Quote:
Originally Posted by JohnnyB
for now I don't think the advanced stats can be consistently counted on to provide better insight than the pattern recognition abilities of a human brain trained on a tonne of game-watching & game-playing data.
|
Stats as "simple" as NHLe and corsiREL are better at identifying useful players than the vast majority of scouts out there. Just look at signings of ours such as Troy Brouwer, or draft picks such as Hunter Smith. That's the "pattern recognition abilities of a human brain trained on a tonne of game-watching & game-playing data" totally flopping.
Of course I'm not saying not to have scouts or that stats of any sort should be overriding the obvious, but these simple stats often focus on what really matters, and those human brains often have such strong biases towards factors that don't really matter, that don't influence winning. Biases like size, hitting (I'm a Bennett fan, but how many times this postseason has he or Lucic went for a hit on the forecheck and the other team was off rushing to the other end? Too much), soft dumpins (see Burke's two minute montage of Corey Perry making pretty dump-ins that accomplished nothing in terms of shots towards the net, though Perry's been great, I'm just talking about the overfixation on some details over others), etc etc. We're all guilty of it. The simple stats kind of filter a lot of the noise out and focus on whether those details can predict results.