THE BOOK--Playing The Percentages In Baseball

Tuesday, September 01, 2009

WOWY: does Ichiro hate the stringers?

As you know, I love the WOWY concept for fielders.  I reason that if Roger Clemens is pitching, then we should expect Derek Jeter and Adam Everett to see the same batted ball distribution.  No longer do I need to know where the balls were actually hit.  All I need to know is who the pitcher (and batter and park) the shortstop faces, have a large enough sample size, and I’m set.

Let’s take the case of Ichiro, who is considered by the Fans as the best fielder, bar none, in the last 7 years, but who does not get the same level of love from the fielding metrics.  When we look at the 32,696 batters he’s faced (through 2008, excluding HR, HBP, SO, BB, bunts), those batters get 250 more outs when Ichiro is on the field than when he is not.  That’s roughly 30 more outs per season.  If we look only at his pitchers, there are 339 more outs recorded with Ichiro.  When we look at the parks he plays in, it’s 278 more outs.  Ideally, we want to run a model that takes all three parameters.  But, it’s not important for this illustration.  Whatever number you want to give Ichiro, it’s going to be somewhere north of 250.

We can also do a WOWY on batted ball types.  We see how often a flyball hit by a LHH with a RHP gets turned into an out by a RF (or CF as the case may be), and compare that to what Ichiro actually did.  Remember though that a batted ball type is what the stringers record as whether they think a ball is a flyball, or linedrive or groundball.  And, there’s a huge difference in terms of out rates if a ball is marked as FB or LD.  If you have a bias among the stringers, we have a problem.  Ichiro’s WOWY on batted ball types is +110 more outs than average.  This is very inconsistent with all his other WOWY evaluations. 

What this tells me is that the batted ball types, when Ichiro is on the field, are not being marked fairly.  It’s possible based on how he’s positioned, and based on how often he gets an out on a line drive, that those balls in air are being classified as flyballs instead.  That is, rather than being major credit with getting an out on a tough line drive, he is instead getting minor credit on a typical fly ball.

When I read Harry’s great article here, it’s more fuel to the fire:

So, do the Mariners hit a lot of line drives, or do the stringers like to tag hits as line drives? Or should we blame their pitchers?

I’ve got 33,000 balls in play that says that Ichiro is being biased against.  Indeed, when MGL ran his UZR against BIS data (bUZR) and STATS data (sUZR), there was a huge difference in runs.  Why would that be?  The only difference in the data is the stringers. 

So, if you have a fielding metric, be it UZR, Dewan, or Pinto’s PMR, that treats the batted ball type as a piece of factual evidence (to the same extent that the identity of the batter, pitcher, park, base/out configurations are fact), then you need to add an uncertainty level to those estimates.  The worse part is that the bias will be severely skewed toward a small group of players, and the problem is, we don’t know who those players are.  Unless of course you do as Harry does and create batted ball type “stringer factors” of some kind.

That is, when Ichiro is at home and when he’s on the road, how many line drives are being hit compared to flyballs?  Is it the same proportion (after accounting for the possible league-wide home bias)?  If not, then we know we have a stringer bias.