Agreed. Why not model order latency in your backtest? Estimate your round trip order latency from market data to order fill, then in the backtest data, look for the tick that would trigger the order to be placed, wait the round trip time, then look for a fill. It woudl avoid false fills on price spikes that have little chance of getting a fill.
Why not run it on a simulator? Since liquidity probably wont be an issue (200-400 shares) that shouldnt be an issue.