Your fill assumption is doing more work than your entry logic

Two backtests of the same idea can differ by more than most entry tweaks, purely on what price you assumed you got. If you assume the mid, you're assuming someone met you halfway on every trade. If you assume you crossed the spread every time, you get a pessimistic number that's usually closer to reality for anything fast. The honest approach is to test both and see whether the strategy survives the pessimistic version. If it only works at the mid, it doesn't work, you've found an idea that requires a favor from the market on every single entry.