http://sites.google.com/site/steamerprojections/2009-ops-forecast-standings

 

Chone Wins! ZiPS looks like a playoff team and Oliver had a strong season to build on going forward. Steamer is taking a cue from the Mets and blaming this one on injuries. We’d fire the manager but he’s also the owner.

 

How should you interpret these standings? The W-L records reveal how well each system did when compared to Tango’s Marcel system (the baseline projection system). How likely is it that Chone was really no better than Marcel this year and simply got lucky? Just as likely as it is that an 93-69 team was really an average team that got lucky. Not at all likely. And, based on other studies that have been done, it seems that Chone has been something like a 90-win team for a few years running.

 

For the statistically inclined, what I did was calculate a Z-score for the hypothesis that each system was just as good as Marcel and matched that up with a Z-score for the hypothesis that a team is a .500 team given it's actual winning percentage over 162 games.

 

You might notice that Steamer took a dive compared to my previous rankings where it had finished ahead of PECOTA. Steamer’s projections did have a higher correlation with actual OPS than Pecota's. The reason for it’s inferior record is that Steamer’s projections were considerably more similar to Marcel’s and therefore, given that it finished below Marcel, we can be more confident in rejecting the hypothesis that it’s just as good as Marcel. Rest assured, however, that Steamer has had a very strong offseason (adding park effects, aging, minor league stats and more thoughtful regression means) and has every intention of making the playoffs in 2010.

Nice work!! Did you do this yourself? I feel pretty comfortable doing relatively complex regressions and would love to see more of your work.

Nice work!! Did you do this yourself? I feel pretty comfortable doing relatively complex regressions and would love to see more of your work.

 

 

See PM. Wish I could say that I did.

Anyone ever wonder why these projection systems consistently get the Marlins record wrong?

 

I found a couple of articles that show that these projection systems explain roughly 40-50% of the variation in record. That means that 50-60% of the variation in record is explained by other factors and luck/random noise. Based on what I've seen, my hunch is that these projection systems are pretty good at projecting really bad teams and really good teams. The Yankees and Red Sox (the good) and the Pirates and Padres (the bad) of the world. We know that the Yankees will be good. Maybe they won't win 100 games, but they'll be in the playoffs. We also know the Pirates/Nationals will be bad. They may not win 60 games, but they'll win under 81 games for sure. As teams' levels of talent approach the mean, it becomes harder to project performance and also very hard to really gauge the level of talent on the team. If you look closely, most of these projections have a standard deviation of 6-7 games won per team. This means that if the Marlins are projected to win 82 games, there is a 95% probability that they will win somewhere between 70 and 94 games. Yeah. That's a big range!

 

In short, if a projection system has your team in the 75-87ish win range, anything can happen. Let's see how the season plays out!