Hi all!!
A question, maybe little dumb, but I have to ask.
I have a system, the system works well in all timeframes. There is only one sell rule, Rank<95. The system picks 10 stocks, monthly rebalance.
The screener, after the buy rules, returns more than 100 stocks above 95 in the rank.
And here my question:
Can I choose the 10 stocks in that group? Or I must ALWAYS take the first 10 positions in the rank?
If I do that (choose the stocks by myself in the group of stocks higher than 95 in the rank) instead of taking the first 10 positions,
Am I altering the result of the backtest? Am I corrupting the system?
Thanks!!!
Corrupting the system? Absolutely not. You are simply changing the nature of the information the backtest is giving you and the conclusions you can draw.
If you stick with the top 10 stocks no matter what, then you are testing a fully-automated portfolio.
If you are picking your favorite 10 stocks from a group of 100 or whatever that the model is showing you, then you are testing the potential for the model to create a sort-of designer mini-universe (a lot smaller and more manageable than the big one) from which you can make your own active choices.
The latter is a perfectly valid way to proceed, and in fact, the way many in the real world do proceed.
Backtesting is about the past, not the future, so it can never be precise no matter what one does. And nowadays, given that the world, including the investment world, is changing under our feet, there is a lot to be said for the second, less rigid, approach.
In any case, you would not be corrupting anything. It’s simply a matter of which choice you want to make.
Thanks for your answer Marc, it help me a lot.