You are here
By Alex Piazza
The score is tied 4-4 in the bottom of the seventh inning.
Bases are loaded with two outs when a senior outfielder steps up to the plate.
She has struggled to hit in recent weeks, so does the coach trust the senior will break out of her slump and knock in the game-winner? Or does the coach instead send in a pinch hitter to seal the victory?
This scenario is nothing new for Carol Hutchins. The University of Michigan softball coach has had to make tough decisions like this for the last 32 seasons.
The only difference is that Hutchins today has an arsenal of research that shows how the senior outfielder performs when the bases are loaded and how she fares against left-handed pitchers. In 1985, Hutchins did not have access to this sort of complex data.
“In our sport, statistics are a big deal because we keep track of everything on the field,” said Hutchins, who is on track to become the winningest coach in NCAA softball history this season. “Analytics are a huge part of the game, especially nowadays because of the amount of information we have access to. It’s another one of those tools we use to get a better look at our opponents.”
Hutchins often goes with her gut instinct in pressure situations, but in recent years, she admits that analytics have played a greater role in the clubhouse.
And she is not alone. Analytics and big data are revolutionizing sports, as coaches and athletes often turn toward research to gain a competitive advantage.
Pick and roll
When a game is on the line, the Detroit Pistons’ head coach knows where to turn.
“(The pick and roll is) what we’re going to be in when the game’s on the line … I don’t care how good you are, you can’t take away everything,” Stan Van Gundy said during an interview when he previously coached the Orlando Magic.
So how do teams defend the pick and roll?
They can start by reviewing research from U-M faculty Jenna Wiens and her former student Avery McIntyre.
Wiens and McIntyre, along with colleagues from the Massachusetts Institute of Technology (MIT), identified more than 340,000 screens from five NBA seasons. The researchers then tabulated how each screen was defended and the outcome—an essential gold mine for basketball coaches.
“The thing that excites me most about this sort of work is that we can automatically identify discrete parts of the game, like the pick and roll, and then produce some really interesting analysis that teams throughout the league can use to gain a competitive edge,” said McIntyre, who graduated from U-M in December with bachelor’s degrees in computer science and business.
McIntyre and Wiens, U-M assistant professor of computer science and engineering, traveled to Boston this month to present their research at the MIT Sloan Sports Analytics Conference. It was a prestigious invitation considering hundreds of researchers worldwide submitted their work for consideration, yet only eight finalists were selected.
It marked the third research presentation at the Sloan conference for Wiens, who has been studying sports analytics since 2010. Her research initially focused solely on analyzing large health care datasets to help identify patterns in patient outcomes.
Then the Boston Celtics called.
“They were really interested in the techniques and methods we had developed for health care—they just wanted to utilize them in the context of basketball,” Wiens said.
"One of the holy grails in sports analytics would be predicting injuries because teams spend so much money on players who can’t play."
This was before the NBA adopted a league-wide system in which six digital cameras now record 25 frames per second of action on the court. That equates to about 800,000 data points per game.
Their initial inquiry was one that baffles many NBA teams: “Is it better to position ourselves for an offensive rebound, or should we simply get back on defense?”
Wiens and her colleagues analyzed 6,500 missed jump shots from one NBA season, and found the optimal choice was for a team to crash the offensive glass instead of getting back on defense.
“Every team nowadays is trying to find a way to capitalize on data like this,” said Wiens, who someday hopes to tap big data to help predict and prevent sports injuries. “One of the holy grails in sports analytics would be predicting injuries because teams spend so much money on players who can’t play. I think that could have an immense impact, not only at the professional level, but also at the collegiate level.”
Yay or Nay
Charles Barkley is blunt when it comes to analytics.
“I’ve always believed analytics is crap,” Barkley, an 11-time NBA All-Star, said during a 2015 airing of Inside the NBA on TNT.
Barkley is not alone, as many fans, athletes and coaches from baseball to hockey have rejected analytics and their influence on sports.
Yago Colás is not one of them. The U-M associate professor of comparative literature, who also teaches in the Residential College, has spent years studying the history of basketball, focusing recently on the emergence of basketball analytics as a cultural phenomenon.
“Basketball analytics is, without question, the most important development in the game over the last 10 to 15 years,” Colás said. “Analytics has contributed a lot to the game, and it’s exciting in itself to watch athletes and coaches try to find ways to make decisions in high-pressure situations using information that was developed by people who are not under that same intense pressure.”
Colás has played basketball since he was 4 years old and has published numerous research articles on the sport, but in recent years, he has recognized how analytics have transformed the game—especially on offense with a surge in 3-point shots.
“Basketball analytics on the whole has in effect developed and furnished the NBA and its owners with powerful tools for maximizing the value extracted from that labor force, in terms of both points per possession on the floor and profit margins in the ledger books,” he said.
Big data not only influences coaching strategies, but it also impacts how athletes are paid.
"Every step forward makes the game more interesting—in terms of both the decisions that managers and players make on the field, and also the strategies that front offices employ off the field."
Mark Pieper (’93 LSA) runs a sports agency that represents more than 70 major leaguers, including Detroit Tigers’ stars Miguel Cabrera and Justin Verlander. As CEO of Relativity Baseball, the U-M alumnus works directly with general managers throughout Major League Baseball to ensure his clients are adequately compensated for their production on the field.
“At every level of negotiations, but particularly in arbitration, the use of analytics has simply become a part of the process,” Pieper said. “On the one hand, the increased information has made it easier in that there is more data available to discuss a player’s overall value to a club. On the other hand, it can complicate matters in that both parties often disagree on the value of some of the data and also, which stats are most important in the determination of a player’s financial worth.”
When Pieper first became a sports agent in 1995, teams primarily focused on the numbers inked on the back of baseball cards: home runs, strikeouts and batting average. A lot has changed over the past 21 years.
“Every step forward makes the game more interesting—in terms of both the decisions that managers and players make on the field, and also the strategies that front offices employ off the field,” he said. “And on top of that, in the ways that media and fans can analyze and interact with, and understand the game. It has improved baseball on a lot of different levels, for everyone involved.”