The Decision To Sit Swisher

From Dan Rosenheck:

In switching Jerry Hairston Jr. for the slumping Nick Swisher, Girardi placed far too much weight on Swisher’s disappointing 38 postseason plate appearances so far — and not nearly enough on the fact that Swisher’s combined on-base and slugging percentages were 159 points higher than Hairston’s in 2009, and 117 points higher over their careers. Girardi chose to sit his powerful right fielder in favor of a player who barely hits enough to play shortstop. Similarly, concerned about Burnett’s supposedly poor synchronization with Jorge Posada, he swapped out his Hall of Fame-caliber catcher for Jose Molina, who has a decent claim to be the worst hitter in the major leagues.

The question of whether “streaks” have any predictive value — whether we should significantly adjust our forecast of a hitter’s performance because he has been “hot” or “cold” in the preceding few weeks — has been studied exhaustively, and analysts are all but unanimous in concluding that we should not.

As Tom Tango writes in “The Book: Playing the Percentages in Baseball,” “Always assume that a player will hit at his projected norm [adjusted for the park, weather, and pitcher he is facing], regardless of how he has performed in the very recent past.”

I did not agree with the decision to sit Swisher, as I felt he was a better bet both offensively and defensively than Hairston even while slumping. However, the reasoning that Rosenheck provides, a favored precept among sabermetricians, seems specious to my admittedly untrained mind. Their studies show that a player in a slump should not be expected to deviate from his career norms in future games. However, just because this idea holds true on a broad level does not mean that a manager should use it when dealing with an individual slumping player.

I would suggest there are two kinds of slumps. The first type is when a hitter is not having any balls fall in, meaning he has not changed his approach at the plate and is not doing anything that would warrant time on the bench. The second type of slump is when the hitter is clearly mechanically off, and could benefit from a day off. If a manager determines that the player is in the second group, does it really make sense to keep him in the game because on a global level, slumps are not predictive of future performance? I would think not. The manager has to address the individual situation, and try to figure out whether this player is doing something clearly anomalous at the plate that is causing his poor performance. In fact, isn’t it possible that the data in the study is skewed by manager actions that help the player out of said slump?

Personally, I thought the fact that Swish was leading the team in pitches per plate appearance in the playoffs meant that he had not changed his approach, and that the hits would therefore come. However, if Girardi felt that he was in the second category of slumping players, a day off probably made sense.

0 thoughts on “The Decision To Sit Swisher

  1. I agree with Rosenheck on this one. If something is mechanically off, that doesn’t mean it will be mechanically off the next day. In fact, if this were the case, it would show up in the study. I don’t necessarily see why a day off would change that. The only slumps that have proven to be predictive are ones that are related to injuries. If Swisher had some nagging injury, I could see him being given a day off, but I haven’t heard anything suggesting that.

    I think that benching Swisher was the wrong move, but using Hairston was way worse. Hinske and Gardner are both superior options to Hairston, so that is the major crime in my eyes. With the defensive improvements of Gardner in CF and Melky in RF, the difference there isn’t that stark. Substituting Hairston in for Swisher is inexcusable, though.

    I highly recommend reading through this thread:

    MGL is pretty harsh, but what he says is all backed up with data.

    • Greg F.: If something is mechanically off, that doesn’t mean it will be mechanically off the next day. In fact, if this were the case, it would show up in the study.

      Why? The study does not say that every single slump exhibits the same tendencies, just that the number of slumps with predictive value is so small as to be statistically insignificant. I’m suggesting that those slumps that are anomalies are the ones where the hitter is off mechanically and therefore might need a day off.

      I like the Book blog and MGL’s work, but sometimes I think he has a slavish devotion to the “macro” type data that may not be applicable in every “micro” case. I’m suggesting that if the manager identifies this slump as the 1/100 type slump that needs a day off, he should give the player a day off. Also, managers have been giving slumping players a day off since the dawn of time. Who is to say that the results of the study are not skewed by the fact that a day off or something of that nature helped the player from his slump?

    • Oh, and as to your second point, I fully agree. Gardner should have gotten the nod.

  2. Really, nobody cares that Hairston had better numbers against Pedro than anyone in the lineup? And Swisher is not a better outfielder than Hairston. Hairston 12.5 UZR/150 in the outfield this season, Swisher -1.9.

    Plus Hairston got a big hit late that lead to a bigger insurance run. So the move worked. I mean he didn’t hit 2 home runs or anything, but he helped score a run in a game that only had 4 of them.

      • Two careers, one guy always moves around the outfield and has always been a plus defender, another guy has always been average-below average. Now I’m supposed to believe that it was a bad choice, at least defensively, to go with the guy who has always been good in the outfield? Please.

        • Yeah, but Hairston has barely played the outfield in his career. Less than 200 games started in the outfield over an 11 year career, so I wouldn’t go by his UZRs. You say that Swisher has always been average-below average, his career UZR/150 right around +5.

        • Swisher has a better career UZR in RF (UZR/150 of 4.9, compared to 3.0), so I’m not sure what you are talking about. The numbers say he is better out there, he plays out there a lot more consistently, and Hairston barely played there at all over the last few seasons.

  3. I also disliked that article because in it the writer quotes the study saying that past slumps don’t indicate future numbers, but totally ignores the line about how the pitcher will effect the numbers and ignores Hairston’s numbers against Pedro.

    • Well, the fact that Hairston was good against Pedro in 27 at-bats doesn’t mean that much. As with most hitter-pitcher matchups, the sample size is too small to really conclude anything. Also, Hairston hadn’t faced Pedro in over five years, making that even more irrelevant.

    • Head to head numbers from 6 years ago are largely irrelevant, especially when the pitcher has changed as much as Pedro has.

        • He uses his fastball a lot less. He threw 10 more changeups than fastballs yesterday!!! Far from the same guy.

  4. I disagree with Rosenheck, if it was the regular season where he has the time to make up for the bad stretch then fine, but with only 6 games left in the season can you take that chance.

    We all know that Swisher is streaky he’s either red hot like he was in April, or awful like in May. He also has times where he is in between, but he hasn’t looked good at all this post season. Yes he is seeing pitches but he’s also doing nothing with them, I can only recall two hard hit balls the double against the twins, and the single in game 6. He looks a lot like the May version right now, I agree with sitting him for a game. I wouldn’t have used Hairston, but the move worked out.

  5. If you watch the games on, you can see that Swish is like a .150 hitter against low and in and low and away balls. Which is obviously terrible.

    Could it be that against better pitchers, who can locate, he just stinks?

    • This is a really interesting comment. With all the micro- analysis of hitting- by ballpark, count, position in the batting order, high/low leverage situation etc.- it would be fascinating to see how a hitter,such as Swisher does against various levels of pitching. Specifically, divide all the pitchers in each league into 4 (10?) groups based on WHIP or ERA or some combination of the two. Then look to see how a batter, like Swisher, does against the various levels. Does he feast against the bottom feeders in WHIP or ERA or does he hold fairly steady against all levels of pitching? Taking a look at all the players in the league based on this analysis, you would get a good idea of who the real studs are- the guys that can handle all kinds of pitching including the elite- Greinke, Doc, Lester, etc.- as well as the guys who do most of their heavy lifting against 5th starters and 11th and 12th men on the staff. Are there guys that actually hit better pitching better and a whole myriad of other questions can be uncovered. Additionally, this might be of some predictive value when the playoffs come to see who might be more susceptible to severe drop offs in offensive output as the pitching presumably is better than the league as a whole because they built their numbers against the lower end of the group.

      • Those stats are being kept right now but, not (I don’t think) at the big league level. As they come up through the minors they are charted. So if one guy is hitting .400 in A ball, then goes to AA etc., and hits (almost) as well or better, I think one can assume that the guy is a good hitter and may even get better.
        This is not to say some of the teams don’t keep these stats, if they do…they aren’t talking. It sounds like some equation could be made to work those numbers into a reality.