Baseball Therapy: Paul Molitor is the Twinspiration

We now pause for a moment to ponder the mystery of the Minnesota Twins. Up until last week, they were in first place in the AL Central. In the offseason, some made the case that the defending AL champion Royals would win the division. Others argued for the defending division-champion Tigers. A few picked the up-and-coming Indians or the White Sox after they signed Zach Duke. No one expected the Spanish Inquisition the Twins to be in first place for any of 2015. This is the team that has specialized in losing 90 games over the last four seasons.

Not only that, but it has been pointed out that the Twins' success this year came despite not being particularly good at hitting or pitching, at least one of which is usually required to win games. The team seemed to have a knack for bunching hits together. And they have a new manager in Paul Molitor, after years and years of working under Ron Gardenhire. It's just that it's the same old Twins. It's not like they went out and got Torii Hunter in free agen… oh right.

Fire up the Mad Lib narrative machine! Molitor, being a (adjective) manager, got (preposition) the (body part) of the Twins and through (encouragement technique) and (vague platitude about fortitude), and really (verb) home to them the importance of doing their best in key situations. But because of Molitor's (character attribute), the Twins are able to (verb) deep and play above their (body part). Molitor isn't just a manager. He's the meaning in their life. He's the inspiration. Gardenhire just didn't have it. He was too much of a (adjective) (animal) (type of spaceship).

Really, this is just a new take on the old clutch debate. Are there hitters who can turn it up a (unit of measurement) when the (cliché that means “pressure situation”)? In general, we find that, statistically, players generally perform about the same whether it's a pressure situation or not. There are certainly clutch hits, and there is some evidence that some sort of clutch “skill” might exist for hitters, but the effect is rather small and we'd need about ten years' worth of data to tell who is good and who isn't. So, like a lot of things, the answer is, yes there is some reality to it, but the magnitude is way overblown.

Is there something that a manager could do that might make a difference? We know that managers can have a consistent (and big) effect on their hitters (and pitchers) when it comes to helping them stay sharp over the course of a long season. Maybe there is something to all of those movies where the manager gives a big speech right before the key at-bat. Then again, maybe Casey's manager gave him a pep talk too, and we all know how that turned out for Mudville.

Warning! (Adjective) Mathematical Details Ahead!
When we talk about pressure situations, we usually go to a measure known in sabermetrics as “leverage,” which was formalized by Tom Tango. It's a way of mathematically saying how important a particular at-bat in a baseball game is. When the score is 15–1, the result of this at-bat means virtually nothing to the outcome of the game. If it's 2–1 in the bottom of the ninth with two runners on and two out, the game could be won or lost right there. Leverage produces a value that allows us to directly compare two situations. A situation with a leverage value of 2 is twice as important to the outcome of the game as one with a value of 1. Players may not think directly in these terms, but it's also a pretty good proxy for what's going on.

I took all regular season events from 1993 to 2014 and generated the leverage value of each. I also found managers who had been managing for more than 30,000 plate appearances during that span. (They didn't have to be consecutive.) That represents about five full years of experience. In part, that high cutoff makes sure we had enough of a sample to work with for each manager, but it was also to give my poor processor a break. There were 45 managers who qualified.

For each plate appearance, I coded it as ending in a strikeout or not. To make sure that I didn't give managers credit for having amazing (or awful) players on their teams, I created a variable based on the batter and pitcher matchup present in each plate appearance to control for how likely that pairing is to produce a strikeout. (This is called the log-odds method, which I have previously described.) I did the same for a number of other outcomes (walk, home run, on-base event, out on a ball in play, and so forth). To be included in the sample, the at-bat had to feature a batter who had 250 PA in that season against a pitcher who faced 250 PA in that season.

I created a series of binary logistic regressions, predicting whether the at-bat would end in a strikeout (or some other outcome). I inserted the control variable first. If that variable—and only that variable—were significant, then it means that the outcome of the at-bat is just a function of the relative skills of the batter and pitcher. But also entering the ring was the leverage value of the event (perhaps some events are just more common when the leverage is higher?), the name of the manager (categorical), and the interaction of manager and leverage. The really important variable here is that interaction term. If that one is significant (and for the initiated, I looked at the overall Wald value for the variable), then it means that for certain managers, there are differences in the likelihood of outcomes as leverage changes, over and above just the guys who are standing on the mound and in the batter's box. It wouldn't necessarily mean that particular manager is amazing when it comes to inspiring his players to (cliché) in a pressure situation, but it also doesn't mean that he's a (animal) (orifice) either.

The result? A complete shutout. The control variable came up strongly significant (it always does), but nothing else reached significance. At all.

Okay, well, maybe managers can't get their hitters to perform when the game is on the line, but what about the pitchers? Is there some magic that managers can weave to help them rise to the occasion? I ran the same analyses, only this time coding for the pitcher's manager as the guy of interest.

Same results. Nothing to see.

Luck Conquers All
From what we see here, there does not appear to be any effect the manager has on his players that makes them better in pressure situations than they we would otherwise expect them to be. Molitor wasn't in the sample, so we don't know specifically whether he would have turned the regression, but it doesn't look like the effect is there. Even if there is an effect, it is likely very small and not worth bothering over.

It's not that it's impossible to believe that at some point, Molitor (or one of his coaches) may have called a player aside and said “Hey, if you see (name of player) do that thing with his right (body part) it probably means a (type of pitch) is coming, so be on the lookout.” And then because of that, someone may have gotten a key hit. Maybe one day, Molitor did say something particularly uplifting to someone. It's possible that a few of the Twins have learned new skills from Molitor that have helped them during all parts of the game. It's also possible that the Twins all really like Molitor and really want to do well for him. But there's no evidence that a manager—any manager—can make his players come through in key situations more than we would otherwise expect of them.

So, the answer to the question of why the Twins put together two months of amazing is most likely “They got really lucky for a couple of months.” Yes, it just happened to be right after Paul Molitor took the reins, but repeat after me: Just because two things happen together doesn't mean that they are linked.