You are right, of course! If the number of trades considered in your sample is high and in line with MAs of the system itself, due to direct proportionality you have a level of granularity that allows you to have a more concrete evaluation, IMO 150 per year spread over your Time Frame is sufficient for the purpose. My opinion, as you have correctly understood, it was to keep track of disconnected components, just for the scientific purpose, also out of your actual algo-composition, since it is already complicated to distinguish a negative year for a type of strategy (regardless of the type of entry) compared to the exhaustion of performance of a component taken individually, then this year is perhaps to be considered a separate year for trading itself? Dunno, it is an eternal struggle. I'm sure you've done all your assessments/work of the case for which I don't allow myself any tipe of judgment. Good luck for all!