BIG difference in Strategy Tester from Build 198 to 200 (bigger in M1)

 
Same data, same expert running on 1 minute timeframe everytick.

Build 198:
Profit factor=11.15, Total trades=1115, Profit trades=95.96%

Build 200:
Profit factor= 0.81, Total trades= 185, Profit trades=73.51%

This is the case with all my experts, all build 198 big winners are build 200 big loosers. *** Higher timeframe back tests show less difference ***.

This difference is hugh, Why please? Makes me think that I have been wasting soooo much time to develop with this platform & backtester.

Please do get technical as I am very technical.
 
More accurate modeling
Reason: