12/21/24
February 16, 2008
ZEUS: is a computer program that analyses NFL coaches based on the plays they call across the season. Belichek's notorious 4 & 13 call in the superbowl was the right choice, but coaches lose their teams 0.5 to 1.5 games a year on blown calls. Interesting interview with the programmers. via kottke
So the Rams went 3-13 but a). they had the least percentage of games lost due to error, and b.) they made the 2nd best calls in critical situations. Interesting. Scott Linehan has been getting pummelled by the St. Louis media and fan base. The system seems to forget that offense only wins games, defense wins championships. I'd like to see a comparable defensive ranking system as well.
posted by BoKnows at 02:46 AM on February 17, 2008
Well the Lions were only 16th on the list for blowing games due to bad calls. That either means the coaching isn't terrible or they are so bad that the calls have no impact on the game. I'll choose the latter, especially after seeing how the Lions are the worse team in the league at making calls in critical situations with a 68.9% error rate. It is interesting how the Patriots were ranked 9th and 8th respectively while the Rams were first and second.
posted by Ying Yang Mafia at 06:54 AM on February 17, 2008
Interesting to see Baltimore and Billick ranked so highly, considering how painful it was to watch Baltimore games this year. Then again, this computer isn't really accounting for play-calling; it's more about aggressiveness vs. passiveness (punting, field goals). I have long agreed that just about every coach is too conservative in these 4th down situations, but I am still not totally sold on this method. I guess this is partly because I don't entirely believe in the simulations and the GWC numbers. I would like to also see numbers about the expected returns in yards and points in these situations, for both alternatives. (For example, it's hard to wrap my head around a -1.4 GWC if a guy kicks a field goal on 4th and 3 from the 10, but it's a lot easier for me to analyze and agree with it if I see figures like a 2.7 point expected return on a field goal vs a 4 point ER by going for it, plus the 10 yard field position (which is worth roughly a point) advantage gained even if they fail. (And yes, those numbers are made up.)) I guess to me, that kind of result just makes more sense. Especially since I don't fully trust the simulation of the rest of the game, which is obviously important to the GWC number. Still, very interesting, and the more people like this that can publicly call NFL coaches chickens, the better.