May 8, 2015
I want to try something a little different this week. Oddly enough, I have to confess that it was inspired by Yordano Ventura. Ventura and his Kansas City Royals, who apparently think that they’ve won an American League Championship recently, have gone from the feel-good story of October 2014 to the feel-kinda-creepy-when-you-watch-them story of April 2015. Ventura is now serving a seven-game suspension for his role in a brawl with the White Sox and earlier was involved in a beanball war with the A’s.
There is, oddly enough, an etiquette around beanballs and even brawling in baseball. Retaliatory beanings are supposed to hit the batter in the butt, where he will simply wear a bruise for two days. At that point, everyone knows that throwing at each other will not be tolerated. If a brawl is necessary, everyone should run out onto the field and make angry faces at each other so that everyone can see that they would fight, if they had to. If punches are actually thrown, it’s recommended that a few players from each team break up the fight before Don Zimmer gets in there and nuclear weapons start getting launched. When you think about it, it’s a rather elaborate dance, but it serves a purpose. Everyone gets a chance to look like they’re not backing down and they can all get it out of their system in a way that keeps anyone from getting seriously hurt.
Afterward, everyone is supposed to talk about how “We’ll stand up for our guy. We’re all looking out for each other.” To the casual observer, it makes no sense for a pitcher to intentionally give the other team a runner on first and to potentially get himself thrown out of a game and suspended, not to mention the senseless violence aspect of it. Within the culture of the game, it all makes sense. Baseball is a game where there is very little protective gear and there’s a hard little projectile that travels around at 100 mph and the shoes have spikes on them. One little idiot who wanted to prove… something could seriously injure someone and cost that guy his livelihood. Knowing that the injured guy's teammates would not be pleased and would have equal access to those weapons means that baseball often polices itself into a détente. But it also means that you need to have a good esprit de corps. If everyone hates each other so much that they won’t back each other up, then suddenly a baseball game becomes a very scary place.
If it was only for this reason, we would still see a lot of talk about the idea of team chemistry in baseball. Leave aside the fact that a baseball team is like being locked into a cargo van with 24 other guys for six months or that they all have to work together. Or at least not murder each other.
Team chemistry is still one of those great mysteries of baseball. We have some idea that it matters, but little definition of what it is or how to measure it or how much it’s worth. In the past, I’ve fantasized about the idea of measuring chemistry by doing all sorts of fancy social network analyses in clubhouses and trying to correlate that to performance on the field. The problem is that I’d basically need a research team of 25 people or so and the majority of MLB teams to grant me full access to their locker rooms. That’s not likely to happen.
But, what if we had a way to measure chemistry from things that could be observed by the public? Or at least things that we think would point toward good team chemistry. We don’t even have a good definition for what chemistry is or what’s involved. Maybe what we call “chemistry” is more than just one thing. So we’re left in the uncomfortable position of not knowing what we’re looking for and not knowing how it can be measured.
Warning! Theoretically Gory Mathematical Details Ahead!
Today, we’re not going to end with a fully formed conclusion about the subject. We’re going to play a bit fast and loose with our #GoryMath and it’s all going to be very theoretical. What we measure may not even bet chemistry. It’s all very exploratory. But sometimes, it’s good to remember that not everything has to be thought out down to the third decimal place and hyper-rational to be useful.
Let’s start by making a list of things that can be measured (ideally, from publicly available data) that we can reasonably assume would be associated with chemistry. After that, we’ll talk about some ways that we could analyze those data to see if we’re really getting at something.
I’ve previously tried defining chemistry as “The answer to the question ‘Why should I bother?’” That’s rather vague, but the idea is that because baseball is played over such a long season and requires such intense concentration just to play right, both within the game and outside the lines. There’s the game itself, but there are meetings to go over pitching matchups, proper diet and conditioning, training, coaching, and occasionally having fun. There’s bound to be some temptation to not go 100 percent into every meeting or to mail in an at-bat or two.
It’s not always obvious from the data whether a player is giving it his all or if he is lollygagging out there. (Maybe Statcast data will tell us some interesting things on the subject.) But we do know that a player who has motivation to do so is more likely to give his all. I think the idea of game theory actually has something to say here on this matter. A player on the team is one of 25 in the room. He could give it his all, even when he’s tired or he can give into the fatigue and potentially let everyone down. Then again, if he looks around him and sees that none of the other 24 guys seems to care, why should he bother either?
Some players solve this by just being personally driven to be their best. Some just want a big contract when they hit free agency at the end of the year. Some people just love the uniform that they’re wearing. But for some guys, that only takes you so far. But there’s one extra source of motivation that we need to think about. Winning. There’s a school of thought that winning produces chemistry, rather than the other way around. There might be some truth to that and there’s nothing wrong with that arrow pointing in both directions. But here’s a testable hypothesis: Many BP readers will be familiar with the idea of the win curve. The idea that there’s precious little difference between winning 60 games in a season and winning 61. The reason is that in either case, you’re in last place. But as you go up the ladder, the wins start meaning more. The difference between the 87th and 88th win might be the difference between a playoff spot and missing out on October. We often critique moves that general managers make as being wise or foolish based on where they are on the win curve. Might we see similar behaviors from players? A player might look around and think to himself “Even if I went 100 percent into this, we’re only gonna win 70 games. I’ve already made my money and I’m signed for two more years. Why bother?” We might need to control for that in our equations. Players might have a win curve in how much they invest in the team and it might make perfect sense.
But what behaviors should we even be looking for? A few suggestions (and feel free to add more in the comments):
· “Suspicious” beanballs. If beanballs are a good sign that a team is sticking together, we might identify “suspicious” situations, for example, the half inning after a member of the team was hit by a pitch, and see how often a guy on the other team goes down. It’s not that all of them will have been intentional, or will have been team-building activities, but perhaps we can see if some teams seem to do that more often than others.
· Manager ejections. Supposedly, those are the manager sticking up for his players. Maybe he’s showing how much he cares. Again, an individual ejection might just be that the manager had a poor choice of vocabulary, but if there’s a little bit of signal, maybe we could pick it up.
· Players who seem to beat the grind. We’ve seen that certain managers seem to have a repeatable skill in keeping their players from falling prey to the grind over the course of a season. We could look to see how many players seem to be doing well in staying alert over the course of a season.
· Players who show improvement. We have reasonably decent tools on how to tell whether a player has significantly changed his spots from last year or even within a year. It could be that a team just has really good coaches, but maybe guys are a little more likely to apply themselves to actualizing their talent when the chemistry is good.
· The team signs free agents in the next offseason for less than the usual dollar-per-WAR value, based on some reasonable projection system. Players do talk to each other. Maybe a guy took a little less money to go to a place he heard was good.
· Analyzing how often players give positive quotes about other teammates vs. neutral ones. These data would be hard to collect, but they could be gotten. There’s the standard question of “Hey, what did you think about Smith’s performance tonight?”
· And then there’s Statcast. I’m looking forward to seeing some response time data and some maximum velocity readings for fielders. As we gather more data, we’ll get to know what a player’s general abilities are, but soon after that, we’ll be able to look at how he responds to different situations and maybe we’ll see from day to day not only certain players have a bad day, but that certain teams are performing below their averages. Things not so happy in the clubhouse that day?
Once we’ve collected some data, we can look to see whether any of these things correlate with one another, either by using simple correlation or something more complex like factor analysis. If we see that teams that have a lot of suspicious beanballs also tend to have better results against the grind and also tend to show improvement over the course of a season, that tells us something interesting. That might not be “chemistry” but it would be something. We could then identify teams that were high (or low) on these sorts of factors and see if we could find any commonalities about them, either statistically or qualitatively. That wouldn’t answer the entire chemistry question, but sometimes that’s not the point. Right now, we have almost nothing to work from. Sometimes you have to lay some groundwork.
The standard view of chemistry in Sabermetrics has been that because it’s so hard to measure, there’s not much that we can do with it. Maybe it has an effect, maybe it doesn’t. But if we can see that there are 30-run differences between good and bad managers, maybe there really is something to the whole chemistry thing and it might just be worth chasing. It seems strange that people don’t try more often. I realize that some of the data are hard to get at and they might be noisy, and that there’s not much of a road map to go with. And there. aren’t. decimal. points. But the ever popular “new frontier” never has a road map. So, consider this a challenge. Go a little off-road.
Care to come with me?