April 15, 2014

Baseball Therapy

Why Sabermetrics Needs Translational Research

by Russell A. Carleton

the archives are now free. All Baseball Prospectus Premium and Fantasy articles more than a year old are now free as a thank you to the entire Internet for making our work possible. Not a subscriber? Get exclusive content like this delivered hot to your inbox every weekday. Click here for more information on Baseball Prospectus subscriptions or use the buttons to the right to subscribe and get instant access to the best baseball content on the web.

Subscribe for $4.95 per month Recurring subscription - cancel anytime. Subscribe for $39.95 for one year a 33% savings over the monthly price! Purchase a $39.95 gift subscription a 33% savings over the monthly price! Already a subscriber? Click here and use the blue login bar to log in.

My father is fond of saying that a thousand “attaboys” is worth one “aw crap.” You can do a thousand things right, but if you get one wrong, all of the goodwill you built up over those thousand successes is now gone. It’s completely irrational, but no one ever said that human beings made any sense.

I’d argue that sabermetrics has a similar problem. One of the most common complaints that I hear expressed (to use a gentle term) is that while most teams proclaim that they have fully embraced the analytics movement, they still do things like bunt and put horrible hitters in the second spot in the lineup. We’re still waiting on optimal bullpen usage to become common and for managers (and general managers) to stop over-reacting to small sample sizes. In fairness, it’s not that the research has gotten things terribly wrong. There are strategic issues where, when you do the #GoryMath, it’s clear that things are being done inefficiently now, and where teams could benefit from changing things around. So… why don’t they?

I mean, we did the math.

Consider for a moment how many people out there do things that we know are horrible for them to do. People overeat, put off preventative medical care, drive too fast, and sleep too little. All of these are demonstrably bad ideas. If it were as easy as showing someone 10 different studies by 10 respected experts at 10 different medical schools, we’d have a country of health nuts. We don’t. Come to think of it, I’ve been guilty of all of the things I’ve listed—and my day job is in public health!

There’s a branch of research in public health known as “translational research.” Once we know that something is bad, how do we actually get that message out there in a way that actually makes people change their behavior? It’s easy to say that you know what’s wrong but so much harder to make it right. I think it’s about time that sabermetrics got its own translational research wing.

The Broccoli Problem
If I may frame the problem in a slightly different way: Suppose that someone asked you to do something differently in your life—perhaps increase your intake of fruits and vegetables to five servings per day, because the scientists all say that “five will keep you alive.” Right now, you maybe get two per day, but there aren’t that many fruits and vegetables that you like (he begins a series of questions with “but have you tried…”) and you don’t know a lot of recipes that contain vegetables (he hands you a cookbook).

You politely inform the veggie vendor that very few of those vegetables are available at the grocery store that you frequent (Try this great place across town!) and that you prefer to eat the foods that you grew up with (It’s your funeral, dude.) It’s not that our vegetable-loving friend was incorrect. There probably are health benefits to his plan, and on the surface it sounds really easy: just eat more fruits and vegetables. But notice the requests that are really buried in there. In order to follow the veggie guy’s plan, you’re going to have to change your grocery store, probably increase your grocery budget to buy all that produce, read a big cookbook and learn a lot of new recipes, eat foods that are unfamiliar, and potentially give up some comfort foods. Where’s the line of people to sign up?

By insisting that you must eat five servings per day and basing his entire pitch around that, he’s created demands that very few will sign up for. He’s got great research to back him up—and it’s completely useless, because few will actually follow it. Worse, when you see him next time, you’ll probably avoid him and his broccoli. What if, instead, he had said, “We’re trying to get people to increase their fruit and vegetable intake. So, we’re asking people to do one of three easy things. One is to pack a piece of fruit (pick what you like!) in your lunch or for the dreaded 3:00 pm low-energy crash. Another is to include a serving of some microwaveable frozen veggies with dinner. The third is to drink some fruit juice in the morning with breakfast. Now, he’s offering the person a choice (“pick one of these three things…”), asking for only one change, tying the behaviors to times when you are already eating, and making the ideas very easy to implement and with minimal costs.

If you take this literally and pick only one strategy, this will not get you to five servings a day (failure!), but you are more likely to increase your fruit and vegetable intake somewhat and to stick with it. Not only that, but if that one change works and you see results, there’s certainly nothing stopping you from using the other two strategies. Next time you see broccoli guy, you might say “Yeah, I started bringing an orange for the three o’clock crash. It actually works. Thanks, man!” By not basing sales pitches on dogma, you might not get all of what you want, but getting something is better than getting nothing at all.

The Third-Base Coach Problem
Let’s bring this back into baseball. One place where there is a major inefficiency that could be exploited is aggressive baserunning. I have previously argued that teams would actually be better off without third base coaches. Instead, just send everyone playground-style. Consider the following: Last week, everyone marveled at the audacity of Billy Hamilton, who tagged up from third on what was essentially a pop fly right behind second base. But let’s consider the situation. After Jay Bruce’s pop up was caught, there were two outs. Had Hamilton stayed, in order to score, he would have needed either Cardinals’ pitcher Shelby Miller to uncork a wild pitch or for on-deck hitter Todd Frazier to at least not make an out in his next plate appearance (if he had walked, then Hamilton would still have been on third—Frazier actually popped to short). Let’s generously estimate that the chances of one of those two things happening are 40 percent. If Hamilton believed he had a 50/50 shot at scoring on the arm of John Jay, even at 150 feet, then it was entirely logical that he take that chance.

In reality, the evidence suggests that this is not how most go/no go decisions are made. They’re mostly made based on whether there is a high degree of probability that the runner will make it, rather than whether the probability of making it is greater or lesser than the break-even point. Third-base coaches send the guys who are very likely to be safe so as to minimize the number of outs made on the basepaths. In doing so (as I estimated), the average team leaves about four runs on the field every season just on potential sacrifice flies.

I suppose that I could demand that third-base coaches turn from their evil ways and repent. But before I do that, I need to think about all of the other factors. The reason that third-base coaches are so shy is because of the thousand attaboys problem. If one runner gets thrown out, everyone rides him about that one and ignores all of the other runners who were safe. Yes, this is completely irrational on the part of the fans, and it’s completely irrational of the third-base coach to let such concerns drive his behavior, but that’s how decisions are really made. We have to deal in reality, not uber-logical fantasy land.

Asking the third-base coach to send everyone won’t work. He’ll take some perfectly mathematically justifiable chances, and a few runners will be thrown out. His TOOTBLAN count will mount quickly, and people will blame him for being reckless. What if, instead, a team showed the third-base coach the research suggesting that he should be more aggressive (a reasonable man should be willing to hear a reasonable argument) and said “We want you to do this, but only in this one situation.” This way, there’s not a big spike in the number of runners thrown out on his watch. This doesn’t recoup all the potential runs that are left on the table now, but it’s something that is much more likely to happen. Want that put mathematically? Something is better than nothing.

The Closer Problem
To take another example, much has been made of the fact that using a closer to get a three-out save is a misuse of resources. A closer, the argument goes, should be used at the point of highest leverage in a game, and sometimes that’s a tie game in the seventh inning. Mathematically, this is correct. But of course, there are the objections. Relievers prefer defined roles so that they know when they will enter a game. Plus, they prefer to collect saves, because saves still drive reliever salaries, and saves are available only in the ninth inning. Managers also like the ninth-inning closer role because it promotes predictability and because people respond more negatively to games lost in the ninth than in the seventh. And why completely re-organize something that’s already working decently well?

What if we asked managers (and closers) to make one small change? The first three priorities for closers at this point are “ninth inning, up by one run”; “ninth inning, up by two runs”, and “ninth inning, up by three runs.” If we swapped out “ninth inning, up by three” for “ninth inning, tie game,” this would get the best reliever on the team (or is he?) into a higher-leverage situation. This solution has the virtue of giving the closer a defined role, and it doesn’t make demands that he suddenly become a two-inning pitcher or a guy who comes into the middle of an inning. Plus, he’ll still get a lot of saves (perhaps not as many, but he can always point out that he nailed down the one- and two-run saves, so he should be okay for the three-run types). It’s not ideal closer usage, but it’s better than what is being done now, and it has a much better chance of actually happening.

The Shifting Solution
There’s a road map for teams who want to implement a “non-traditional” idea, and it’s been drawn somewhat by accident. Ten years ago, the idea of the infield shift was downright silly. There was a central line that ran from home plate to second base. There were two infielders on the left and two on the right. For the longest time (and continuing today) the biggest barrier to teams implementing a shift is a sense that “that’s not how gentlemen play baseball.” But… well, whenever David Ortiz hit a ground ball, it always went to the right side, and often into short right field. The logic behind the shift isn’t hard to explain: We know where hitters like to hit the ball, so why not put the fielders in those spots?

At first, the infield shift was an “Ortiz Shift” only and was used by a couple of teams. There were back then, as there are now, other hitters for whom a shift made sense. There may have been a strong traditional pull not to shift, but Ortiz presented such an extreme case that it was hard to ignore. But then something happened. The world didn’t end because one of the infielders was on the wrong side of second base. The only thing that happened was that more would-have-been-singles turned into 4-3 grounders.

Now, a lot of teams shift fielders around and do so for more hitters than just Big Papi. More and more teams have joined the ranks, and now shifting is just part of the game. The logical thing would have been to demand that all teams shift on an entire subset of players and do so immediately. Instead, baseball took a test case and used it to learn how to get over the guilt of asking the shortstop to take too many steps to his left. I understand that the guilt was never rational, but it was real and it needs to be respected.

If the sabermetric movement wants to make further inroads, then it’s going to need to work on the Broccoli Problem. It’s not enough to simply point out the places where people are being inefficient. You need to understand the reasons keeping that inefficiency in place and work on those. You can do all of the cool research you want, but if you don’t pay attention to how it can actually be implemented, it’s a pointless exercise.

Now go eat your broccoli.

Russell A. Carleton is an author of Baseball Prospectus. Follow @pizzacutter4
Click here to see Russell's other articles. You can contact Russell by clicking here