Quote:
Originally Posted by CaptainCrunch
If the money ball concept worked in Hockey then Edmonton who bought in a basement dwelling stats nerd would be much better then they are, and guidos like Purcell and Poulliot would be all stars this year.
I still believe that there is a completely random element to the game (ie lw'er going down the right side of the ice, depth of backcheck gaps) that kills a lot of the advanced stats.
I'm not saying that there is no use for them, but I'm saying that too many people use them in arguments that there is no point in arguing.
|
Exactly.
Not only is there a massive random element that has to be considered, but also, many of the more 'useful' stats have very small discrepancies. In other words, people talk about this player's Corsi being 53% and that player only 49%. In order for there to be a real difference between 53% and 49% (for example), you would need a
lot of data. And that's
ignoring the large amount of random noise that's going on.
With all of the players, bounces, randomness and 'luck' that is a part of hockey, it requires a
crazy amount of data in order to be able to have any confidence that there is any real information in the stats. Yet people throw them around like they are gospel.
Not saying they are useless, just that they are over-valued, over-used and miss-interpreted more often than not.