Because I have been testing the veracity of fills on a R3000 system I ran a backtest for this month against the actual fills and got a discrepancy of around 20% on the number of fills….This didn’t feel right to me as I don’t remember on the daily checking it being out so much.

So I went back to my R1000 MOC and checked that. Now I compare a daily backtest to the actual fills every day at the end of trading and there is rarely a discrepancy. However I ran a back test from Aug to Nov this year and there was a noticeable difference.

Aug -4.3 backtest -4.46 actual
Sept 2.9 backtest 1.22 actual
Oct 4.8 backtest 3.85 actual
Nov -1.4 backtest -1.71 actual

This is with me checking results against backtests on a daily basis to compare; I do this by checking the P/L total is comparable, not by checking each trade.

I guess there is the odd occurrence of partial fills and the odd missed fill on the low of the day. Also a backtest will include fills after 15:40 pm which an actual will not, so that is probably the cause.

It made me a little more conservative with my thinking about testing/expectations though