I found one bug at strategy backtesting. This happens at multi-session instruments, like Wheat (ZW), session CBOT Agricultural ETH. If your primary bars is Minute and secondary bars is Day, the strategy bars flow act this way:
- First goes bars from first Minute session UTC-06 [18:00, 7:15],
- then goes Day bar
- and after that goes bars from second Minute session UTC-06 [9:30, 13:15]
In another words, in second session I simply know, where the price is going to close, so I simply may decide if go long or short.
Let's consider, that my timezone in PC is set to UTC+01, so my sessions are [01:00, 14:15] and [16:30, 20:15]
I attach strategy, which demonstrate this problem. In my case the strategy is not 100% as should be expected because of some incoherence between minute data and day data, but hope that equity speaks for itself.
Comment