First off, here are the final 2009 standings:
Not much has changed since we last checked in. Most of these got a bit lower, as you'd expect with the slightly larger sample size; CAIRO ended up gaining a decent amount of separation from the pack.
Of course, there's all sorts of noise when we're just looking at one season of projections. Below are the systems I have win totals from for each year since 2005, plus CHONE, which I have for '06-'09:
It's essentially a dead heat between CHONE and MGL over the last four years, with MGL holding the five-year crown on the strength of his '05 performance. PECOTA was the clear leader before 2009, but because of both its terrible showing this season and questions about the system's current methodology, I'd have a hard time relying on its 2010 projections too heavily. Considering that MGL doesn't make his win projections public prior to the season, that means we should probably be focusing on CHONE when analyzing the preseason divisional odds next March.
4 comments:
How does Marcel rank from 05-09?
I can see where this is going...A projection system for win projection systems. We've gone meta.
I understand why you use deviation for comparing accuracy but for shits and giggles I compared the Chone projections vs the O/U used in this years VW contest and Chone would have gone 17-13. Nothing brilliant, but if someone was crazy enough to blanket bet they would have shown a profit.
I also thought it was interesting that three times Chone vs the spread had a difference of more than 6 and all three won. How much do you rely on chone and the like with making your futures bets?
What was the mean and 1 standard deviation of all the projection systems. Seems like dumb luck would have one of them 2 SDs away from the mean.
vr, Xei
Post a Comment