RealClearSports
Advertisement

MLB Insights


July 22, 2009 2:50 AM

Debunking A Small Portion Of The UZR Myth

After the Tampa Bay Rays emerged as one of the top teams in baseball last year due to their improved defense, stats geeks and baseball fans alike fell in love with the Ultimate Zone Rating statistic.  While the stat is the most comprehensive one we have to quantify defence, and there has been proven correlations between increased UZR and wins, many writers are taking it way too far. 

In an article on Sports Illustrated today written by Ben Reiter, he talks about how an improved defense has helped the Los Angeles Dodgers in one-run ball games.
Dodgers players and coaches believe that their newly stingy defense has not only allowed them to prevent runs on the whole, but also to prevent them when it matter most: in tight ball games. "When it comes down to close games, that's when you really notice it," Kershaw said. "The sure-handed teams seem to win those close games." L.A. is a remarkable 19-9 in games decided by one run.
Yes, LA improved their UZR/150 from -6.0 to -.6 from 2008 to 2009, and seen their record in one run games improve from 19-24 to 19-9.  However, this is only one team and there are a variety of other factors that could explain the improved record.  The Dodgers have also lowered their Fielding Independant Pitching from 3.86 to a MLB best 3.67 and their bullpen ERA from 3.34 to a MLB best 3.21.  These are other stats that can have huge implications on a team's record in one run games. 

My other problem is that 71 games is a fairly small sample size.  I want to take a look at how UZR/150 correlated to win loss record during the 2008 season for all teams.  This is 681 games worth of data.  For starters, I ran the numbers in Excel and there was no correlation between winning percentage and UZR/150.

We can also take a look at these numbers with the naked eye.  I split the league into three tiers, the top ten UZR/150 teams, the middle ten and bottom ten.  What a surprise.  The top tier had a winning percentage of .499 in one run games, the middle tier .489 and the bottom tier the best percentage with .513.  Good defence does not necessarily win close ball games.  I am sure if I ran the data in different years I would have different results.  A team's record in one run games cannot be predicted with UZR/150.

So please baseball writers, stop asking baseball players and managers who are not statistic oriented to comment on something that they do not understand.  I am sure Kershaw thinks that his team's new and improved defence is winning them ball games, and it makes for a great story, but the numbers do not back it up.  It also looks silly when your entire article is based on how UZR improvement correlates to more wins.

If there is one downside to the statistical revolution in baseball it is that everyone thinks they are an expert, and I am not excluding myself on this one, which leads to some pretty dumb things being said and written.

A Member Of