View Single Post
Old 08-27-2013, 09:32 PM   #85
Jay Random
Franchise Player
 
Join Date: Aug 2005
Exp:
Default

This is a reasonable way to go about doing statistical analysis:

Quote:
Originally Posted by V View Post
I would try something like taking a random list of players, maybe 10% of the league over 4 years, and then I would give each of them a subjective gritty rank. Then I would create a table consisting of each of those players gritty score and every single variable that I thought contributed to grittiness. Then I would run a multiple regression model and refine it until it provided a statistically significant result with an acceptable correlation. Then I would use the model to predict the gritty score of all other players in the league. Then I would try to correlate a team's aggregate gritty score to its winning percentage. Then maybe I would be comfortable at least sharing my findings, even if it's total bunk.
This, on the other hand, is just wanking with numbers:

Quote:
Originally Posted by ricardodw View Post
Yelle 2005-6, first year stats were kept, 72 games 101 hits 56 blocked shots 42 take aways and only 18 Give aways.

Personification of a gritty player and totally born out by the stats.
Verbum sap.
Jay Random is offline   Reply With Quote