I am currently backtesting my system and facing a problem. The problem is whether I should treat my long&short trading strategy as one system or separate them as two. My set-up for both long and short are the same but the filters for each are set differently.
If there's any chance of triggering both long and short entry signals at the same time, then you know you have a problem. But if your filters prevent this from happening then you've already achieved the separation you need.
This is an interesting problem with no simple answers. Let us think of the following possibilities for LongS and ShortS and S (combined system): 1. M of LongS+ShortS > S 2. M of LongS+ShortS < S 3. M of LongS+ShortS = S where M is some metric (return, drawdown, Sharpe, etc.) The problem is to find a set of M that allows you to make a choice between LongS+ShortS and S (using the long and short separately as opposed to the combined system. I would choose CAGR plus max DD plus Sharpe. Ideas are welcome.
Separate systems. For ES longs move in a different way than shorts do, better to optimize each one separately.
Thank you for all the replies here. The reason I asked this question is for the subsequent validation of each system. As I thought more about it, I realized that anytime I use a logic OR in my system, I start to separate it into two or more systems. Now I know why OR has a lower priority than AND in logic.
Instead of Sharpe Ratio, I would take the Sortino and or Omega one. MAR Ratio is a good measure as well. Basically, every measure, which emphasizes the risk of your strategy, is a good one, in my point of view. I would also take into account trading analysis such as Profit Factor. I hope that helps. What kind of dynamic methodology do you use to analyse your data?
System A: MAR ratio 1 (10%/10%) System B MAR ratio 2 (100%/50%) which one is better? Let me tell you IMO all these metrics are pure junk in comparing systems.