Baseball Is Here

Getting started back on the baseball model. BP and crockpot released season win projections, I'll see if I can't work up a reasonable approximation of what some other systems would say.

BP: http://www.baseballprospectus.com/fantasy/dc/

crockpot: http://playoffodds.blogspot.com/

Chone: http://www.baseballprojection.com/2009standings.htm

Here's some good banter comparing projection systems between the actual publishers of those systems (Nate Silver, tangotiger, etc). I consider PECOTA the gold standard, but it doesn't have a huge leg-up. Its advantage is nearly entirely on the margins. In essence, systems almost always effectively agree, but when PECOTA and another major projection system disagree heavily on a player, PECOTA is going to be more right ~55-60% of the time. The exception to that is probably Chone, where the advantage is very slight.

http://www.insidethebook.com/ee/index.php/site/comments/forecast_evaluations/

I'm working on evaluating the systems for the past couple years and coming up with custom projections that just rip off the current systems, but adjusted for demonstrated bias. It might be fruitless, but I'm hoping the sum of the parts, properly evaluated, is a slight improvement over just starting with PECOTA projections to start the year.

I feel like a surprised chicken whenever saying PECOTA in my head.