January 26, 2004

I hate the phrase "team defense." There’s no such thing. A few plays such as the double play and relay throws are team-dependant, but other than that, defense is just like hitting – one ball in play, one man making a play on it. It is the next step in baseball’s majestic mano a mano confrontations – pitcher-batter, batter-fielder. DER is a nice baseline, but it’s very rough as a descriptive statistic and very few conclusions should be drawn from it.

People hate the Gold Glove Awards. Dan Szymborski is fond of saying that it is easier to win the GG twice than it is to win it once and there is a good bit of truth to that. GGs usually begin rewarding a player a year or two after he’s has deserved one and stop giving it to him a year or two after he’s deserved it.

I agree with some of that sentiment, but once the GGs find a good player, he gets rewarded. I would also submit that the accuracy of our defensive measurements should allow for a guy who has been historically strong to win over a player whose season appears to be better than said glove star.

The team that leads in DER must have good fielders, right? The team that has average DER must have average fielders, right? The team that trails in DER must have poor fielders, right?

No, no, and no. The team that leads in DER may have allowed a ton of fly balls or just lucky ball-in-play distribution. The team that trails in DER may have given up a ton of line drives. The team around average may have had just the wrong mix.

The St. Louis Cardinals stumbled home with a slightly above league average DER. They also won four GGs. Okay, one was catcher, and that doesn’t go much into a team’s DER. What’s more, there is very little dispute about Scott Rolen’s defense, Edgar Renteria’s defense, or Jim Edmonds’ defense. The shortstop, the second baseman, and the center fielder have significantly more defensive chances than any of the other positions. The third baseman has a reasonable number (in the 300s as opposed to the 500s)

So what does that mean? How does a pedestrian defensive "team" warrant half of the non-pitcher GGs? Answer: DER ignores chance distribution.

Dave Studemund’s data helps me to look at the individual pitchers:

DER is the defensive efficiency rating for that pitcher, in the simplest form: IP-K divided by IP-K+H-HR.

GB/FBip is the ratio of ground balls to fly balls in play (home runs subtracted from fly balls).

Notice in the table that there are several things at work: high ERA, high FB rate, good DER; low ERA, high GB rate, good DER; low ERA, low GB rate, average DER; high ERA, high GB rate, bad DER.

So which is it? Are the Cardinal defenders just having good days behind Stephenson? Do they all show up sick behind Simontacchi and Tomko?

Probably neither – the balls are distributed when those guys pitch in a manner to maximize or minimize their DER.

What I suspect is the issue is the line drive rate. Simontacchi and Tomko are probably their own worst enemy, allowing line drives for hits at a higher than league average rate – or at least than they as GB pitchers should (GB pitchers should allow less than average number of line drives because of the GB thing…). And when Stephenson could keep the ball in the yard (he allowed 30 home runs in those IP), he wasn’t giving up line drives, relative to what you’d expect a severe fly ball pitcher to allow.

I looked at the pitchers’ ball in play distribution at www.tendu.net. As ground ball pitchers, Simontacchi and Tomko both allowed a lower percentage of line drives than fly ball pitcher Stephenson.

What was even more interesting was the absolute abuse Brett Tomko took at the hands of his infield defense. Based on the hit distribution, it appears that the St. Louis third basemen and shortstops "pinched" the hole. Tomko gave up a relatively high number of balls down the left field line for hits, and according to the zone in which the balls were hit, they could have been fielded. The data indicated they were hard hit grounders, but the bunching wasn’t right down the line and there were very few hits in the hole. It is also very apparent the shortstop was missing a good number of plays up the middle, but still very playable for most shortstops.

My suspicion for Tomko went, not just unconfirmed, but essentially refuted. Without more specific data (I could watch all of Tomko’s games on MLB, but I’m not going to – it’s Brett Tomko and the frickin’ Cardinals), I’d have to say Tomko had one of two issues – his defense played poorly behind him, or the balls were just hit too hard to be fielded by even the best infielders.

Simontacchi did allow more line drive hits, but his sample is also smaller.

Mike Emeigh has pointed out to me that fly balls are turned into outs about 80% of the time. Ground balls are turned into outs about 50% of the time [75% of the time - ed.] and line drives about 20% of the time. I think those are the rates… Mike has also being doing some interesting work with line drive percentages – particularly with the NL Central.

Pitchers that throw mostly fly balls allow more home runs, and get more outs on balls in play. At least that is my theory. Have a pitching staff that throws fly balls and thus have a team with a better DER.

I have not run a massive study on this but I’m going to lean towards DER overrating the defensive play of teams that throw an atypical number of fly balls and vice versa. Here’s more from www.baseballgraphs.com.

I glanced at 2003 DER/GBFBip data, and it was largely inconclusive, as one season of anything can often be (AL correlation very high, NL correlation negative). This will result in several other things: the outfielders for those teams get more chances and those OF will get better reputations, and those OF will win GGs.

Which brings me to: DER is like batting average – you need to know the distribution of the balls in play to determine the quality of the performance.