Saberseminar 6

[markos]

I was lucky enough to attend Saberseminar 6 at Boston University two weekend ago (Aug 13-14). While none of the presenters addressed Twins-specific issues or players, several talks did touch on topics that were applicable to the organization in general or specific players. I'll have a few more comments in the days ahead from the seminar, but to start here are my thoughts on presentations or Q&As given by current front-office members.

 

Tom Tippett - Senior Baseball Analyst for the Red Sox.

This talk was amazing, and unlike anything I've seen from a Twins front office member. His presentation was basically a retrospective of the 2011 season - the season where Red Sox finished with an epic September collapse and where they lost out on the playoffs on the last day of the season. He walked through their thought process during the prior off-season when they traded for Adrian Gonzalez and signed Carl Crawford. This led to a slide describing their internal win projections going into that season and tracked how the team was performing relative to those expectations. He spent some time talking about their strategy going into the trade deadline including a list of players available and how they actually performed.

 

Like I mentioned in the beginning, this was a talk that I've never seen from a member of the Twins front office. He was completely comfortable with the sabermetric jargon and was very open about how their projections and models were included in the decision making during the off-season and trade deadline. He said that their initial internal projections had them 3rd in the AL East (behind the Rays and Yankees) for the 2011 season despite projecting them for 95 wins, and that this deficit was one of the factors going into the Crawford signing and Gonzales trade.

 

Quick aside - he mentioned almost in passing how the game has changed a lot with the advent of the second Wildcard spot and the current parity in the league, and that it would be crazy today to be disappointed with a 95 win projection.

 

John Baker - Player Development for Cubs

Baker is a former catcher who now works in player development for the Cubs. His primary focus is to help develop mental skills for players in their minor league system. He didn't give a presentation (he was part of a Q&A session with Brian Bannister), but my takeaway is that he is working to instill the following three mental skills for the Cubs players:

1. Mental focus for each pitch. This seemed to deal with creating checklist for players to think through prior to each pitch, and then a mental step of intense focus for the 2-3 seconds just before and during each pitch.
2. Compartmentalization of baseball. Once the game is over, stop thinking about baseball. Have another mental outlet available after games to wipe the slate clean.
3. Ability to deal with failure. For almost every player, the ability (or lack there of) to handle failure is one of the most critical skills for players.

He was a little New Age-y but also Science-y as well. It was an interesting combination. And it fits with Epstein's reputation that he is willing to try new things.

 

Q&As with Dave Dombrowski (President of Baseball Operations for Red Sox) and Bryan Minniti (Assistant General Manager for Diamondbacks)

Both of these Q&As were closer to the generic baseball front-office talks that I am more familiar with. Neither was particularly focused on analytics, and for the most part the audience questions gravitated towards other topics. Both talked about how there is just so much stuff going on in the front office that the GM (or head of baseball operations) has to trust their subordinates to get the job done. Minniti worked with Neil Huntington and Mike Rizzo prior to going to Arizona, and he said the same about both of them. Dombrowski talked about how he is the face of the front office and ultimately has responsibility for the decisions, but it is imperative to hire good people and let them do their jobs. Baseball operations departments are too big these days to micro-manage.

 

Looking at each talk from the perspective of a fan of an organization actively searching for a new GM, they both shed some light on the upcoming transition. Namely, that the process will be slow. Dombrowski specifically said how it has taken him almost a year to learn about everyone in the front office and the job that they do, and that he hasn't made a lot of personnel changes in the front office. I'm not expert on baseball front offices, so take this all with a grain of salt. But in addition to this seminar, I have (coincidentally) read and listened to various baseball sources discussing front office changes. And reading between the lines, it seems likely that the transformation of the Twins front office will be slower than many people expect. I've certainly adjusted my own expectations. Many new GMs are restricted from bringing colleagues with them to a new organization - essentially they have to deal with non-competes. If, for example, the Twins were to hire someone from the Cubs, it may be months or even years before the new GM is able bring in other members of the Cubs front office that he or she wants in the organization. Also, while teams are generally very good about letting front office members move for a promotion, lateral moves depend contract expiration. In the interim, I'm guessing that the Twins front office will mostly stay intact for the first 12 months.

 

One final note: I did meet two members of the Twins front office at the conference.

[big dog]

Thanks- really, really interesting report. 

[Old Twins Cap]

It took Dave Dombrowski a full year to learn everyone in the GM's front office and the job they did.

 

Heck, Pohlad was able to get that done in a couple days by reading other Club's media guides.

 

 

[markos]

Though Saberseminar didn't have any official topics, one area of emphasis this year was visual perception for batters. Three separate talks looked into this topic.

 

The first talk was by Dr. Art Shapiro from American University. He specializes in visual perception and presented research regarding the ability of people to correctly identify spin axis when exposed to a glimpse of a baseball rotating at different spin rates and direction. Not surprisingly, he showed that people had a hard time distinguishing spin direction when the spin was at a higher RPM and when the exposure time was shorter. This was strictly laboratory work tested on the general population, so the results may not be applicable to major league hitters. Another area he talked about was optical illusions. First, objects that appear to be rotating are often perceived to be moving even when they are not. Second, the perception of a rotating object is strongly impacted by contrast differences. Neither of these examples were baseball specific, but it isn't difficult to see how they could be extended to hitting. Batters often talk about how certain pitches seem to 'move' differently than what PitchF/X indicates. And it also would explain how pitch perception is impacted by lighting, background, etc.

 

The second talk was by Dr. Michael Richmond. He was presenting results from a study by a team of Japanese researchers. They created a system that connected a pair of special pair of glasses to a pitching machine. The batter would wear the glasses, which would turn opaque at differing times after the pitched had been released. They discovered that there is a certain point (approximately ~150ms before the pitch crosses the plate, if my memory is correct) at which point vision no longer matters and the quality of contact is just as good as with full sight. This passes the smell test, as the time it takes for a big league batter to go from 'loaded' to 'bat in zone' is ~120ms-130ms.

 

Piggy-backing on the previous talk, Dr. Peter Fadde talked about his system for testing and training pitch recognition. He created an video training system that tests players to identify pitch type and pitch results (ball or strike) based on showing the player only the first 150ms after pitch release. He has collected data at the Cape Cod summer league and worked with some teams at the big league and college level. He contends that his system can predict BB% and BB/K rates, diagnose specific pitch identification weaknesses and provide prospect advancement criteria. Further, he suggested that pitch recognition was a trainable skill that players can improve on, to a certain degree, with practice. Fadde specifically name-dropped Buxton as a player he would like to test.

 

Overall, I think visual perception testing/training will rapidly become standard procedure across the league for player development and draft preparation. My only hesitation for the latter is that the Players Association (or more specifically agents) may object to this kind of testing prior to draft. But if it is regularly used by big league teams to test/filter players, I would imagine there will be a demand for training programs at the high school and college level.

[bird]

Wonderful stuff, markos. Thank you!

[SarasotaBill]

Excellent work - thank you 

[ashbury]

Numbers are important, but people get the wrong idea about baseball analytics, and this conference summary is a good antidote. Folks should be thinking in terms of Systems Analysis, not simply coming up with the latest all-encompassing number to rank players by.

 

And I believe that teams that try to wing it, with only generalists on their staff, will come out behind the teams that know what specialists to hire as consultants or bring on board.

[Jack Goin]

 

Numbers are important, but people get the wrong idea about baseball analytics, and this conference summary is a good antidote. Folks should be thinking in terms of Systems Analysis, not simply coming up with the latest all-encompassing number to rank players by.

 

And I believe that teams that try to wing it, with only generalists on their staff, will come out behind the teams that know what specialists to hire as consultants or bring on board.

You are exactly right John. The goal is to have a systematic decision-making process. I guess a fair comparison would be automated trading. Our job isn't to just look at statistics. That just tells us how a player is performing or has performed. The goal is to judge all players at the same time, attempting to come up with a value.

[Otto von Ballpark]

 

You are exactly right John. The goal is to have a systematic decision-making process. I guess a fair comparison would be automated trading. Our job isn't to just look at statistics. That just tells us how a player is performing or has performed. The goal is to judge all players at the same time, attempting to come up with a value.

Doesn't he mean it really goes beyond that (judging players)?  Don't you want analysts looking for ways to improve players too?  And judging the systems you use for judging/improving players?

[Jack Goin]

 

Doesn't he mean it really goes beyond that (judging players)?  Don't you want analysts looking for ways to improve players too?  And judging the systems you use for judging/improving players?

Yes it does. But I think when most people think you need to look at the stats. When acquiring a player that is a small portion. Player development is probably one of the biggest areas of impact. I read his post to pertain to player acquisition I guess. 

[Parker Hageman]

The second talk was by Dr. Michael Richmond. He was presenting results from a study by a team of Japanese researchers. They created a system that connected a pair of special pair of glasses to a pitching machine. The batter would wear the glasses, which would turn opaque at differing times after the pitched had been released. They discovered that there is a certain point (approximately ~150ms before the pitch crosses the plate, if my memory is correct) at which point vision no longer matters and the quality of contact is just as good as with full sight. This passes the smell test, as the time it takes for a big league batter to go from 'loaded' to 'bat in zone' is ~120ms-130ms.

 

 

This was the thing that Joe Mauer was doing when he was wearing the strobe glasses earlier this year: http://twinsdaily.com/_/minnesota-twins-news/minnesota-twins/return-of-the-swing-r4667

 

[Loosey]

 

Though Saberseminar didn't have any official topics, one area of emphasis this year was visual perception for batters. Three separate talks looked into this topic.

 

The first talk was by Dr. Art Shapiro from American University. He specializes in visual perception and presented research regarding the ability of people to correctly identify spin axis when exposed to a glimpse of a baseball rotating at different spin rates and direction. Not surprisingly, he showed that people had a hard time distinguishing spin direction when the spin was at a higher RPM and when the exposure time was shorter. This was strictly laboratory work tested on the general population, so the results may not be applicable to major league hitters. Another area he talked about was optical illusions. First, objects that appear to be rotating are often perceived to be moving even when they are not. Second, the perception of a rotating object is strongly impacted by contrast differences. Neither of these examples were baseball specific, but it isn't difficult to see how they could be extended to hitting. Batters often talk about how certain pitches seem to 'move' differently than what PitchF/X indicates. And it also would explain how pitch perception is impacted by lighting, background, etc.

 

The second talk was by Dr. Michael Richmond. He was presenting results from a study by a team of Japanese researchers. They created a system that connected a pair of special pair of glasses to a pitching machine. The batter would wear the glasses, which would turn opaque at differing times after the pitched had been released. They discovered that there is a certain point (approximately ~150ms before the pitch crosses the plate, if my memory is correct) at which point vision no longer matters and the quality of contact is just as good as with full sight. This passes the smell test, as the time it takes for a big league batter to go from 'loaded' to 'bat in zone' is ~120ms-130ms.

 

Piggy-backing on the previous talk, Dr. Peter Fadde talked about his system for testing and training pitch recognition. He created an video training system that tests players to identify pitch type and pitch results (ball or strike) based on showing the player only the first 150ms after pitch release. He has collected data at the Cape Cod summer league and worked with some teams at the big league and college level. He contends that his system can predict BB% and BB/K rates, diagnose specific pitch identification weaknesses and provide prospect advancement criteria. Further, he suggested that pitch recognition was a trainable skill that players can improve on, to a certain degree, with practice. Fadde specifically name-dropped Buxton as a player he would like to test.

 

Overall, I think visual perception testing/training will rapidly become standard procedure across the league for player development and draft preparation. My only hesitation for the latter is that the Players Association (or more specifically agents) may object to this kind of testing prior to draft. But if it is regularly used by big league teams to test/filter players, I would imagine there will be a demand for training programs at the high school and college level.

Does this system by Dr. Peter Fadde test players current ability to recognize pitchers or does it also teach them or train their eyes and brain to recognize the pitch as well?

 

Even before I read your line about him wanting to test Buxton I was thinking this would be a great tool to train Buxton on if in fact his greatest weakness right now is pitch recognition.

 

Great post by the way.

[ashbury]

I read his post to pertain to player acquisition I guess. 

My response was in overall reaction to poster "markos"'s original post and his followup post in this thread - all of which was well worth reading - and not so much to any particular area of application.

 

If markos is correct that a couple of Twins folks attended the conference, I hope they found their time well spent. I wish this conference hadn't flown under my radar, as I currently live about 10 miles from the BU campus.

[Jack Goin]

If markos is correct that a couple of Twins folks attended the conference, I hope they found their time well spent. I wish this conference hadn't flown under my radar, as I currently live about 10 miles from the BU campus.

 

We had two people there this year. We have been a few times. It is an excellent conference. Well run. Very good presentations. Smaller than Sloan & SABR Analytics.

[markos]

 

Does this system by Dr. Peter Fadde test players current ability to recognize pitchers or does it also teach them or train their eyes and brain to recognize the pitch as well?

 

Even before I read your line about him wanting to test Buxton I was thinking this would be a great tool to train Buxton on if in fact his greatest weakness right now is pitch recognition.

 

Great post by the way.

My memory is a little fuzzy at this point, so I may be mis-remembering. But his presentation (videos and slides) mostly focused on the testing aspect. I'm pretty sure that he said that it could be used for training purposes, but I don't remember him presenting any results from a training program.

[markos]

Valuing catcher defense has been one of the more, um, fluid issues within the sabermetric community over the past two decades. I won't rehash the entire history (and I'm probably not able to do it justice), but I don't think it is unfair to say that the value allocated to a catcher's defense has ranged anywhere from zero (It doesn't matter at all! Why do these no-bat catchers still have jobs?) to 50+ runs per season (Why isn't someone paying Jose Molina $20M a year?!) over the years. Many baseball analysts (certainly those who presented at this conference) seem to have gained a bit of humility throughout the entire process. The folks that talked at this conference certainly didn't exude any sort of confidence that they have solved this problem. It was more of a "we like our methods, and we've got some number that we think are pretty good. They seem pretty reasonable, but hey, we are probably still missing stuff." Case in point, just last week Joe Sheehan tweeted the following:

 

 

With that background behind us, the folks at Baseball Prospectus have spent the last several seasons refining their catcher defense metrics. They are using sophisticated statistical techniques to attempt to model the interplay between pitchers, catchers, batters, base-runners and umpires in order to tease out the contribution of the catcher for the results of each pitch. These metrics currently broken down into pitch framing, throwing out base-stealers, discouraging steal attempts and blocking pitches. Inspired, in part, because teams continue to employ catchers that, by their metrics, cannot hit or field, they have started to look into try to quantify the softer skills of catching.

 

Harry Pavlidis and Jonathan Judge presented their work at quantifying "game-calling." Harry gave the first half of the presentation, and he mostly looked at how much a catcher influences the pitch selection and pitch sequence. Using similar methodology to what they used for for the above statistics, he was able to determine that pitcher-batter-catcher breakdown on both pitch selection and pitch sequences is roughly 80%-10%-10%. So the pitcher dictates 4/5ths of the pitch selection, and the batter/catcher split the rest. The rates slightly in various situations, but that was the general gist. This makes intuitive sense, as many catchers (including one at this conference) will say that they focus on a pitcher's strengths rather than attack a batter's weaknesses.

 

Jonathan Judge's half of the presentation focused on quantifying the value of a catcher's "game-calling." His methodology was rather clever. Basically, after adjusting for all of the catching influences that they are already explicitly calculating (the framing, throwing and blocking), does the mere presence of the catcher had any impact on events? And the result was that yes, the mere presence of different catchers will impact a pitcher's walk rate, strikeout rate and home run rate even after adjusting for the catcher's framing skills. The effect isn't large (basically ranged between +/- 15 runs in 2015), but it does correlate positively from year-to-year. Again, there isn't any claim for what exactly this is specifically measuring. It might be pitching selection, or maybe pitch sequencing, or maybe the confidence a pitcher has with the catcher, or maybe an ability to settle down a rattled pitcher, or maybe an ability to fix subtle mechanical issues on the fly. It is probably a little bit of all of those things.

 

I bet you are wondering why I'm taking the time to bring this up. Well, it turns out that the 4th best catcher in "game-calling" in 2015 - squeezed in between Jason Castro and Yadier Molina - was Kurt Suzuki at 10.3 runs better than average. Unfortunately, even that extra added value isn't enough to move him above replacement-level in 2015 (at least from BP's perspective). The BP folks indicated that they will be posting this research online.

[markos]

The final talk of Twins-related note came from Dr. Brian Mills from the University of Florida. He has been tracking the strike zone changes over the past year, and his presentation dove into the changes as they may (or may not) relate to the increase in scoring (and home runs).

 

One can find the slides of his presentation here. Borrowing liberally from the the slides, he showed that umpires are calling few low strikes on the edges and more high strikes.

 

 

Further, pitchers are responding by throwing more pitches inside.

 

 

If you follow the rest of his talk, he delves into the details of the change in contact locations and how much impact that will have on the current run scoring environment. The short answer is that is probably explains ~10% of the change. But sitting in the auditorium, my immediate thought after seeing these slides is that this might explain part of Dozier's success this year. Fewer outside and low pitches combined with more high and inside pitches seems like the perfect combination for Dozier. Dr. Mills indicated that the overall change is rather small - approximately 1-2 pitchers per game. The Twins have played 130 games, so we are talking only 130-260 pitches for the season. But Dozier leads the team is pitches so far, and he has seen over 11% of the total pitches. So he has possibly had somewhere between 15-30 extra inside pitches this year compared to last year. That might explain some of the extra home runs this year.

[tobi0040]

Really good stuff. 

[rukavina]

 

The final talk of Twins-related note came from Dr. Brian Mills from the University of Florida. He has been tracking the strike zone changes over the past year, and his presentation dove into the changes as they may (or may not) relate to the increase in scoring (and home runs).

 

One can find the slides of his presentation here. Borrowing liberally from the the slides, he showed that umpires are calling few low strikes on the edges and more high strikes.

 

 

 

Further, pitchers are responding by throwing more pitches inside.

 

 

 

If you follow the rest of his talk, he delves into the details of the change in contact locations and how much impact that will have on the current run scoring environment. The short answer is that is probably explains ~10% of the change. But sitting in the auditorium, my immediate thought after seeing these slides is that this might explain part of Dozier's success this year. Fewer outside and low pitches combined with more high and inside pitches seems like the perfect combination for Dozier. Dr. Mills indicated that the overall change is rather small - approximately 1-2 pitchers per game. The Twins have played 130 games, so we are talking only 130-260 pitches for the season. But Dozier leads the team is pitches so far, and he has seen over 11% of the total pitches. So he has possibly had somewhere between 15-30 extra inside pitches this year compared to last year. That might explain some of the extra home runs this year.

and possibly why gibson has had such a pedestrian year.