The purpose is to make sure that the system is yours, that you understand the rules, and you know why the parameter values are chosen where they are. You’ve satisfied yourself that that’s okay. You like the performance, all of these things. It’s sort of the handoff from me to you and taking ownership and responsibility is, “Yes, this is my system.” Your system now, and you like those parameter values.

The purpose is not necessarily to try and make the system a whole lot better. You may be able to, but it’s mostly about building confidence. Then the question of compounding backtests for optimisation versus unconstrained capital backtest for optimisation. What I do is I optimise the whole system on compounding, so one parameter at a time, and if I’m pretty comfortable and confident in that, as in there are lots of trades, I don’t feel like there’s much risk of curve fitting, that’s all I’ll do. Then, I’ll do an unconstrained capital backtest to check its stability over time. Just one backtest, not an optimisation, because of what we want to see. For example, when we do the unconstrained capital optimisation, we want to see that the equity curve is nice and straight, and it’s not doing something different, where it’s decaying. We’re looking for this sort of behaviour on the unconstrained capital backtest. Sometimes you need to optimise on the unconstrained, but generally, not. So I’m going to suggest that you go through it as I’ve said, optimise on the compounding backtest, do the one backtest on the unconstrained, check its stability, and share the results in one of these sessions. If we need to do anything different, we can discuss it at the time.

Pin It on Pinterest

Share This

Share This

Share this post with your friends!