The 2012 MIT Sloan Sports Conference has long been dominated by basketball, but baseball is finally achieving a presence at the annual sports analytics conference in Boston. On the first day of the conference Friday, baseball had several presentations, with the baseball analytics panel by far the most engaging.
Super agent Scott Boras, who impressed the crowd with his intelligence and communication, appeared alongside former player and current Rays special assistant Rocco Baldelli, Astros GM Jeff Luhnow, president Mark Shapiro of the Indians and sabermetrician Bill James, the founding father of the movement.
While there were many engaging topics, Boras stole the day after also appearing earlier in the opening panel of the conference and explained why analytics struggles to take hold across all baseball teams. He spoke of “languages” people have from inside the game, ranging from player to business, executive, scouting, psychology, and how many people come from different backgrounds that influence how teams are run.
Take the Indians, for instance. Boras said when he deals with the Indians, he knows they will be heavy on applied metrics. (“Analytics help take the emotion out of decisions and stick to strategy,” Shapiro would later say.) On the opposite end of the spectrum, you have the Astros, who didn’t even have a comprehensive analytical database before noted stats nerd Jeff Luhnow took over the reins following his implementation of a database in St. Louis in 2003. He said that was the reason the Cardinals won the World Series this past season.
Luhnow cited the ability to create a database in Houston as one of the more exciting aspects of his new job, as he will start with a “blank slate,” a luxury most teams do not have these days. Take the Cardinals. The world has change dramatically since 2003 with respect to technology, new statistics and new ways of culling information. It is difficult to implement those effectively in preexisting databases, which have already gone through the creative process and implementation. Might this mean the Astros will have an advantage once their database is built?
Boras has his own extensive database he says he doesn’t share in negotiations, preferring to keep the information close to his vest, allowing him to make decisions more effectively. That’s unfortunate, because one of the metrics that would be fascinating to see is how Boras values closers. It’s certainly not Wins Above Replacement, as he believes WAR doesn’t show the true value of closers due to innings-pitched limitations. Of course, Boras is also an agent, so it’s not that surprising to see him rail against a statistic that reduces a closer’s earning power.
One of Baldelli’s comments opening the panel resonated throughout the discussion. He said most players don’t even realize analytics exist, and it’s important not to discount the human element of the game. Boras said 20 to 30 percent of players are wholly focused on sustaining their careers, and nothing else. Not loyalty, not the labor agreement, not statistics. As long as analysis remains mostly confined to the front office, it simply won’t take hold among the playing and coaching personnel.
That might change eventually, however. Two other sessions highlighted this in particular: “Predicting the Next Pitch (link to a PDF file) and “SportVision’s Baseball Data Platform.”
In the first session, an MIT team looked at ways to improve predicting pitches. This has broad implications; a player’s abilty to correctly predict the next pitch would change the way pitchers attack players after a period where the hitters dominate. Currently, the only real model available is that based on a pitcher’s overall history, but the presenters introduced a machine-learning-based predictor that looks at the current count, current game state, pitcher tendencies, batter tendencies and more to predict the next pitch, while also using weighted classifiers.
Using 2008 data and applying it to the 2009 season, the pitch predictor tool correctly predicted the pitch 70 percent of the time, with a mean improvement of 18 percent. (It was also found that pitchers do not tailor their pitch selection to stadiums.) While the tool did improve the predictions of pitches over the current model, most of the improvement came from fastball counts.
The study also found that it is easier to predict the next pitch as the count moves into the batter’s favor. That’s not completely surprising, as fastballs are thrown with more frequency in these types of counts, but the model did show an improvement over the current method, which has not caught on. If you put a proven pitch-predictor tool in front of players, that could change their opinion on using such data.
Other data that could impact players and coaching was discussed in the SportsVision presentation, which ran through its four major platforms for baseball: PITCHf/x, HITf/x, COMMANDf/x and FIELDf/x. An example showed how these data could help coaches and players by showing them exactly how their pitches vary on a pitch-by-pitch basis.
Instead of asking why a pitcher’s pitch moved less than the previous pitch, they can look at exactly what was different and adjust. This technology is currently available in PITCHf/x, but it is difficult to get it into the hands of coaches. The example given during the presentation was having live information, most likely on an iPad, available for the coach to track immediate results during practice.
Unfortunately, PITCHf/x is the only publicly-available system right now, and it is unclear if the other systems will ever be publicly available. While FIELDf/x has often been called the next major advancement in the game for its broad applications, COMMANDf/x is also a very intriguing tool that measures the location of a catcher’s mitt at the time of a pitcher’s release and where the ball ended up. It would finally answer the question of which pitchers have good command and can put the ball exactly where they want it. Roy Halladay might be the king here: His fastballs hit its intended target 26 percent of the time, as compared to a league average of 10 percent.
Back in the baseball analytics panel, moderator Rob Neyer asked what was the next necessary adjustment in the game. James suggested the next step is to find a way to get data from leagues outside of the majors. That means the minors, internationally and in the amateurs. More information will allow better ability to judge players, as there is “no real concept of how different levels of competition fit together at the MLB level.”
Boras concurred, saying that all good athletes start out playing Little League before moving to other sports, and there needs to be a way to track players 8-12 years of age to catch stars early and develop them into baseball players.
Interestingly, SportVision may have a solution. The company will begin collecting amateur data starting with the 2012 season, using PITCHf/x and FIELDf/x, dramatically broadening the amount of data available. That will allow James more information to predict baseball ability as well as give teams a heads-up on which players they may want to start tracking closely.
Another potential advancement in the game could come in the fields of biomechanics and biochemistry, as Gil Blander demonstrated in “Achieving Optimal Athletic Performance through Blood Biochemistry.”
Blander posited that many athletes, instead of injecting themselves with steroids to make them better, could improve themselves through natural means via blood testing.
For example, 12 percent of athletes have an iron deficiency, which leads to decreased endurance, which affects ability over the course of a game and a season. Using a system called InsideTracker, players could monitor their blood results to find out what they may be lacking that could boost athletic performance, as well as ways to boost it. Another interesting statistic: More than half of athletes have a vitamin D deficiency, and that percentage is higher in African-Americans. Such a deficiency leads to muscle weakness, depression, slower reaction time and weight issues.
The baseball analytics panel also discussed at length an aspect of the new labor agreement that Boras strongly disagrees with: the new restrictions on spending for amateur talent through the draft and international free agency, which “penalizes success,” he said.
“The unfairness of the (spending restrictions) is that it deviates from intellect,” he added. Boras specifically cited small- and large-market teams needing the freedom to choose how to allocate their money. The Yankees, for example, are not a developmental machine because they can spend money on free agents. Boras wants smaller markets to have the same consideration in the draft and the ability to outspend larger markets in the draft if they want. Instead, all teams must now play by the same rules in amateur procurement, giving large markets an advantage overall.
Boras also cited ever-shifting talent pools as another reason why it doesn’t make sense to cap draft spending and keep it uniform every year, as drafts will vary in talent. The talent in the 2011 draft, for example, will outstrip the talent coming in 2012, but both draft pools will have the same money allocated.
James agreed with Boras, calling the restrictions heavy-handed, saying that the new rules will cause misallocation of resources. It will work well for seven to eight years, he predicted, before the bubble will burst.