Keyword: "theory".

In practice, well, it's only a small percentage of the whole "thing".

So, let's take some points in turn:

1. "probability", and all those fancy statistics are based on past events. In case one didn't know, past performance is not a guarantee of future performance. Therefore, to "calculate" future performance based on past statistics, and form a conclusion based on that is the wrong thing to do.

2. Fund performance has factors that statistics cannot take into account. a) Fund managers (as in manager persons, not fund house) can change, and the new fund managers can have different view-points and can do something very different. That can stuff you your nicely calculated standard-deviations and Sharpe ratios.

Statistics is nice and useful, and is one of the many, many, things one needs to consider.