At the recent Sloan Sports Analytics Conference, Santa Claus (dressed cleverly as Bob Bowman, CEO of MLB Advanced Media) delivered a new toy for us all to share—or at least the promise of a new toy, in 2015 or so.
Santa Bowman announced that MLBAM has begun a project to put revolutionary—dare I say Orwellian?—tracking technology in every major league ballpark. For now, I’ll name it the HINZO System (High Information Nerdy Zootropic Online System) after former Indians second baseman Tommy Hinzo. It needs a name, and I grew up in Cleveland in the 1980s. Take that, Bill Pecota!
HINZO is already active in Milwaukee, Minneapolis, and Queens and capable of tracking the movements of all the players on the field, including all nine fielders (and potentially all four baserunners) plus the ball. It will also provide information on swings, hangtime for balls, and all sorts of other fun numbers. No word yet on whether we’ll also get HOTDOGVENDORf/x out of this, but I’ll send them an e-mail and let you know.
In his speech revealing HINZO to the world, Bowman stated that while of course the system will provide evidence to answer questions that were unanswerable before, he hoped that the new data stream would start more debates than it ended. Already, there’s plenty of buzz about how this new data set will revolutionize fielding metrics, and it probably will. We will soon be talking more intelligently about positioning, reaction time, acceleration, sustained running speed, and everyone’s new favorite term, “route efficiency,” rather than just crude measures of “range.” For all we know, we’ve been doing it wrong all these years. Maybe it’s not a good idea to focus on finding the guy with sprinter speed to play center field when there are a bunch of other candidates who would do just as well with slightly slower wheels.
It’s easy to see how HINZO will provide answers to questions that we already have. It’s not as easy to see how it will spark entirely new questions. After all, how do you solve a problem like Maria when you’ve never met Maria? But in the way that no one expected PITCHf/x to yield the observation that we have completely misunderstood a gigantic part of a catcher’s job (and how much it can matter), there will be questions that the new MLBAM toy will generate. I promise you.
The roadblock right now is that those of us who analyze baseball, either as hobbyists or as professionals, have been working within the confines of the tools that are already out there. Retrosheet and PITCHf/x are wonderful data sources, but they have their limits. It’s easy to base all of your thinking around those limits and to have those walls become part of your thought process. In psychology, we call it functional fixedness. We not only have to come up with brilliant ideas about what the new toy can do, but we have to learn how to come up with brilliant ideas within the headspace of “We have a database that tracks the movement of everything on the field.” That takes time.
The other roadblock is the simple fact that while we now have the broad contours of what HINZO will provide, we don’t know exactly what the technical limitations will be. Will it be beset with errors? Will some parks have more reliable data than others? What exactly will be in those XML files (assuming we ever see them)? It seems like rank speculation to dream on something that isn’t really there.
But I vote that we dream a little. (On Friday’s episode of Effectively Wild, Sam Miller kindly voted for me to dream too. Little did he know that I had been scribbling on this topic already!) What could we do with our new little toy, at least with the knowledge that we have about it at this moment? Most of that knowledge comes from this video of Jason Heyward making an amazing catch. If you haven’t watched it yet, watch it once for the catch (wow!) and twice for the data that come with the play.
A couple of things in this video are worth noting. One is that we get distances traveled in tenths of a foot. Since a tenth of a foot is 1.2 inches (and they are likely rounding), this suggests to me that HINZO has inch-resolution on where players are at any minute, and presumably at a frame rate of 25 per second. In other words, if Dustin Pedroia shifts slightly to his left, HINZO will know. We also get reaction times to the tenth of a second. Also, when they show Heyward’s route efficiency, if you look closely behind him, you’ll see a straight line representing the most direct route to the ball and then another line that bends a bit representing Heyward’s actual path. Presumably that’s all in there.
So—with the appropriate hedge that some of these may turn out to be un-answerable (or impractical) with the data available once we actually see it—here are five questions that we might look forward to answering.
1. Are fielders tipping pitches?
I don’t think people fully understand the power of being able to “see” every movement on the field. It’s always a big scandal when a pitcher is “tipping” his pitches. There’s something that the pitcher does that gives away his intentions. A batter, armed with that intelligence, has an advantage. Most pitchers take special care to guard against tipping (and advanced scouts surely try to identify a heavy tipper). What about the other eight guys on the field? Fielders commonly “cheat” in one direction or another just before a pitch, because they are in on the catcher’s signs as well. In the pre-game meeting, they discuss strategies like “We’re going to try to get Smith to reach and slap a ground ball the other way.” If you know it’s coming, you cheat a bit.
It’s hard to watch eight fielders with one set of advance scout eyes and establish this sort of pattern, but it should be possible to see what a player was doing a few seconds before the pitch is even thrown. If fielders are doing it right, they should be making their subtle moves at a point where it’s too late for the batter to see and use that information to adapt. But there are eight guys out there, and all you need is one weak link who has a predictable pattern. Now we can watch all nine fielders in all of their games for all of their pitches. That’s a beautiful data set for figuring out who that weak link is. If the “tell” is visible to the naked eye (the right fielder moving five inches to the left will not be visible from the batter’s box), it’s simply a matter of going over it in pre-game meetings. Right before the pitch, watch what the third baseman is doing. If he takes a couple steps to his left…
2. Are players paying attention? Are they sad? Are they nervous?
The trained eye can tell a good deal about someone by how they move. People who suffer from depression move more slowly than average. They also have slower reaction times. People who are suffering from anxiety move around in a more agitated way. People who have problems paying attention show inconsistent response times. HINZO will offer information about how a player is moving, even during the times in the game when nothing is really happening and he’s just alone in left field with his thoughts and the drunk guy who’s taunting him.
We don’t talk about mental health issues in baseball (or other professional sports), in large part because in United States male culture, that’s a taboo topic. I have to wonder how many players are suffering through periods of depression and not telling anyone at all. Indeed, it could be a physical injury that the player doesn’t want to cop to. Men are good at withholding that information as well. This new data set makes it easier to see those patterns, and from a team’s perspective, a way to perhaps intervene to help a guy out both as a human being and as a player. Troubles paying attention and slowed reaction time can cost real runs in a game of inches. They might be related to an undiagnosed attention problem or troubles with sleep. Maybe we can’t make a diagnosis from HINZO data, but we can certainly tell a bit about a player’s mental state just from a simple behavioral observation, and maybe get a lead that saves someone’s season.
3. Who feels it in the clutch? Does it matter?
While we’re on the subject of mental state, HINZO also offers a rather interesting opportunity to look at an old favorite: clutch hitting. Yes, I know that the general consensus is either that clutch hitting doesn’t exist or that there is so much noise around it that we don’t have the ability to measure it (and that if there’s that much noise around it, it can’t be a big contributor to explaining the variance). So why bother reopening the investigation?
The problem with a lot of those old studies is that in the past, we’ve relied on leverage index as a proxy for “pressure situation” and used a large-N database query to test whether leverage effects are present. Leverage is a fine measure of the relative importance of a situation, but it’s also not intuitive. It’s not likely that players view the importance of a point in the game in line with what the leverage index says. Years of whining about closer usage shows that managers sure don’t.
It’s also not clear that players respond twice as “nervously” to a situation with a leverage index of 2 as they do to a leverage index of 1. If we look at what situations actually bring about behavior changes, we would have a better idea of what really drives nerves in most players. But beyond even analyses that start with “most players”—there’s the tyranny of the large-N database query again—we might study individual players to see what situations lead them to change their behavior. Most people fidget when they get nervous, and HINZO has the resolution to pick up fidgeting. That’s not a perfect proxy (some people freeze!) but with HINZO data, we’d have another way, maybe a better way, to ask the question. Maybe we’ll get a different answer. Once we have a better understanding of how pressure affects players, we can start to look into whether that actually affects what happens when he steps up to the plate.
4. Is it really incredibly hard?
Well, Wash says it is. When a player switches from one position to another, is there a period of time where his reaction times and route efficiency suffer? Probably. How long does it take the average conversion to fully work? It’ll take a few years of data, but we could start to answer that question, whereas before we were left with poor fielding metrics. Until we have enough data to do a full study, we could have some “report card” studies on recent converts. We could also look at utility players and see how “portable” skills are from position to position.
While we’re at it, HINZO data will no doubt include a lot of information that would be useful in evaluating a player’s learning process. Between the positioning and fielding data, baserunning data, and swing data that the system will provide, we’ll be able to see much deeper into some of the “fundamentals of the game.” Are players progressing in what they need to learn? How much do those fundamentals matter? Not only that, but with the new positioning data, we can start to pull apart whether a player’s “great defense” is the result of fantastic positioning (thanks to his manager? his catcher? his team’s brilliant front office analysts? his own amazing instincts?) or because of his amazing athletic ability.
5. Is speed or efficiency more important?
One other thing that HINZO looks like it will provide is baserunning route efficiency, which appears to be simply the shortest possible distance between where the batter was when the ball was hit and the base he ended up at divided by the actual distance that he ran. I’m guessing that “shortest distance” is based on a ruler-drawn direct bee-line to one base, followed by a full-speed 90 degree pirouette to turn toward the next base. That’s pretty much impossible, so no one will ever score a perfect 100 percent. But it’s also reasonable to suggest that some guys have better baserunning technique (like a tighter turning radius) than others.
Let’s do a little #GoryMath while we’re in the neighborhood to see how important that might actually be. Our BP prospect team recently wrote about scouting speed and provided a handy chart for figuring out whether a prospect has elite speed (an “80” on the 20-80 scale) or just average speed (“50” on the 20-80 scale, although it’s been discussed in the past that in practice, the majority of big leaguers actually rate as below-average runners). The prospect team suggests that (for a right-handed hitter) a time from home to first of 4.0 seconds is elite, while 4.3 seconds—a difference of two blinks of an eye!—is average. Let’s assume that an elite runner can go 90 feet in four seconds. That’s a rate of 22.5 feet per second. Now, let’s say that because he’s not an efficient runner, he takes a route that adds an extra two and a half feet between bases (an efficiency drop of a little less than three percent), but that his foot speed doesn’t drop. Over 90 feet, the drop in efficiency will cost him an extra tenth of a second, give or take. It might not seem like much, but that tenth of a second is the difference between a 70 and an 80. It’ll be interesting to see what the actual spread is among runners as to their baserunning efficiency. It’s entirely possible that we might find that some players add value with how well they turn left, rather than how fast they run in a straight line. You might not be able to teach speed, but you can probably teach runners to be more efficient.
No, This Will Not Ruin Baseball
The list of questions above is neither exhaustive nor final. It’s a partial list of things that popped into my head in the last week. As the first data from HINZO start to trickle in and we (hopefully) get a glimpse of what the database structure looks like, there will be those who play around with it and become inspired (and write code for it). Plenty of new questions will come up. I think Bob Bowman is right. This is not the end of baseball analysis. This could settle debates that are open, but it could also open debates that were considered settled. More importantly, it will open inquiry on questions that we didn’t even know we had.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.Subscribe now