Hi Rainer,

You reply that there wasn’t the OOS feature when you created pathfinder. I’m not talking of WFA (what I agree is hard to do with such a seasonal strategy). A simple OOS test does not require any special feature from PRT. The whole story is just to optimize and test on two different datasets.

Testing on the optimized dataset like you do is IMO not saying much about how the algorithm might perform in the future. And I say it again, I don’t doubt that pathfinder is a good system. You seem to have ran it for some time live and this is of course the ultimate OOS-Test. The only thing I say and I stick to it, the curve fitted backtests in this thread don’t say anything about the quality of the algo and I think it is a pity that there are no tests shown that really show that pathfinder is as good as one is inclined to think after seeing this incredible backtests.

I personally would not dare to set an algo live based on this results. Of course if you have ran it for some time on demo and this was profitable then one might dare to try but based on this tests – no. But this is my personal opinion and others might of course have a different opinion.