Aceboogie
Registered User
- Aug 25, 2012
- 32,649
- 3,896
Not a fan of refining a data model with "the eye test." A data model is objective with vastly higher sample size. Take the model and results for what its worth. Dont muddy the waters tryign to introduce highly, highly biased subjective eye test analysis (with an untrained hockey eye and minuscule sample size for most players).
The best models are the models that have some surprising players much higher in the rankings, because it signals a divergence between conventional wisdom and objective results. That player shouldnt be arbtraily ranked lower because it might make the regular fan scoff at your WAR model. Instead, it should be the fan who has the question their own evaluation of that player (are they watching enough of them, do they play a style of game a fan has a bias against)
The best models are the models that have some surprising players much higher in the rankings, because it signals a divergence between conventional wisdom and objective results. That player shouldnt be arbtraily ranked lower because it might make the regular fan scoff at your WAR model. Instead, it should be the fan who has the question their own evaluation of that player (are they watching enough of them, do they play a style of game a fan has a bias against)