"Mathe-manic-al" Students Analyze Stats to Aid Their Basketball 'Cats
Here's a warning to all of Davidson's men's basketball opponents this year: The Wildcats' game is fueled not only by hard work and excellent coaching, but also by the power of math!
A group of basketball-loving math majors is gathering and analyzing reams of micro-level game and player statistics to discover information that might give the team an edge on the court. The students, who call their enterprise "CatsStats," meet weekly with assistant coaches Matt McKillop and Will Riegel to review their findings, and also tweet in-game analysis to @CatsStats.
They're part of a broad movement to apply mathematical analysis to improve sports performance. The field has grown rapidly over the past few years, and spawned its own professional association called APBRmetrics (Association for Professional Basketball Research Metrics). Professional basketball teams began hiring statistics consultants in 2004. Davidson's own former varsity standout Jason Richards '08, a coach at the University of Pittsburgh, now serves as that team's coordinator of video and analytics.
Professor Specializes in Sports Analytics
At Davidson the practice has been promoted by Associate Professor of Mathematics Tim Chartier. He has taught several independent study projects in the field, and he co-organized last spring the inaugural meeting of Carolina Sports Analytics.
Chartier has mentored several individual students on sports analytics projects. One is Seth Kindig '14, who worked last summer with Chartier to develop a basketball team ranking system. They tested it out on the results of last March's 64-team NCAA Tournament, and their system picked the winning team in the second round of the tournament 75 percent of the time. Not bad considering there are more than nine quintillion possible outcomes of a 64-team bracket!
Chartier said, "Statistics can help a team prepare for an opponent. It's essentially the Moneyball system of basketball."
CatsStats operates at a level of statistics far deeper than what's found in newspaper box scores. By today's analytical standards, statistics such as rebounds, fouls, and shots made and missed, are rudimentary. The 'Cats employ information on esoteric statistics like a player's chances of hitting a shot when guarded or unguarded, the scoring percentage of certain lineups, and points scored off of turnovers, timeouts and rebounds. Armed with spreadsheets, shot charts and Web resources like KenPom.com, the students review the outcome of recent games and report them to the coaching staff, along with suggestions for improvement.
Video Service Tracks Performance
There's also a great deal of information now available from a subscription video software program called "Synergy." It categorizes a vast array of information on individual players, and can show all instances of a player's performance in every game throughout the year. For instance, subscribers can dial up and view all instances of a player scoring off of a pick and roll, a spot up shot, post-up move or isolation drive.
The CatsStats team spends hours during and after games gathering and categorizing their statistics. Miles Abbett '14 charts each shot taken during the game using a computer program written by Chartier, and reports the stats of opposing team players. Ford Higgins '14 also reports on upcoming opponents. Kindig calculates the efficiency of the various permutations of Davidson players on the floor.
The CatsStats team focuses on data pertaining to "The Four Factors" that have been historically and statistically proven to correlate most closely to winning percentage: 1) Effective field goal percentage, with extra weight for three-pointers; 2) Offensive rebounding percentage; 3) Turnover percentage; and 4) The number of free throw attempts calculated over the number of field goal attempts.
Players are evaluated by their offensive rating, which is computed by the formula "(points scored divided by individual possessions) times 100." Efficiency is computed as the number of points divided by the number of possessions. These statistical methods allow comparison of players in games occurring at very different paces.
Measuring "Pace-less" Stats
Kindig explained, "You'll notice that these are all ‘pace-less' statistics. We measure a player's and team's efficiency at scoring without considering the number of possessions. That's because some teams may play a fast-paced game and have 80 to 90 possessions per game, and another team might only average 45 to 50. We're more interested in how efficient players are with each possession rather than the number of possessions the team has."
There's also an emphasis on shooting effectiveness by location on the court and time on the shot clock. These statistics allow analysts to measure contributions not accounted for by traditional statistics, particularly at the defensive end of the court.
Assistant Coach Matt McKillop said statistical analysis is especially valuable in revealing which combination of players on the floor is most efficient. One of the current trends inside statistical analytics is computing a team's efficiency while a player is on the floor. Kindig has been providing analysis of efficiency not only for a full complement of five players, but also for subsets of three and four players as well.
"It isn't always just what one player is doing as an individual, but what he brings out in others on the floor," Kindig said. "Coaches like McKillop have become really good at the ‘eyeball test,' seeing the situation on the floor. But analytics help measure what the eyeball test might miss. It's new eyes to see what's happening."
Assistant Coach McKillop said CatsStats analysis also has been helpful in looking at positions the Wildcats play on the floor. "Our players are versatile and can play different positions," he said. "The statistics allow us to figure out what positions they play best."
He said the coaching staff is glad to have the CatsStats help. "If they can get us a couple of points per contest, it'll pay dividends," he said. "The CatsStats crew is now offering a depth of statistics unavailable before."