Why would test results differ if inputs are the same?

 

Re-starting the same EA test with the same symbol, same date range (historical), same timeframe, same parameters produces different results. I've eliminated things which could change, like MarketInfo values and everything else that is left should be static.


But different test runs produces subtle differences in Price, SL, TP of a few pips which cause the total profit to be significantly different. What have I missed?