Chances for the deviation could be explained by:
1) Poor estimation of system expectancy
2) Back luck for system B
Assuming your estimation of expectancy is correct, I'll go with B, since the equity should revert to mean. Can't theorize without more data.