Skip to content

Memo to mainstream media: Girardi relying on hitter-pitcher matchups is not sabermetrics

Oct 25, 2010, 2:16 PM EDT

Texas Rangers v New York Yankees, Game 3 Getty Images

With a few exceptions the media members in New York are ripping Joe Girardi to shreds following the Yankees’ lackluster showing in the ALCS and many of them are focusing on the fact that Girardi places a great deal of importance on hitter-pitcher matchups, often mocking the material contained in his “binder.”

Along the way many of those same media members have somehow convinced themselves that Girardi is engaging in sabermetrics by focusing on those matchup numbers, which allows them to do the two-birds-with-one-stone thing and rip both Girardi and stat-heads.

Here’s a prominent example, from Bob Klapisch of

Girardi’s over-reliance on numbers failed the Yankees time and again in the final week of the season, revealing a lack of trust in his own instincts. As a result, the Yankees never found that “on” switch in Game 6, because Girardi didn’t know how to ask for it. It’s not in his nature to peel away the layers of his players’ psychological flesh. Instead, the manager relied on matchups, trends and data–all the trimmings of the sabermetric era.

Here’s the problem with that notion: One of the basic tenets of sabermetrics is that individual hitter-pitcher matchups with track records consisting of some small number of plate appearances has very little predictive value. Girardi isn’t alone in relying on those small-sample numbers and in fact it’s a common practice among big-league managers, but doing so is the exact opposite of sabermetrics.

Klapisch sees Girardi looking at numbers and simply assumes those numbers equal sabermetrics, but that couldn’t be further from the truth. Any decent sabermetrician would tell you that one batter being 1-for-6 with three strikeouts and another batter being 3-for-8 with two homers versus the same pitcher over the span of multiple seasons has close to zero value and in fact Girardi’s reliance on such data has been criticized by various stat-friendly Yankees writers and bloggers.

If media members want to criticize and mock Girardi for making decisions based on hitter-pitcher matchups and for keeping that data in a binder, go right ahead. Just don’t attach that type of thinking to sabermetrics.

  1. mcnater - Oct 25, 2010 at 2:23 PM

    Well said. Is this stuff really THAT hard comprehend?

  2. Mr. Jason "El Bravo" Heyward - Oct 25, 2010 at 2:29 PM

    The Yankees lost because they spent one dollar too few on their payroll. One more dollar spread through the team and ARod would have hit that HR, Burnett would have pitched that perfect game, Jeter would have gone 4-for-4 every game, CC would have had two no-hitters, and Girardi would have been manager of the year. Next year, Yankees fans, next year. Just open those wallets for once and spend some money! Gosh!

  3. smparsons - Oct 25, 2010 at 2:31 PM

    except that he is using “Sabremetrics” – he’s just not, if you believe these accounts, using them correctly or with adequate controls.

    Kinda like you are your BABIP.

  4. Mark Armour - Oct 25, 2010 at 2:37 PM

    Sabermetrics is the search for, and use of, objective knowledge about baseball. As such, I would argue that Girardi’s use of his binder is very much sabermetrics. He is using facts and knowledge at the expense of his gut. You may think he is overusing or misusing these facts, but he is using facts just as surely as John McGraw and Earl Weaver did.

    • okobojicat - Oct 25, 2010 at 3:01 PM

      You aren’t even listening to Aaron’s argument. He’s arguing that those stats aren’t “OBJECTIVE KNOWLEDGE!” because the sample size is so small. They are pointless. Let’s say I road the subway to work, and that 6 of the first 10 times I sat next to a white guy with a big beard, he smelled of meth. Does that mean I should start reporting to the cops every single time I see a white guy with a beard as a meth user? No, it means I should probably find a different way to get to work.
      Girardi is using stats, but if he’s using X batter vs. Y pitcher, that’s not good application of stats. However, if he’s using stats like Y pitcher vs. LH batters, or Y pitcher vs. hitters who hit curveballs well, then that would be a sabermetric approach.
      Just using stats /= sabermetric. You must use them intelligently and most importantly, know where stats don’t provide any knowledge and should be downweighted or even perhaps discarded.

      • Mark Armour - Oct 25, 2010 at 3:49 PM

        I disagree.

        If you go outside and observe where the moon is, and you write the co-ordinates down in your notebook, that is “object data” and the recording and analysis of that data is “science”. Whether there is enough data there to determine where the moon will be tomorrow is an interesting question, but this does not make the data any less factual, nor the process any less of a science

    • The Dangerous Mabry - Oct 25, 2010 at 4:09 PM

      By that argument, the fact that he’s wearing a hat means he’s involved in fashion, the fact that he writes out a lineup card means he’s a calligrapher, and the fact that he walks out to the pitcher’s mound means he’s a hiker.

      • Mark Armour - Oct 25, 2010 at 4:14 PM

        It means nothing. Sabermetrics is a science, and as such it is based on objective and measurable facts. Each event that takes place on a baseball field is a fact worth measuring and using those facts is sabermetrics. The more facts you have, and the more facile you become with those facts, the better the sabermetrics is. But the fundamental premise of sabermetrics from the beginning has been to use objective data over your gut or your instincts. This is not difficult.

    • cktai - Oct 26, 2010 at 2:33 AM

      Gotta hate statheads like Joe Buck and McCarver who keep whining about Howard’s lack of RBIs and all the writers wile Paul Hoynes who care about nothing then the win stat.

      Actually sabermetrics is the analyses of baseball statistics. Part of this analyses is noticing when data is insufficient or just plain wrong. This is why sabermetricians are cautious in using batting average and shy away from RBI and pitcher wins altogether. Sabermetricians will acknowledge the danger of small sample size, and (s)he will know that the larger sample of the whole season against multiple pitchers is more valuable then the small sample against a specific pitcher. Proper analyses will tell you not to follow hitter/pitcher matchups. If you decide to follow it anyway then you are as much a sabermetrician as an astrologer is a physicist.

  5. Jonny 5 - Oct 25, 2010 at 2:53 PM

    I guess he should hire a Phsychiatrist and a fortune teller instead of looking at the facts? It’s funny how soooo many people could always do so much better of a job than the guys who get paid the big bucks. Prepare yourself for 20 different ways the Yanks and Phills SHOULD have been managed, You know, for the gauranteed win they both deserved. As I roll my eyes….

    • rbabaseball - Oct 25, 2010 at 4:00 PM

      Respectfully, using data in this way is exactly what makes it “not science.” Statisticians abhor the use of small sample sizes to make any predictions and in fact would be loath to say *anything at all* based on 10, 15, 20 plate appearances. To follow the analogy of the moon, using data the way that Girardi does it (or as is espoused here) would be like looking at the moon’s position on August 23 at 8:43 PM, November 2 at 10:51 PM, and January 14 at 9:44 PM and then trying to predict where it will be on May 5 using just those 3 readings. Sure, the data collection is scientific, but the *use* of it is far, far from it.

      This is what sunk Elias vis-a-vis Bill James; Elias was always trying to make the most minute diagnoses (which are useless) while James was trying to approach it scientifically. Unfortunately, those garbage statistics have spawned a whole generation of people like Girardi who thinks that they mean something.

      • Mr. Jason "El Bravo" Heyward - Oct 25, 2010 at 4:22 PM

        If a Sabermatician eats a big piece of poo on three separate occasions and determines the poo is not good to eat at least 2/3 of the time, does that mean we should eat poo?

  6. rbabaseball - Oct 25, 2010 at 4:36 PM

    So the question here is whether the term “sabermetrics” implies “sound baseball analysis.” If you think it doesn’t necessarily imply that, then you can believe that the small-sample-size, Joe Girardi tactics are sabermetrics. I, however, disagree. It may be semantics, but I believe that “sabermetrics,” like the term “statistics,” implies sound analysis.

    For example, if I flipped a coin 5 times, collected that data, and then used it to presume that the coin would come up heads on the next flip, that would *not* be sound statistics. If I had put that on one of my tests in grad school, I would have likely been kicked out of the program. Thus, I wouldn’t say that I was using “statistics” to make that conclusion. I see this in the same way. If the data is used incorrectly, or at least to make a conclusion that isn’t sound, then it’s not sabermetrics under my definition of the term.

  7. quintjs - Oct 25, 2010 at 5:15 PM

    Any amount of data, even one piece of data is valid and a fact, but it has no predictive value, and you need to look at what the data is.

    PRIME example and the worst by far of all of Girardi’s decisions – starting Hughes in game 2 (& therefore 6) in Texas because he had good numbers in Texas.

    He was a rookie then, unknown. It was years ago, it was against a Texas lineup where half of the guys he faced are out of baseball because they were so terrible. Texas now have a really good potent lineup with speed and power.

    While the numbers were good, if you looked into the numbers, you would see it was a small sample size, and of no real value. And that was the stated reason by Girardi.

    • Kevin S. - Oct 25, 2010 at 6:14 PM

      In Girardi’s defense, he actually did that to allow Pettitte as much time to rest as possible.

  8. churchoftheperpetuallyoutraged - Oct 25, 2010 at 6:48 PM

    It wasn’t just the previous starts against Texas. Hughes is also an extreme fly ball pitcher, and in YSIII that’s not a good thing (SSS alert). He was also better on the road than at home (mostly due to the previous comment).

    • Kevin S. - Oct 25, 2010 at 8:28 PM

      As Dave Cameron over at Fangraphs noted, the same issues applied in Texas as in YSIII.

  9. lordd99 - Oct 26, 2010 at 4:33 AM

    Not only is this not fair to sabermetrics, it’s probably also not fair to Joe Girardi.

    Girardi does have a lot of statistics at hand, and studies probably as intensely as any MLB manager. He was an engineering major at Northwestern. I think we all know engineering types. They like their numbers. Let us also not forget that Girardi was a professional MLB player and a catcher. He of all people knows that small sample sizes mean nothing. A player being 1-6 against a specific pitcher is simply a data point, but it not the determining factor. As I’ve heard Girardi explain at varies times, he’ll look at the match-up of hitter vs. pitcher. Does the pitcher throw a slider? Is the hitter good against sliders? He’ll consider how the hitter actually performed in the ABs against specific pitchers, not necessarily the final result, because a small sample size means nothing. The player going 1-6 might have actually scorched the ball all six times against the pitcher. He considers all these aspects, but just because he binder says the player was 1-6 doesn’t mean that stat determined anything.

  10. obo1892 - Oct 26, 2010 at 8:51 AM

    Suppose a=b

    Does misusing math demonstrate a flaw in the use or does it demonstrate a flaw in mathematics itself?

    • obo1892 - Oct 26, 2010 at 8:56 AM


  11. fquaye149 - Oct 26, 2010 at 9:55 AM

    Let me get this straight, all you commenters arguing that Girardi is using “SABRmetrics” when he uses the splits book. You are arguing that Girardi is practicing SABRmetrics by using numbers because he is using numbers when he looks at splits rather than looking at non-numbers like season long Batting Average or Home Runs like other non-SABRmetrics-using, non-numbers-using managers might have? Gotcha….

Leave Comment

You must be logged in to leave a comment. Not a member? Register now!

Top 10 MLB Player Searches
  1. B. Crawford (2979)
  2. C. Correa (2690)
  3. G. Stanton (2685)
  4. Y. Puig (2622)
  5. G. Springer (2583)
  1. H. Pence (2443)
  2. H. Ramirez (2327)
  3. M. Teixeira (2276)
  4. J. Hamilton (2270)
  5. J. Baez (2165)