake at night a little.
It’s hard to tell if you’ve tortured your backtest too much or just right.
But here’s one thing that is easy to do and I found out by accident helps a lot.
Among all the data points (a.k.a. columns, indicators, or features) in a backtest, add a column with random values.
Then continue with your analysis as you normally would.
If you find your random column appearing in your analysis as “important”
Then you know you’re trying too hard and it’s time to stop the torture.
This is something I include in every strategy assessment I do.
Here are the results of a recent strategy assessment (see image below).
This trader removed the trades in the red line and is left with the trades in the green line.
The average profit per trade is up over 50% while steering clear of the torture threshold.
If you’re interested in learning more about making your strategy better, hit reply – happy to chat.
