There are two men in front of you who want to be your team’s manager. One of them is fully up to date on all the latest baseball research. He reads Baseball Prospectus religiously, and that’s not a metaphor. He actually has a shrine to Dan Brooks in his bedroom. (We have a support group that meets on Wednesdays, that’s how I know.) He’s fully on board with the analytical movement, dabbles in his own research, drops the phrase “run expectancy matrix” into sentences, and has pledged that he will make sure that the supercomputer is in the dugout with him every night. He’s also rather boring. Not a jerk, just…boring.
The other guy is a complete troglodyte when it comes to the numbers. He doesn’t need a guy with a Ph.D. who quit playing after seventh-grade summer rec league softball telling him how to manage a baseball team. That guy’s just a talk radio caller who knows more acronyms. Bachelor no. 2 promises to hit the fast guy first and the guy who is really only good at bunting second, use his closer every single time he has a three-run lead in the ninth, and write headlines with bad puns on WAR in them. But my oh my, after listening to him speak, you would run through the Green Monster if he told you to. He embodies every cliché about being a “leader of men” that you’ve ever heard.
If you’d hire the first man as your manager, please turn to page 27. If you’d choose the second, please turn to page 28. Oh…and don’t do the thing where you keep your thumb there on the page so that if you didn’t like the outcome, you could come back. Cheater.
Warning! Gory Mathematical Details Ahead!
OK, let’s build the perfectly saber-savvy manager. I must give credit for the inspiration of this piece to a conversation between Rob Neyer and Jonah Keri on Jonah’s podcast last week. On that show, Mr. Neyer suggested that one thing that we don’t fully understand is how big a contribution a manager actually makes to the outcome of a game. After all, he’s not pitching or hitting or fielding, but he’s making a lot of strategic decisions, right? I’m not going to fully solve that problem today, but I think that there’s enough research out there that we can set some boundaries around how big or small that contribution can be.
For the time being, we’re going to look only at the in-game strategic decisions that a manager makes. We’re also going to play within the bounds of reality. We’re not going to give our saber-savvy manager any superpowers, just the knowledge he would have from the latest research. We’re also assuming that while our old-school option isn’t all in on that research, he’ll follow the same strategic “book” that everyone else seems to.
So, what does a manager actually do to affect the outcome of a game?
1) He decides who’s playing today.
Well, at least he decides which position players are playing today, because the starter is whoever is next in the rotation. Maybe you could make the case that our manager should be more willing to try some exotic rotation option, but many of those just aren’t realistic. The system that might be worth a try is that the manager would skip over the fifth starter when his team is not playing five games in a row and give those starts to the four better starters. Teams might be able to re-assign 10-12 starts away from the fifth starter (1/3 of his seasonal workload?). Assuming that he’s replacement level anyway and that a team could get an average “extra” performance equal to a league-average two-ish win starter in those 10-12 starts, maybe you clear 2/3 of a win.
As far as position players, it’s usually pretty obvious who should be starting, assuming no one is hurt or needs a rest. The manager is in charge of resting his players once in a while, and rest does have a very small positive effect (players who have had a recent day off hit ever-so-slightly better), but to get a day off, a player needs to be sat down and replaced by a bench player for the day. For an average regular starter, that value lost about cancels out any gain that might be made. I’d say it’s a wash.
Maybe there’s a case to be made for a manager being more willing to embrace platoons, although that’s very much dependent on having two players on the team who make sense as platoon partners. Even then, I’ve estimated that simply having a platoon on the roster is worth something like half a win at most over the course of a year, and that’s if you break the rules of substitution. And it’s not like the idea of platoons was some major sabermetric breakthrough, either.
2) He makes the lineup.
There are two things that you need to know about lineup construction: managers still make significant errors in putting together their lineups and, all told, those errors don’t hurt all that much. Tom Ruane did a study years ago in which he simulated every possible lineup, based on a team containing the average leadoff hitter, the average no. 2 hitter, the average no. 3 hitter, etc. He found that the difference between the perfectly optimized lineup and the “he’s clearly intentionally trying to sabotage the team” lineup was about a tenth of a run per game. The difference between the traditional lineup and the perfectly optimized lineup was roughly one hundredth of a run per game. Even over 162 games, you squeeze an extra run and a half by optimizing the lineup. That’s about it.
3) He calls for the sac bunt.
Sabermetric orthodoxy has been hard on the sacrifice bunt for years, but recently, I found that the way that it is commonly employed now is almost a break-even strategy. Bunts could end up as base hits or throwing errors, and the hitters who are asked to bunt are generally not run generators anyway. Teams lose an average of about .02 runs each time they bunt with a non-pitcher. Even over 25 sac bunts, that’s only half a run. Not half a win. Half a run. So, if our manager completely eschewed the bunt or just dramatically reduced his usage, he’s half a run better than your garden-variety tactician. Wow.
4) He calls for the stolen base.
The commonly accepted wisdom is that the offense needs to succeed on roughly 70 percent of its steal attempts to break even on SB attempts. Wouldn’t you know it, the success rate for stolen base attempts is around 70 percent each year. The problem with saying that managers should steal more (or less) is that stolen bases are subject to game theory. Sure, the manager can be more aggressive, but the pitching team can also devote more time to throwing over to first, pitching out, or changing the pitch sequence to give the catcher a little extra cheat time. Over time, offenses and defenses have played each other to a stalemate. For every move there is a counter-move.
Even still, in September of 2012, when Billy Hamilton was just a minor leaguer whose only real tool was that he was fast (he is now a major leaguer), Sam Miller looked at what Hamilton might do in a designated pinch runner role. Hamilton’s only job would be to steal second. Sam allowed that the manager could somehow magically intuit what the highest-leverage situation in the game to use Hamilton was and, although Sam never said this, it’s implied that the guy whom Hamilton replaced could come back into the game. The result? Billy Hamilton was worth a tenth of a win in the month of September.
If Billy Hamilton plus psychic powers plus breaking the rules of substitution is worth around half a win a year, there’s probably not a lot of blood to be squeezed out of this turnip.
5) He calls for the hit-and-run.
Mike Fast did a study on the benefits of the hit-and-run, although in his data set (2003-2011, before Mike went to that great juice box
in the sky in Houston) he found that teams don’t use it a lot. The average team used it 35 times per year, and somewhat by definition, did so only in situations where it made the most sense—like having a high-contact hitter up and a decently fast runner on base. Their reward was about .06 runs per attempt over and above what you might expect from a situation with runners on and a decent hitter up, according to Mike’s math. Even if we make a silly assumption that a manager could pull out that much extra value any time he used the hit-and-run, and he pushed that button as much as the most hit-and-run happy manager (Mike Scioscia, who averaged 60 attempts), congratulations, you just won a run and half.
6) He calls for his fiddlers three.
I just wanted to see if you were still paying attention. He also calls for ball four in the form of an intentional walk. There are some howler intentional walks issued, but those are so notable because they’re rather rare. Much like bunting, intentional walks are handed out in specific circumstances, and the cumulative effect is likely much lower than we might imagine.
7) He decides when the starting pitcher has had enough.
It’s hard to really rate a manager here, because the decision of whether to give the starter the hook will depend on how rested (and how good) the bullpen is. Even after off-days, when the bullpen is more rested, managers in general don’t seem to have a quicker hook, possibly because they have to think about the long-term bullpen workload as well. There probably is a correct solution for how long a pitcher should go if we make the assumption that this is Game 7. It’s harder once you figure in that it’s not usually Game 7. Maybe there’s room for efficiency here, but I haven’t seen it yet.
8) He decides who pitches in relief, and when.
This is a favorite of the acronymati. We look forward to the days when the three-run, ninth-inning save will become a curious fossil. We decry the manager’s reluctance to use his closer in extra innings on the road until he gets a lead. Those are, of course, mathematically incorrect moves, but how much do they really matter? On the first matter, I once studied a system where closers no longer pitched in the ninth inning with a three-run lead, but rather yielded that honor to their set-up men. Instead, they handled the eighth and ninth innings of games within one run (and the ninth inning of two-run games). I estimated that it would save, on balance, about a tenth of a blown save over the course of a season. Not a tenth of a win. A tenth of a blown save (which could still turn into a win).
As to using the closer on the road in extra innings, I looked at all ninth and later innings in which the home team was batting and entered either down one (when, presumably, the closer would be pitching) or were tied (when, for some reason, it seems to be a good time to bring in the long reliever). In save situations, the closer gave up a run 14.6 percent of the time. In tie situations, whoever was out there gave up a run 25.5 percent of the time. Managers, when they absolutely need a goose egg on the board, are instead opting for pitchers who give away 10 percentage points of win expectancy, although they’ll probably give some of that back in the fact that if they do get a lead, they’ll have to send someone not as good as the closer out there to protect it. In general, the closer is really only going to pitch one inning. In 2013, the average team was involved in about 16 extra-inning games, which suggests that eight of them were on the road. Let’s say our saber-savvy manager is able to grab an extra eight-tenths of a win there.
That’s not an exhaustive list of the things that a manager does in a game, but it covers most of the things that sabermetricians like to complain about. If you roll all of these into a big ball, including the faulty assumptions, and squint really hard, we’re talking about two wins’ worth of inefficiency when it comes to manager tactics. Most of it comes from two moves (skipping the fifth starter more often and using the closer in tie games on the road) that most managers—even the really saber-savvy ones—don’t seem to be too keen on. The real value that we might expect is a handful of runs. To be sure, a lot of it is free money sitting there, and every little bit helps.
But let’s go back to our other managerial candidate who will leave that money on the table. We said that he may not be up to date on the latest research, but his charisma could power a cult. Let’s see if we can put a value on that. Let’s start with the idea that on the team might be a position player who has fallen into some bad habits around getting enough rest. He’s 25, a multi-millionaire, and enjoys going out after the games to experience the wide range of cultural events that each city has to offer. He’s not staying out all night, mind you, nor is he doing anything illegal or particularly harmful to himself. (Gasp! He might enjoy a beer, something that no 25-year-old has ever done after work.) He’s just out a bit later than he should be. He knows that he should get seven or eight hours of shut-eye per night, but…well, six is about the same, right?
Not really. He’s entered into what sleep scientists call chronic sleep restriction or what you call “Oh, just one more re-run of Perfect Strangers.” There’s a research study in which participants slept an hour less than the recommended amount per night over the course of two weeks. They took a number of neurological tests during that time, including a sustained attention task. This is a pretty standard test. Letters or shapes or symbols (it varies based on which test you’re using) flash on the screen, and when you see one, you have to hit a button. The computer keeps track of how long it takes you.
After one night of only slightly interrupted sleep, the rate at which participants had a more than a 500-millisecond (half a second) reaction time more than doubled. It might not seem like much, but that sort of reaction time is a bad sign. And the effects build up over time. After a week and a half, the rate of these lapses is roughly four times that of a well-rested person, and equivalent to someone who stayed out one night and did not sleep at all. Or someone who is drunk.
A perfectly reasonable reaction time is something on the order of 250 milliseconds, so we’re talking about someone who is more prone to being a quarter-second slow. What can a quarter second buy you in baseball? Well, a 90 mph fastball crosses home plate in about 10 milliseconds, and the entire length of the batter’s box in 45 milliseconds. If you are slow, the ball might already be by you, and if there’s something that research on catcher framing has taught us, it’s that the ability to steal a few extra strikes might be the most powerful force in baseball, because turning a ball into a strike is worth 0.15 runs and there are 150 pitches or so per game. If a batter is having very slow reaction times on two out of what I assume is 120 targets on the reaction time test on his first day of sleep restriction, then he’ll be having reaction time issues on 1.6 percent of pitches. Over the course of a season—say, 2000 pitches faced—he’ll be ultra slow on 33 of them. If half of those become strikes that would have been balls (ignoring that some of them would have been balls put into play), he bleeds away 2.4 runs of value. The more sleep he loses, the bigger the impact. And that doesn’t count the effects on his fielding.
Now, our charismatic manager comes along and sits down with our sleepy player and says, “I don’t want you to give up going out after games. You’re a big boy and what you do is your own business. But, I’m asking that you do think about the team and come in an hour earlier than you have been.” I once defined team chemistry as the answer to the question “Why should I bother?” Our manager gives the player a reason to bother. And in doing so, he saves the team a quarter of a win in what would otherwise be lost value.
And that’s just one player, doing one thing that can bleed away value. And it doesn’t take into account all the other things that lack of sleep can do to a player. (Lack of sleep impacts the pre-frontal cortex, which is also involved in attention, pattern recognition, planning, and higher-order thinking.) And it isn’t all that hard to believe that the manager could make it happen.
Beer, Tacos, and a Good Night of Sleep
The careful reader is probably already yelling at the screen accusing me of setting up a false choice. One can be a truly inspirational leader (or at least have some good people skills) and be up to date on the latest and greatest research (or at least have an open enough mind to listen to a reasonable argument). I started out by trying to answer the question of how much of an impact a manager might make on a team with his strategic decisions. It turns out that while the answer isn’t zero, most of the things that we obsess over with managers aren’t really that powerful. But we do obsess over them.
After I had answered that question, it occurred to me that a manager might be able to do something just as powerful (or more so) through his work behind the scenes. Whether the value of the behind-the-scenes work is greater or lesser than that of the strategic pieces is unimportant. (Also, we haven’t even talked about the manager as the leader of a coaching staff—something that we’ve seen can have very big effects on a team.) A reasonable examination of the numbers shows a decent case that the two pieces are on the same order of magnitude, and yet when we evaluate managers, we focus only on the strategic matters. Yes, the strategic matters are so much easier to study, but that doesn’t excuse us from trying to assess performance in other areas. And now that we can at least make the case that the behind-the-scene work is at least on par with the strategic issues, we should probably reserve judgment on whether a manager should be fired based solely on his in-game decisions. Sure, maybe he should be fired, but is he being given a fair evaluation of his entire job description?