I'm testing an EA in strategy tester using 1M data (90% modelling quality and I used accurate M1 data rather then the simple "download" button available).
The current settings I have for the system work extremely well over 10 years on the 1H chart. However, on other charts (15M and 30M) it goes from extremely profitable to just plain losing. If it was performing well and suddenly crashed or just had smaller returns I could understand, but it just starts losing money from the start.
Should the settings (number of bars indicators are calculated off, mostly) have a massive difference in performance on different timeframes? Because I was under the impression that only TP/SL would differ due to the size of the market swings being smaller. Or do I have an issue of overoptimization on the 1H charts? Or is MT4 backtesting over timeframes smaller then 1H charts simply inaccurate and should I run the EA on the 15M chart in demo?
Of course I'm planning to do this anyway, but if the settings should always require tweaking for 15M or if the EA is overoptimized then I would just be wasting my time.
The current settings I have for the system work extremely well over 10 years on the 1H chart. However, on other charts (15M and 30M) it goes from extremely profitable to just plain losing. If it was performing well and suddenly crashed or just had smaller returns I could understand, but it just starts losing money from the start.
Should the settings (number of bars indicators are calculated off, mostly) have a massive difference in performance on different timeframes? Because I was under the impression that only TP/SL would differ due to the size of the market swings being smaller. Or do I have an issue of overoptimization on the 1H charts? Or is MT4 backtesting over timeframes smaller then 1H charts simply inaccurate and should I run the EA on the 15M chart in demo?
Of course I'm planning to do this anyway, but if the settings should always require tweaking for 15M or if the EA is overoptimized then I would just be wasting my time.