Recap of the SportVision Pitch F/x Summit

Research often inspires more questions than it answers. That’s acceptable because asking the right questions is such an important part of doing the right research. Every presentation at SportVision’s 2011 Pitch F/x Summit either asked or answered a worthy question, making the summit a great way to spend an day talking about baseball.

That said, of course a couple presentations stepped to the fore.

In terms of pure surprise factor, there might not have been more surprising nuggets of information than those Alan Nathan included in his presentation about the College Softball World Series. For one, his calculations suggest that an elite fast-pitch softball pitch appears as fast to the hitter as a 90 MPH major league fastball. Because of the shorter distance from the mound to the plate, the pitches have similar hang times. Other factors make batting difficult at the highest levels of women’s softball. For one, the lower release point means that pitches are often actually rising. Also, some of the best pitchers in softball have pitches that look identical until about halfway to the plate. Respect level heightened. Maybe there’s even a chance that this sort of research can help major league submariners perfect their craft?

Max Marchi’s presentation about quantifying catcher defense as a whole was an inspired piece of research. By identifying the main components of catcher defense and then using Pitch F/x to lead the analysis, he came up with results that were both surprising and affirming. The main components were blocking pitches, framing pitches, controlling the run game, and fielding balls in play, and each featured about a six-to-seven run observed swing in each direction. That the results featured Russell Martin‘s 2008 as one of the best defensive seasons in the Pitch F/x period (+27.4 runs), and Ryan Doumit‘s 2008 as one of the worst (-32.6 runs) served as an affirmation of what we’ve seen with our own eyes. Still, these would represent three-win changes to these catchers’ values as calculated to date. Are we comfortable with their defense being worth that much? How much more refining must be done if the highest year-to-year correlation on any of the components was .68 (for framing)?

Perhaps the most interesting catcher-related connections could be made between Marchi’s work and Graham Goldbeck’s summit-opening presentation. Goldbeck, in trying to analyze the axiom that pitchers can take something off in order to improve their control, worked with a concept called Command F/x. By identifying the catcher’s mitt as a target, and then finding distance of the final location of a pitch from that target, he came up with a ‘command’ number. This allowed him to compare the command on pitches of different speeds from the same pitcher. His results were inconclusive, but with the critique of the audience both live and on-line, it seems that Goldbeck could refine his study for more impressive results. Could a more refined Command F/x number be as useful for catcher evaluation as it is for evaluating pitchers?

When discussing the question of blocking and framing pitches, Max Marchi admitted that using more of Goldbeck’s ‘Command F/x’ type of analysis would help him lock down some of his own results for catcher defense. In other words, he felt that there was some use to using the catcher’s mitt to help analyze the ability of the pitcher to hit a target. One of his own findings was that the distance between the location of a pitch and the catcher’s set target was smaller for strikes than it was for balls. That seems to suggest that there is something to this method, even if it might be uncomfortable for some to use the catcher’s mitt as the pitcher’s target. But how do we deal with the individuality of catcher approaches with respect to target-setting?

The simplest presentation of the day offered an insightful new way to evaluate pitcher command from an entirely new angle. Instead of dealing with the catchers’ glove, Matt Lentzner created three zones to replace the strike/ball dichotomy of the past. These three zones are completely intuitive: True strikes, or the 20% of pitches that are ‘down the middle’ and are batters’ pitches; True balls, or the 30% of pitches that are definitely balls and are pitchers’ pitches; and Borderline pitches, which are comprised of the 50% of all pitches that ended up in a zone that could be called either way. By changing two zones into these three zones, Lentzner created a new way of evaluating both batters and pitchers. A sample finding from this analysis was that a contact-heavy player like Jason Bartlett had virtually no difference on his slugging percentage on balls in each zone (87, 86 and 85 SLGcon+ in each of the three zones), while sluggers like Adam Dunn showed very different splits based on where the pitch was (160 SLGcon+ on True Strikes, 128 on Borderline and 131 on True Balls). This line of reasoning has many implications and we are only now scratching the surface. For example, what’s the ideal hitting approach given these zones?

There was at least one time when the questions inspired by the presenter were slightly melancholy. When Alan Nathan showed us his best fits for batted ball trajectories, viewers couldn’t help but wonder about the black box that is Hit F/x. Later, when Greg Rybarcyk updated his excellent work on true defensive range, that same question popped up. What about Field F/x, too? Rybarczyk had access to the time stamp at the beginning of a play, flight time of the ball, impact point, world-x and world-y positions of the defenders at the beginning and end of the play, and the time stamp at the end of a play. How many more analysts will have access to similar information in the future? If SportVision wants to continually promote excellent work using their tools, they might want to find a way to give the public some of this information in an easily accessible and processable format.

The Korean contingent and their presentation on the laudable and difficult process of bringing Pitch F/x to a league with some parks of questionable quality was also excellent. In building from the ground up, they are aiming to link Pitch F/x data with video of each pitch, which is a lofty aim that we could do well to mimic ourselves (beyond, of course, the Bloomberg Sports pro tool, which does so now).

Other presentations and conversations were notable and enjoyable and cannot all be reproduced here. The post-summit baseball game may also have inspired some of the presentations of tomorrow. Overall, the SportVision Pitch F/x Summit is an event that does much to further the quality of baseball research around the world.




Print This Post



Graphs: Baseball, Roto, Beer, brats (OK, no graphs for that...yet), repeat. Follow him on Twitter @enosarris.


16 Responses to “Recap of the SportVision Pitch F/x Summit”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. WiersNRAF says:

    Without looking, there’s what, 9 pro Korean teams? How many have Pitch F/x already installed, and is that data available to all of us? Any word on if the Japanese leagues are installing it too? I would LOVE some more info on Yu Darvish

    Vote -1 Vote +1

    • Mike Fast says:

      Four stadiums had installations, and no, the Sports2i presenter said that they were still working on their back-end database infrastructure and did not have the data available online at this time.

      Vote -1 Vote +1

  2. AK707 says:

    Re: Catchers – “Are we comfortable with their defense being worth that much?”

    Why not? They play a part in every pitch during the game. If we are willing to give starters the benefit of the doubt (ie, they can rack up value just as well as a position player considering how many batters they face per individual game, even if they only play 1/5 of the time: via Eric Seidman “Consider this: Verlander has already faced 830 batters, while a league-leading hitter might step to the plate 750 times. The disconnect deals with the cross-section of time, perception and impact.”), why not catcher defense? They should be involved with many, many more plays than a average fielder, and have more opportunities to accrue value. We can’t see this with traditional metrics, and F/x tech could help with this.

    Vote -1 Vote +1

  3. Mike Fast says:

    It’s important to observe, too, that the +/- 30 run seasons are the very maximum performance observations over the course of three years. The standard deviation of observed performance is more like 10 runs, and the standard deviation of the actual talent would be less than that, even if one assumes that Max’s method did a perfect job of measuring what he was trying to measure.

    I would refer people to Sean Smith’s article in the 2011 Hardball Times Annual where he used a WOWY method with pitcher-catcher pairs to evaluate catcher defense. I don’t have it in front of me at the moment, so I don’t remember the exact values he found, but the range was not that different than what Max found.

    Vote -1 Vote +1

  4. Choo says:

    Have any results of TDR been published for our viewing pleasure?

    Vote -1 Vote +1

  5. Bill says:

    Softball v Baseball

    No, 60 mph (at 40 feet) is NOT harder than 90 mph (at 60 feet).

    Yes, it gets to you just as quick. But what EVERY SINGLE one of these studies FAILS to recognize, you don’t have to just have quick reaction times, you have to meet the object at precisely the correct time and space. The ball traveling 90 mph is SO much harder to meet.

    I can easily go into the cage with the machine throwing 60 mph and hit from 30 feet away. I’d have no chance of hitting a 120 mph ball from 60 feet, even though they are taking the same amount of time to reach the batter.

    Vote -1 Vote +1

  6. Bill says:

    You hit what you see. A 70 mph pitch looks big. A 96 mph pitch looks like a freaking peanut.

    Vote -1 Vote +1

    • DD says:

      Not to mention, the ball is smaller in MLB and they use wooden bats…

      Vote -1 Vote +1

      • Eno Sarris says:

        I believe both Alan Nathan and I phrased this correctly: the ball appears as quickly as a 90 MPH fastball. That is all. It’s correct that the ball is bigger and that the bats are metal. But the hang time is the same, so reaction times have to be similar.

        Also, you should see the diagram he had of one of their pitch flight paths. Might impressive. Like the pitcher could have thrown both piches through a hole halfway to home plate, and one would end up a foot lower than the other at the plate.

        Vote -1 Vote +1

  7. They may arrive after the same delay, but the baseball is not only smaller, but coming from farther away, and thus presents a much smaller visual target to resolve and begin tracking.

    Taking this point to an extreme, a bullet fired from about 1,000 feet away would arrive in 0.4 seconds or so also; clearly we wouldn’t expect a hitter to be able to have a similar reaction time to the bullet as to the pitched softball.

    Vote -1 Vote +1

  8. Me, too! Given the relatively short agenda, I don’t know why they didn’t put some more break time in the middle, or perhaps even do a round of introductions. No doubt there were a lot of people I missed talking to – I may never know exactly who!

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>