Hockey Prospectus Top 100 Prospects

[Freudian]

This is a very interesting and important debate. If a certain percentage of hockey games are determined by random chance, then there is a theoretical cap (which is less than 100%) to the accuracy of any predictive model. If there is no random chance, then the theoretical cap is 100%. Is what we perceive as unpredictable by current modeling methods truly the result of this "luck," a force beyond human control, or with more knowledge, with a better understanding of the variables at play, could we improve our predictions?

I agree that there isn't really any randomness involved. Just a limited number of observations (over a year and possibly over a career) that make it may appear that way.

I don't see any reason to treat it differently than what you do with any data set with a limited amount of observations. Attempting to predict or account for it won't improve the statistical analysis.

There are probably some events in hockey that are so unpredictable that as to appear as random/luck. Weird bounces off the boards are one of those things. But I see no reason why that should be accounted for in any way.