Making Mountains out of Molehills: Teixeira and UZR
Author’s Note: this was originally meant to go on THT Live, I got distracted, and now it’d be too late (in my opinion) to post it, seeing as how the issue has mostly blown over. As poster Campanari noted in my previous post, I’ve been updating quite infrequently. I apologize for this, and hopefully I’ll finish most of my little projects soon. For those interested, you can read one of my entries at THT regarding hitting into the double play.
Mark Teixeira is one heck of a baseball player. Owner of three Gold Glove awards and a lifetime .391 wOBA, Teixeira’s a legitimate threat at the plate and a defensive wizard at the not so hot corner. With a career ISO of .255 and a BB% of 11.3, he’s both patient and powerful—in addition, he puts the ball in play quite a bit and has never hit below 30 home runs in a single season aside from his rookie year. His defense, according to the scouting report provided by The Fielding Bible, Volume II:
Teixeira has outstanding footwork around the bag, allowing him to prepare for throws of any variety, and he shows excellent range to both his glove and throwing sides. He fields flyballs with the ease of a natural-born outfielder. He also has great baseball instincts and a good arm for a first baseman, which he uses to cut down advancing baserunners. (262-3)
By all accounts, Teixeira’s a darn good defensive first baseman. The scouts rave about his defensive abilities, and both the fans and his teammates swear that he’s one of the best—if not the best—defensive first baseman in the game. You’ll be hard-pressed to find someone that will dispute this, and if you do, chances are they’re just trying to stir up some trouble.
According to the media, Ultimate Zone Rating is trying to do just that. UZR claims that Teixeira was -3.7 runs below an average first baseman in 2009. Heresy! The metric must be extremely flawed, or it’s overlooking something special about Teixeira that makes him infinitely better than the numbers suggest. At least, this is the assumption made by Bob Klapisch, who expresses many of the issues Yankee fans and players have brought up in regards to Teixeira’s negative rating. Interestingly enough, UZR isn’t the only system that says Teixeira was below average last year.
I assume UZR is being picked on because it’s regarded as the gold standard of defensive metrics, but it’s worth noting that other systems are in the same ballpark. Total Zone is the outlier, obviously. Baseball Prospectus’ Fielding Runs Above Average even rates him below average, and they’re not even incorporating precise batted ball data into their calculations. The results are somewhat alarming, as the Fans’ Scouting Report rated Teixeira as the second best defensive first baseman in 2009. With such a discrepancy between what our eyes tell us and what the numbers are saying, of course this has raised quite a few eyebrows. So, let’s try and figure out what’s going on here.
Is the problem with omitting scoops?
One of the concerns brought up by Klapisch, along with Teixeira and his supporting cast, is that UZR doesn’t track the countless throws saved by Teixeira. This is true—UZR doesn’t measure a player’s scoops (nor do the other systems). Klapisch writes,
Teixeira (…) says UZR is incapable of quantifying his most significant defensive asset: scooping balls in the dirt.
UZR isn’t incapable of handling scoops; it just doesn’t implement them into its calculation. Both Sean Smith and UZR creator Mitchel Lichtman studied the effects of first baseman scoops a while back, and both found the same general conclusion: its effect is not nearly as strong as some might expect it to be. By using a WOWY (“With Or Without You”) method, they looked at the rate of errant throws made to first base with the first baseman on the field and the rate of errant throws made to first when the first baseman is not on the field. Their findings suggest that a first baseman’s scooping value saves or costs approximately + 4-5 runs per year. But since we’re looking at limited samples for single seasons (Teixeira played 90% of all innings at first base for the Yanks in 2009), it’s hard to really get a feel for just how efficient the player is with this method.
A separate method used to look at scoops belongs to the folks at Baseball Info Solutions, who track “good defensive plays” and “defensive misplays.” The scouts watch every single play made by every single player in every single game—in other words, they’re everywhere and tracking everything. The list presented in late August of ’09 showed Casey Kotchman saved 9 more errant throws than the average first baseman from 2008-2009. Converted to runs, this is about +7 throws saved over two seasons, or about +3-4 runs saved per season. Right around what WOWY suggests. Once again, we’re looking at a noticeable, albeit relatively small impact a first baseman has on saving errant throws. Don’t these runs saved totals seem a bit low, though?
This is because we’re looking at scoops made in relative terms and not absolute. There is no question that a first baseman has to handle a number of errant throws—the question is, “how many throws does _______ save compared to the average first baseman?” We all know Teixeira saved countless runs by preventing errant throws; the question is how sufficient he was compared to the league average. The guys at Stat of the Week go out of their way to point out that over 2008-2009, Teixeira was exactly +1 throw saved above average. Data provided by STATS, Inc. didn’t have Teixeira in the top 8, either. That’s utterly shocking when considering the scores of fans and colleagues that swear to his ability to save everything thrown his way.
Consider this myth debunked. This isn’t the reason why Teixeira was rated below average in 2009, and even if he were at the top end of the spectrum at scoops, he’d still be right around average overall (at least by UZR). Let’s move on to another possibility.
Do the eyes lie?
He’s so good, he makes difficult plays look routine — and no software program could interpret that.
This statement raises the question: what does it matter if a player looks good? Style points don’t mean anything when it comes to value. I could do a backflip and then catch a fly ball, and it’d look much better than just standing in place and letting it fall into my glove. The end result is the same regardless of how we got there. I’m pretty sure there’s a lot more to defense than being really, really, ridiculously good-looking at it. Just because Teixeira looks good doesn’t mean that he is good.
Michael Salfino wrote an outstanding article (I’d call it a “must-read”) recently telling us that our eyes can be misleading. We often use availability and confirmation bias to form our opinions, and we quickly dismiss anything that is contradictory to be incorrect. Teixeira came to New York with a reputation of being a good defender (the defensive metrics support this too), and he replaced a well below-average fielder. Klapisch says,
Much of the Bombers’ affection for Teixeira is rooted in their memory of Jason Giambi, who was as graceful as Herman Munster around the bag.
This supports Salfino’s premise. First of all, Teixeira’s reputation as a good defender came well before he donned pinstripes—so everyone has a preconceived notion that he’s a good, if not stellar, defender at first. Second, he’s replacing a guy that’s well known for being a butcher at the bag. The difference between the two is going to be quite large.
Let’s say, for example, that Giambi’s five runs below the average first baseman and Teixeira is a +5. The swing between the two is 10 runs, which is huge. But Teixeira’s still just +5 runs above the average. He’s a great defender compared to Giambi*, but he’s “merely good” when compared to a normal first baseman.
The point I’m trying to make is this: Yankees fans and players are using Giambi as a reference point and not a league average first baseman. What results is an overstated perception of Teixeira’s defensive value. Is he an otherworldly defensive first baseman compared to Giambi? Absolutely! Is he an otherworldly defensive first baseman compared to the league average? Well, that’s open to debate.
This still doesn’t account for the negative rating though.
So, what’s up with UZR?
If a hitter goes 75 for 250, then he’s hitting .300. If he goes 69 for 250, he’s hitting .276. And if he goes 63 for 250, he’s hitting .252. The separation between a very good hitter and an average hitter is only six hits; the difference between an average hitter and a below average hitter is another six. It takes very little to go from “good” to “bad.” The “250” figure isn’t random—that’s right around the average amount of outs made by a first baseman in 2009. That’s not very many. If 250 at-bats yield issues with reliability, then 250 outs amidst a small pool of opportunities with zones that have different out conversion rates is even more problematic. Let’s look at it another way:
Let’s assume that 700 plays in the field are equal to 700 plate appearances, which is a full season at the plate. If this is true, then it would take a year and a half for a shortstop to equal one season at the plate. It takes a lot more for a first baseman—around 2.75 years—to be on equal footing with his offense. This is, of course, assuming that 700 plays are equal to 700 plate appearances. The number could very well be higher. What I want you to take from this is that we’re dealing with an extremely limited sample size here. We cannot draw any conclusions about Teixeira’s fielding talent from one season in the field. It is very possible that Teixeira simply didn’t perform in the field as well as he usually does, despite what he says. But it’s more likely measurement error more than anything else.
The problem doesn’t reside in the omission of scoops, although that would certainly help improve the quality of the estimation. The problem has something do with our biases, but that doesn’t account for everything, especially considering that Teixeira is generally rated as a good defender. The real culprit, as I just mentioned, is the excruciatingly small sample size we’re dealing with. There’s no doubt that Teixeira’s a good defensive first baseman- over the course of his career, he’s a +2 defender per 150 games- and one little hiccup here and there means nothing other than measurement error. When all is said and done, Mark Teixeira’s a fine defender. Let’s not make mountains out of molehills.
Then again, this is the New York media we’re talking about here…
*Ironically, the scouts showed Giambi as being two throws saved better than Teixeira (from ’08 until most of ’09). Furthermore, this little tidbit really caught my attention:
In 2008, Jason Giambi recorded 29 scoops for the Yankees in 112 games started, and Mark Teixeira recorded 36 for the Braves and Angels in 153 games started. In 2009, Teixeira has scooped 22 balls in the dirt through 95 starts and Giambi has scooped 14 in 58 games at first. While this looks like a victory for Teixeira, fans of Giambi will note that he plays a large number of games serving as his team’s Designated Hitter and doesn’t play as many games as Teixeira at first base. In fact, in 2008, Giambi’s 29 scoops for the Yankees were good for 0.26 scoops per game started, while Teixeira’s 2009 scoops for the Yankees are only 0.23 per game. This isn’t a significant difference, but there is nothing here to suggest that Teixeira is better at scooping errant throws than Giambi.