July 11, 2009

New Tech Will Create Stats for Fielder Arm Strength, Baserunning Speed: Fascinating piece in the New York Times on new baseball technology developed by Sportvision: A new camera and software system is under development that can record the speed and location of the ball and every player on the field, allowing the most digitized of sports to be overrun anew by new statistics such as outfielder arm strength and base-running efficiency. "It will probably become the largest single advance in baseball science since the development of the box score," predicts the Times. "here come the stat rats,” said 35-year-scout Paul Ricciarini of the Houston Astros.

posted by rcade to baseball at 02:00 PM - 12 comments

Not that all players welcome the new numbers. A few lockers down, Wells's teammate Scott Rolen whose excellent defense and base running would presumably be evidenced by the tracking system said: "I don't believe that baseball is a game that can be encapsulated that way. That's the beauty of the whole game."
Fantastic! This software will usher in a new era of refined defensive analysis, allowing us to finally discern between A) players whose slowness or poor positioning force them to make Jeterian dives and tumbles and get an undeserved reputation for being "great" fielders, and B) those whose smart play and quick feet make everything look so easy that most observers don't fully appreciate their gifts.


And I suspect that Scott Rolen fears he might very well be in that first group. :)

posted by hincandenza at 04:19 PM on July 11, 2009

Scott Rolen comes out very well in both traditional and advanced defensive metrics -- at least historically (he has declined with age and injury). To put him in the same category as Jeter is laughable. I suspect Scott Rolen is just a small-town Indiana boy who thinks that you should go out and play the game and observe the game without any fancy technology.

posted by holden at 05:00 PM on July 11, 2009

Pete Incaviglia says:

"What happens in left field stays in left field. Capiche?"

posted by beaverboard at 10:35 PM on July 11, 2009

Whats up with the hating on Jeter? Its harder for him wearing four rings in his hand. I think that should count for something :).

posted by STUNNER at 10:55 PM on July 11, 2009

I didn't say Rolen wasn't a good fielder- my two points were accidentally conflated by how I wrote that. One, that I praise these innovations because they'll show what happens before the camera reaches the guy making the play- the kind of information that won't pull McCarver's lips off Jeter's Little Intangible, but will otherwise help the baseball community increase its accuracy in evaluating defensive talent and accomplishments.

Along with that, I made the point that Rolen might be worried about these new metrics making the game complicated or possibly showing him in a less favorable light- but I'm not saying he's like Jeter, in that Jeter is wildly overrated whereas Rolen is not.

It's not different than Curt Schilling's bizarre hatred of Questec (given that he is a nerdy tech-oriented guy overall); I can only assume Schilling worried that learning the umps and their own flaws/quirks was a part of the game, and a part that gave him an edge with his veteran experience.

Semi-related: with all the data that things like PitchFx has given us, and with innovations like this on the horizon, wouldn't it behoove teams to be throwing machine learning analysis at the data to see if there are trends available? After all, pitchers and catchers are not random actors: whether they like it or not, they have patterns and trends inherent in their behavior, even below the levels of the ones they are aware of. If we have data representing every pitch, type, location, and circumstance thrown in the last few years, surely we have a dataset enough to start to "guess" what pitch a pitcher will throw next, given the count, the hitter, the situation, etc. If that accuracy was anything better than 50%- and wouldn't it be?- a set of computers in a lab could be getting real-time data of each pitch, guessing the next pitch, then being fed the real pitch and adjusting accordingly.

Maybe the culture of baseball would look down on this, but I don't think it'd be against any rules. Imagine even mediocre hitters being able to glance to the dugout and have the hitting coach signal "Slider", and know that there's a ~70%+ chance the next pitch is a slider. It would turn every hitter into Tony Gwynn. Or maybe there are teams doing this now, and they are being very mum about it; I've read I think Schilling comment that being a pitcher in Boston was unique because of the technical tools they have- some he "couldn't even talk about".

It just seems that for the cost of a cheap utility infielder, you could have an assortment of video and 3d analysis, kinesiological analysis, predictive capabilities, and even that one pitching machine that has a human-size screen capable of simulating just about any pitch along with video of said pitcher throwing it (so that hitters could warm up facing a vintage Randy Johnson slider, etc, before facing the real thing). Having those utilities should allow a club to make all their players a little bit better. And a little bit goes a long way- 1% difference in OPS on a team would yield about 3 wins more in the standings over the course of a year.

posted by hincandenza at 11:53 PM on July 11, 2009

So now you won't have sportswriters covering baseball, you'll have accountants.

posted by outonleave at 09:08 AM on July 12, 2009

Baseball has been one of the most stat obsessed sports in the world. It's one of its charms, which is why there are so many sabermetricians -- "accountants" -- following the sport (and a few running teams as well).

posted by rcade at 10:40 AM on July 12, 2009

I can only assume Schilling worried that learning the umps and their own flaws/quirks was a part of the game, and a part that gave him an edge with his veteran experience.

This is a really significant point. The value of a veteran in any industry comes from the knowledge of the system that got him/her there, and if that system changes, he/she runs the risk of becoming less valuable. This happens in all walks of life, but imagine what it must feel like in the majors, with so few jobs available (relative to a "normal" industry). No wonder the reaction to any change is so vehement.

(Not that Scott Rolen has any reason to worry, mind you. In fact, Vernon Wells might very well have more reason to be concerned.)

posted by DrJohnEvans at 04:38 PM on July 12, 2009

Maybe the culture of baseball would look down on this, but I don't think it'd be against any rules. Imagine even mediocre hitters being able to glance to the dugout and have the hitting coach signal "Slider", and know that there's a ~70%+ chance the next pitch is a slider. It would turn every hitter into Tony Gwynn.

Maybe I'm taking you too literally, but no way would this happen. You're assuming the difference between Gwynn and Joe Average is that Gwynn is better at guessing pitches. There's much more to hitting than that.

Not to mention, the majority of the time the count dictates the pitch. It's not difficult even as a fan to watch a game and guess the majority of pitches. It's the exception pitch thats difficult to predict. And if the computer says the next pitch is going to be a curveball, and you trust the computer to completely look curveball even though the count suggest fastball, you better be right because you're going to be helpless against a fastball.

Guessing pitches better would help any hitter, good or bad, but it won't make anyone gwynn,

posted by justgary at 05:48 PM on July 12, 2009

Yeah, I didn't mean it would make every hitter into Tony Gwynn- rather, that if every mediocre hitter had a Tony Gwynn in his ear during the at-bat, would he be more effective? Watching young players play, say an Ellsbury in Boston, and seeing how they don't have patience or good sense about working pitch counts, etc, would they be a little bit more effective if they were getting cues from the dugout of "Sit on the next pitch, it's likely a slider- you don't hit those well, and he likes to follow his slider with a fastball outside which you can hammer into rightfield". No, not HoF effective, but say a .260 hitter instead of a .250 hitter? Hell, as "Bull Durham" notes, the difference between .250 and .300 is one extra hit a week if you're a regular player.

I grant you that many times any decent professional can "guess" with some accuracy, but how accurate? The homerun derby shows us that even with slower pitches thrown for the intention of being hammered, the best sluggers still only "hit" .333-.400 (say, 5-7 homeruns by the time they make 10 outs), with most of the outs being weak grounders or warning-track popups. So it's not like a PitchMaster3000 that was say 90% accurate at guessing the location and type of the next pitch would lead to a lineup of Williams and Ruths. But if it made every hitter one extra base hit every other week better... that'd be a playoff bound team.

I'm not saying it'd pan out, or make a huge difference- but I can't imagine it wouldn't have *some* benefit if the guessing algorithm was sufficiently accurate. And every little bit counts over the long haul.

posted by hincandenza at 06:11 PM on July 12, 2009

I'm not saying it'd pan out, or make a huge difference- but I can't imagine it wouldn't have *some* benefit if the guessing algorithm was sufficiently accurate. And every little bit counts over the long haul.

Yeah, I agree. The question would be how much.

1. Pitchers aren't going to ignore this. They'll know the book on their pitch selection. They can change.

2. Something like 'pitcher x throws a fastball 90 percent on 2-1 counts would helps, but a fastball isn't one pitch. Location is very important. A fastball on the hands is completely different than a fastball that tails back over the plate outside.

Again, I agree, it could help, just not sure how much. If you know a Santana changeup is coming, and you're correct, great. If you guess wrong, you'll never hit his fastball. So being more certain of his next pitch makes you a better hitter, unless you're wrong, then you'll probably do worse.

posted by justgary at 10:36 PM on July 12, 2009

Perhaps- I think the idea is that great hitters do have better baseball minds than average hitters: they are better at guessing right on that Santana change and parking it. Forget the steroids, Barry Bonds got better because he got better at working counts in his favor, then destroying something on the inner half of the plate.

Yeah, when any hitter is wrong, they swing badly and miss- but then that would happen without any fancy software, which is why Santana is a two-time Cy Young winner who's only had two seasons as a starter where he finished with an ERA over 3.00.

But it's not those pitchers you beat up on: the great pitchers make everyone look bad, including great hitters. The great hitters make up for it in spades by demolishing the guys with a 5.20 ERA. I'm just suggesting that it's not about the edge cases: it's about the law of large numbers playing out, of having every hitter on your team "guess" the next pitch, and it's location, correctly for say 80% of the time. Yeah, you fail miserably 20% of the time- but if the guy in the dugout is signaling pitch and a suggestion (lay off the next pitch, likely slider, you have a strike to spare)- you could have a very high BABIP for your team's lineup. And it doesn't have to be much difference: just enough that your hitters on the whole are a little less fooled.

Yeah, the pitchers will start adjusting their games, but that would facilitate their team having their own algorithms and systems to analyze the pitchers tendencies to try to create "perfect" randomness. The trick to me is that a pitcher/catcher battery will lean on certain pitches more often than others, in certain situations (high leverage situation and counts, maybe the middle relief guy is scared to go to his curve because he's not as good with it). The thinking is that the "system" would compute the accuracy and strike % of pitches, and do so in real time, so that the hitter would have a pretty good chance of knowing "He's going to his curve on the next pitch, but he only throws it for a strike 40% of the time when the temperature is below 75 degrees", etc.

I don't know that this would work, that such a system could or would find trends. I'm just surprised that I've never heard of anyone trying. Although I guess to your point #1 above, if you did have such a system... would you tell anyone?! :)

posted by hincandenza at 12:27 AM on July 13, 2009

You're not logged in. Please log in or register.