New camera, software system could help rate defense

Who will stay and who will go now?
2009 Draft: NoMaas catches up with Damon Oppenheimer

Defense: the final frontier. For the past five or so years, baseball statisticians have been seeking a way to rate defenders. They already have measures for offensive value, with stats which reflect a player’s ability to get base hits, get on base, and hit for power. With pitching there are a rash of stats to not only measure how a pitcher did, but also to examine his independent numbers (strikeouts, walks, home runs) to help strip out how the defense helps him. Yet there is so much more that goes into defense that it’s tough to get with batted ball data. Alan Schwarz of the Times notes a new system currently being tested in San Francisco. It uses various cameras, kind of like pitch f/x, to measure the distance, trajectory, and speed of batted balls. It could be the defensive revelation we’ve been seeking.

Bob Bowman, CEO of MLBAM, said that he hopes to be measuring in San Fran by the end of this year, with the goal to get it in all 30 parks by next year. The stats, in some capacity, would be available to the public for a subscription fee — which you can be sure RAB would pony up for. As Schwarz says, “The new camera-tracking system will assess it all to the inch.” I’m pretty stoked to see this in action.

Who will stay and who will go now?
2009 Draft: NoMaas catches up with Damon Oppenheimer
  • Tank Foster

    The UZR is a very “ambitious” statistic in what it attempts to rate, and it relies in part on some pretty spongy subjective assessments by these “stringers” who watch baseball games. I wonder how many people who put faith in UZR know that some of the statistic is calculated using quasi-data collected by someone’s “own two eyes.”

    • whozat

      I think most of us here do. Which is why it’s always funny to me when someone blasts UZR for contradicting what his eyes tell him…because it’s really just an aggregation of data from what some people eyes told them and some other numbers.

      • RAB poster

        Which is why UZR in particular is my least favorite stat.

        • whozat

          Why? Aren’t you one of those people who likes “results”?

          UZR is a compilation of a set of results into one number. The results taken into account for UZR in particular don’t include stuff like scooping throws, holding runners, throwing out base-stealers, or blocking balls at the plate, which is why UZR makes no sense for catchers and isn’t great at rating first basemen.

          • RAB poster

            Buy UZR as it says here seems like a wildly objective stat. Instead of relying on UZR I’d rather rely on what I see.

            • http://www.riveraveblues.com Joseph Pawlikowski

              But you can’t see everything. There are 30 teams playing 162 games. You see the Yanks every day, and other teams three games at a time. That’s not a good environment for observation.

              Plus, observations come from your memory. Allow me to quote: “Memory can change the shape of a room; it can change the color of a car. And memories can be distorted. They’re just an interpretation, they’re not a record, and they’re irrelevant if you have the facts.”

              • RAB poster

                A stat like UZR rates Swisher as an average defender. I do not believe that.

                Anyway,you make a good point about all the different teams, but the teams I care about are the ones we play often, and then I form my own conclusions seperate from a very objective stat like UZR.

                • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

                  This “eyes” vs. “stats” convo is hilarious.

                  All stats are aggregations of eyes. All of them. Batting average is an aggregation of sets of eyes watching what happens every single time a batter has an official at bat.

                  The difference between UZR and your eyes is that your eyes don’t watch every play. UZR does.

                  More data >>>>>>>>>>>>>>>> less data

                • RAB poster

                  I watch all the plays and players I care about. I know enough to say that Swisher is not good defensively, Damon does not have a good arm (duh, I know) Gardner has a lot of range but takes ad jumps, Tex has a ton of range and is terriffic defensively..etc.etc. etc. If I need an opinion on how others are defensively I’ll use UZR, but only with a grain of salt until I see these players for myself.

                • RAB poster

                  ad is bad.

                • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

                  Okay, how about this, then:

                  You have watched Swisher all year long and think he’s a bad defender. I’ve watched him all year long and think he’s a good defender.

                  Our difference of opinion is probably based largely on the biases we carry. I may be more prone to remember his good catches and forget his bad ones because I think he’s a good player. You may do the opposite because you think he’s a bad player.

                  What UZR does is ignore any bias for or against the player and simply compare him, dispassionately, to a neutral standard – that standard being the rest of the league.

                  UZR isn’t agreeing or disagreeing with either of us. It’s just saying, compared to other rightfielders, Nick Swisher makes as many outs as everyone else. It’s taking your data and my data, removing subjective biases, piling it in with everyone else’s data, and making comparisons to baselines.

                  Unbiased large data samples >>>>>>>>>>>>> biased small data samples

                • RAB poster

                  But the problem with UZR is that it’s based on what people see with their own eyes anyway. Making itf fairly unreliable.

                • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

                  But the problem with UZR is that it’s based on what people see with their own eyes anyway. Making itf fairly unreliable.

                  [ facepalm ]

                  No, it’s not. That’s what I just outlined for you.

                  What we see with our own eyes is unreliable for three reasons.

                  1) We don’t see everything, because we can’t be everywhere watching every play at the same time.
                  2) We can’t remember everything we see. We tend to overemphasize the very good and very bad because they’re more memorable and forget the vast majority in the middle because it’s mundane.
                  3) We are colored by our bias and we overemphasize evidence that supports our preconceived notion and we ignore or minimize evidence that contradicts our preconceived notion.

                  UZR, like all other stats, eliminates those three flaws from our anecdotal visual evidence by:
                  1) Seeing every single play at once
                  2) Recording the results of every single play in a perfect, permanent memory bank
                  3) Recording only what did and did not happen, without favoritism for or against any party involved.

                  UZR, like every other stat, is human visual evidence at it’s best without the flaws and shortcomings of human visual evidence.

                  UZR is like thousands of sets of RAB poster’s eyes all doing exactly what you do when you watch and evaluate defense, only without your imperfections and fuzzy memories and mental reservations.

                • ChrisS

                  UZR is like thousands of sets of RAB poster’s eyes all doing exactly what you do when you watch and evaluate defense, only without your imperfections and fuzzy memories and mental reservations.

                  While also watching every other OFer in the league.

  • Tank Foster

    Forgot to add…at the end of that post, that anything more precise, like this system, will be a huge improvement in calculating defensive ability.

    The problem of assessing where the fielder was positioned at the beginning of the play also needs to be conquered, as well as, for middle infielders, where they might have been “leaning” at the start of the play, in plays where runners on base are moving with the pitch.

    • http://www.riveraveblues.com Joseph Pawlikowski

      Hopefully this system can pick up defenders and where they’re positioned.

    • Chris

      It doesn’t really matter where the defender is positioned, unless there is a severe shift in play. Part of defense is positioning yourself correctly. It’s similar to the outfielder that takes bad routes to the ball but is fast enough to make up for it and make the play.

      The question of shifts and when runners are going is slightly different, but that is a relatively small percentage of overall plays (and the differences between players is even smaller).

      • Tank Foster

        “It doesn’t really matter where the defender is positioned, unless there is a severe shift in play. Part of defense is positioning yourself correctly. It’s similar to the outfielder that takes bad routes to the ball but is fast enough to make up for it and make the play.”

        I guess you’re right in a way. But I’d want to know which guys make plays because of speed and which because of positioning. I’d think you can improve your positioning on batters from scouting reports, but improving your quickness or raw speed getting to a ball is much harder to change.

  • A.D.

    Better defensive metrics would be excellent, interesting to know their exact algorithm would be for determining overall defense.

  • A.D.

    Reading the article, its even more than a better UZR, but will include baserunning & throwing arm.

  • ledavidisrael

    LoveMyLife

    We are so lucky to be alive right now..

    Once everyone has this. The next mountain to climb is going to be the medical side of the game!!! Healthiest teams win the most!

    • http://www.riveraveblues.com Benjamin Kabak

      Actually, The Times wrote about the medical innovations earlier this week. Michael Schmidt wrote about the pace of injuries and also about teams’ efforts at predicting injuries. Interesting stuff.

  • ledavidisrael

    The MPH feature makes me feel all warm inside..

  • C

    There’s a great article about this at http://www.hardballtimes.com/main/article/evaluating-defense-using-hit-f-x/ Really interesting stuff.

  • A.D.

    …but they still rate fielders by errors and fielding percentage, which are about as computationally sophisticated as a horse clomping its hoof.

    I just enjoyed this comparison

    • RAB poster

      I don’t get why people are just starting to totally disregard errors and fielding percentage.

      I understand that there are other stats, but doesn’t make more sense to have, say, A-Rod at third, even with the hip injruy, than Ransom who’s made like, four errors already?

      • whozat

        because the other stats take errors into account as well. It’s just that fielding percentage is such a very crude tool, so it makes more sense to use advanced metrics because they’re more likely to give you a more accurate picture of what’s going on. Not necessarily needed in the Ransom case, but not all decisions are as clear cut as that.

        • RAB poster

          Do they really? That I did not know.

          • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

            It’s interesting that you seem very anti-UZR without really knowing what UZR does and does not measure.

            • RAB poster

              Whozat mentioned “the stats”, not UZR.

              • Jack

                UZR is one of “the stats”.

      • Ed

        I don’t get why people are just starting to totally disregard errors and fielding percentage.

        If you have a lot of errors or a low fielding percentage, you’re almost certainly a bad fielder.

        But having a high fielding percentage doesn’t mean you’re a good fielder. It just means you don’t screw up when you manage to get the glove on the ball. Jason Giambi has a career .992 fielding percentage. Sounds great, but if you’ve ever watched him play you’ll realize that he only fields things that are hit within about 2 feet of him.

        I understand that there are other stats, but doesn’t make more sense to have, say, A-Rod at third, even with the hip injruy, than Ransom who’s made like, four errors already?

        Playing Ransom over A-Rod is a necessary evil, not something people want to do. A-Rod isn’t at 100% and his doctors don’t want him on the field every day. Pushing his hip too far at this stage could cause further injury.

  • A.D.

    One step closer in the hope of just playing the games on spreadsheets, as it was intended.

    • ChrisS

      How true.

      In fact, it should just go back to the good ol’ days when they didn’t even have a box score. Just a modified transcript of John Sterling declaring every pop-up and flyball, “it his high, it is far it is … caught at the warning track.”

      Every player could be above average.

    • gc

      Can’t you just smell the freshly dried ink on the page? The warm glow of the computer screen as it washes against your face? The clitter clatter of the keyboard? The aftertaste of the 64 ounce Super Big Gulp of Mountain Dew at your side?

      Ahhh, America’s Pastime. What a great game, baseball. :)

      • Tank Foster

        Funny. In a way, this is capitalism at work.

        Back when baseball talent evaluators were paying attention mostly to those things they could directly observe on the green grass, it was easier to get an advantage over another team if you were savvy with statistics. Today, every teams knows what works (with offense and pitching, anyway), so this is just teams trying to get another edge, finding bargains in defense.

        Trouble is, every team will have it, and the playing field will be leveled.

        To keep baseball interesting, they should do something with the rules such that the modern way of evaluating offensive value is no longer valid.

        Consider, in the extreme, if you could no longer get a walk. The value of players would be turned upside down, and slap/contact hitting guys with high batting average would immediately become far more valuable than they are now.

        The original intent of baseball was for batters to hit the ball and put it in play. They didn’t even have a base-on-balls rule for many years. Through much of the 20th century, the value of the walk was under appreciated, and some baseball people even viewed batters with lots of walks as “lazy.” Today, players go for walks so much that it’s almost absurd.

        The best thing that could happen to baseball would be some sort of change that markedly decreased walks, but did not change the overall amount of offense too much. With modern pitching restrictions, the compulsion to get walks and works pitchers has created a slow, tedious game where power hitters are in control and teams can’t get enough decent pitchers to get through a game with less than 4 pitchers.

        • RAB poster

          I like baseball how it is now. A great game. No need to change it.

          • Tank Foster

            Everybody seems to want the game they grew up with. That’s fine. It’s not my place to tell anyone what they should want. The game I grew up with, in the 1970s, was quite different than today. These new stats have opened my eyes to how the game works, and I think the game is way, way off base now, no pun intended. They’ll never change it significantly, though. Baseball people are way too stodgy, and they’re probably scared to mess with the game. I just envision a game with lots of offense, but with guys actually hitting the ball and running the bases alot. Instead of walking and waiting for 3 run homers. All sorts of athleticism would return to the game if there were more hitters, a few less homers, and more going on on the bases.

            • RAB poster

              Fair enough.

            • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

              You know, batters took walks back in the 1960’s as well.

              • ChrisS

                Ken Singleton said that back then he wasn’t up there looking for walks, he was just looking for a good pitch to hit. Sometimes the pitchers never threw him a good pitch to hit, so he walked quite a bit [like Nick Swisher, Mark Teixeira, and A-Rod].

                I like Kenny, but damn that statement made my head spin.

        • RAB poster

          Of course, the truly great thing about baseball is that you can analyze these ridiculous stats all you want and something will still happen that’ll surprise you.

  • gxpanos

    It frankly strengthens my faith in Man that people can make a living doing this. So so so so great that even a game is not above objective evaluation or disciplined study. If we’re going to waste time experiencing the dramatic ups and downs of a season, a game, an at-bat, so that we can feel the ecstasy of a WS, we might as well watch and understand baseball in the most comprehensive and meaningful way possible.

    If this works, I’m subscribing.

    • gc

      Comprehensive? You bet. Meaningful? Not necessarily.

      • gxpanos

        Hm, that’s interesting. Why do you say that? Just because a lot of the stats (notably UZR) are shoddy?

        • gc

          Because, as someone pointed out above, you can analyze and analyze and analyze. You can discover probable trends and figure out possible outcomes. And all of that is great. But in the end, the game has to be played in real time by real people, and we’ve been shown again and again through the years that amazing things happen that contradict the numbers, trends, and analyses. So while figuring all this stuff out can be, and will likely be valuable, it doesn’t always translate into steadfast “meaningfulness,” if that makes any sense. No stat really does. The unpredictability, in the end, is one of the reasons I love the game so much. Just when you think you’ve seen it all…

          • RAB poster

            In the end, all the stats and trends in the world are useless if Luis Castillo drops the ball.

            • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

              No, they’re not.

              • RAB poster

                What stat in the world-what stat, tell me?-predicted Luis Castiillo dropping the ball.

                Anybody who knew and loved stats would have said that the Mets had that game right up until the ball was an inch from Castillo’s glove.

                • RAB poster

                  For that matter, every baeball fan in the COUNTRY said that.

                • http://www.riveraveblues.com Joseph Pawlikowski

                  This is a strawman argument. No one has argued that stats can predict the outcomes of individual plays. You’re better than this, RAB poster. That kind of thing is what John Sterling would say.

          • Tank Foster

            gc–I think I know what you mean, but I’m not sure the way you’re explaining it makes sense. Your statement “we’ve been shown again and again through the years that amazing things happen that contradict the numbers, trends, and analyses” doesn’t really make sense to me. In some cases, I think you can say that, at least looking over the last 30 years, we’ve learned that the old methods of analysis were faulty. It wasn’t that baseball was “unpredictable,” it was that people didn’t know what to look at in order to make accurate predictions.

            Today, baseball men have a much better handle on how to predict things. No, you can’t predict individual games, because of variance. But people know better than ever how to assess which are the good teams, and good players, than ever before. That, to me, makes things boring…there are few surprises.

            • RAB poster

              You’re right, very boring.

              But sometimes the stats and trends are defied, which makes the games worth watching.

            • gc

              It basically comes down to this for me. Stats are useful, but they don’t always tell the complete story, nor are they always reliable. And let me make this clear: I don’t and never have expected them to be. In the end, all the stats may say that this one player should perform a certain way in a certain situation, and then when it comes down to that situation, that same player can do something which the stats say he shouldn’t be able to do. It’s no big deal. I’m not anti-stats. I use them as much as the next guy. I just take them with a tiny bit of a grain of salt, that’s all. So for me…useful, YES. Meaningful? Most times, but not always. Others are free to feel differently. :)

              • gc

                Hate to reply to my own comment, but I guess I should clarify that a lot of what I’m trying to say has to do with when people try to use stats as predictors of what will happen. To analyze what has already happened can be extremely useful, and that is not where my (pseudo) beef lies.

              • RAB poster

                I feel exactly as you do.

              • Chris

                Stats predicted that the Rays would be a competitive team in 2008, largely because of improved defense. No subjective viewer came up with a similar result.

                • RAB poster

                  The Rays are competitive.

  • RAB poster

    These new stats are pissing me off.

    Why? No idea. Just are.

    • whozat

      we could tell. I’m not sure why you object to understanding things better.

      • RAB poster

        Good question.

        I’m anti learning.

      • RAB poster

        In all seriousness, the reason I dislike these stats is that sometimes they contradict results, which I don’t get.

        For example, things like what Steve Goldman (yes I use him a lot because he’s a major example of a stat guy I know of) say things like “Wang’s strikeout to walk ratio don’t line up with long term sucess”.

        Granted, he’s not being sucessful right now but that’s because of injury. It seemed to me that Wang as a sinkerballer would have a low BABIP and not many strikeouts, and that would be fine since he’s a sinkerballer. No reason not to be sucessful long term.

        So sometimes the stats annoy me with things like “Cano’s BABIP with men on base is very low, that’ll change.

        Well, it may change, but I doubt significantly. Forget stats. Look at results. Cano has trouble hitting w/men in scoring position.

        That’s my take on it. I really don’t think there’s such a thing as a “foolproof” stat.

        • Tank Foster

          I think you’re confused about the meaning of stats. Stats are numbers which described what happened in the past. They can in some cases be used to predict future trends, but they don’t predict isolated events. You’re “begging the question” by saying you can forget the stats and conclude Cano can’t hit with RISP. The stats don’t say he can or can’t, they say he hasn’t. Knowing what is known about BABIP allows you to predict he will do better at it in the future than he has in the past. That’s all it says.

          • Tank Foster

            Wasn’t trying to be snarky or mean; shouldn’t have said “confused.” Just wanted to clarify my understanding of stats, and their use.

            • RAB poster

              No worries.

          • RAB poster

            I know, I do not think Cano will do much better at it in the future. He has bad at bats w/RISP. Some people just have trouble in those situations that a stat like BABIP doesn’t answer. Cano seems to me like one of those people.

        • whozat

          There is no such thing as a foolproof stat, you’re absolutely right. So?

          Just because something doesn’t provide all the answers doesn’t mean that it’s useless.

          “Well, it may change, but I doubt significantly. Forget stats. Look at results. Cano has trouble hitting w/men in scoring position.”

          Yes, but WHY? Stats help us try to understand what’s happening beneath the results, to see if there’s an endemic problem or if it’s just happenstance. You look at LD% and BABiP because a component of the problem may be that a guy is hitting into a lot of bad luck. That WILL change, and potentially significantly. Pitcher’s BABiP helps identify fool’s gold, guys who are a product of their defense and luck, like Brian Bannister a few years back.

          Looking beyond results allows one to know why things are the way they are. Doing that helps you predict how guys are going to play, so you can value players by what they WILL do, not what they HAVE DONE, and that helps you build better baseball teams. You can get Nick Swisher for Wilson Betemit.

          • RAB poster

            I don’t think Cano (for example) is hitting into a lot of bad luck. That’s the point. He’s having bad at bats IMO.

            And sometimes the people are fool’s gold, but sometimes hey are just “stat defiers”, like (hopefully when he gets back to normal) Wang.

            • RAB poster

              hey=they

            • V

              The thing is, though, that more than a handful of times, he’s come up in a big situation and SMOKED the ball RIGHT to the SS, for an EASY double play.

              5 feet to the left, it’s a 2 run single.

              That’s luck.

              Cano doesn’t make weak contact. He makes HARD contact, almost every time he’s up there. Thus, if he hits it to the shortstop or second baseman, it’s easier to turn a DP than if he had dribbled the ball weakly. He can’t control if it’s at someone or not.

              Yes, his plate approach sucks, and he should be hitting line drives or even fly balls instead of grounders, but he has also had considerable bad luck when he’s gotten a hold of one.

              • RAB poster

                He hits it on the ground in those situations far too often, his approach is bad, and yes some bad luck.

                But I can’t see his RISP avg. improving all that much. Say from 200 (right now) to 240, that’s a HUGE jump and still pretty bad.

        • http://twitter.com/tsjc68 tommiesmithjohncarlos a/k/a Ridiculous Upside

          In all seriousness, the reason I dislike these stats is that sometimes they contradict results, which I don’t get.

          No, they don’t. What they do is contradict what you anticipate the result is going to be.

          It’s not that stats don’t correspond with reality. It’s that they don’t correspond with what your own personal bias perceives reality to be.

          • RAB poster

            Like the guy who said Wang will have troubble with long term sucess because of a low BABIP and strikeout rate.

            But Wang is a sinkerballer. Of course those things will be low.

          • Chris

            I think it’s more that stats evaluate the aggregate of what happens, which evens out (and can’t predict) the single or improbably exceptional events. The beauty of baseball is that things generally average out over the season because there are so many games.

            That being said, what bugs me is when people complain about advanced statistics and then quote less useful stats as part of their argument that stats are bad. If stats are so bad, then why are you using them?

  • Mattingly’s Love Child

    Wow! Another step closer to playing the games on spreadsheets! Nerds of the world rejoice!

    /sarcasm

    That would be truly awesome if they could set this system up for next year. Having a truly accurate way to measure fielding would be great. To me, something like this could end the debate between the old school Joe Morgan types and the new school analytic types (bloggers in their basements) when it comes to defense. This system would be evaluating what the players are doing on the field, hard for even a seasoned a-hole like Morgan and Sutcliffe to argue with that….

    • Mattingly’s Love Child

      Wow, I missed AD’s comment that was exactly the same as part of mine….

      -10 for me

  • Glen L

    Please, everyone, bear in mind – stats are simply facts, which have already occurred, reduced to numbers

    • Chris

      As opposed to what I see with my eyes, which are simply observations, which have already occurred, modified by my pre-conceived notions and then haphazardly stored in my own fallible memory.

      • RAB poster

        Gesundheit.

      • The Fallen Phoenix

        +1.

  • http://bronxbaseballdaily.com Matt ACTY/BBD

    I like it. Let’s hope it works out well.

  • http://www.riveraveblues.com Benjamin Kabak

    You weren’t the first or only person to send us this article. Stop yer whinin’.

  • http://www.riveraveblues.com Joseph Pawlikowski

    Fuck off. My friend sent it to me this morning.

  • EDUB

    rrrrright. i’m sure that tip box blows up a thousand times a day

    /eyeroll

  • gxpanos

    Wow cool.

  • EDUB

    comical

  • Glen L

    EDUB – you are the GREATEST!!!!!!!!! thank you so so so much for being the only person ever to email the site admins! seriously, without your deep-digging research into some underground rag no one has ever heard of (the NY Times? .. did i spell that right??) we would never, ever, have come across this article

    from the bottom of all our hearts – a huge RAB thanks and ever-binding debt of gratitude .. keep fighting the good fight!