Look-ahead bias with EMA(21, 0 ,#Vol)

Something is not right.

The factors “sma(10, 0 ,#Vol)” and “AvgVol(10)” gives me the same volume numbers in the screener. But if I compare the two factors in a rank test I get different results.