Could a statistical model detect cheating NBA ref?
BYU professor calls for 'early-warning system'
PROVO The NBA could create an "early-warning system" to detect cheating by a referee, says a Brigham Young University professor who uses statistical models to analyze sports.
Statistics professor Shane Reese said mining data about every call made by officials during games might catch one who fixes games, as did Tim Donaghy, the former NBA referee who pleaded guilty Wednesday to betting on games he officiated.
"I think the NBA has enough information to construct an early-warning system," Reese said. "Would one have caught Donaghy? I don't know, but the granularity of the statistics the league keeps could be used to devise such a system."
The league compiles statistics about which referees made each call and when they made them, but it doesn't share the information. NBA commissioner David Stern has said the league's current system didn't indicate wrongdoing by Donaghy.
Reese, who turned down a job in the front office of the NFL's Philadelphia Eagles last year, said access to the NBA's information would allow statisticians to build a better model.
Without the league's data, several groups have attempted less-sophisticated ways to analyze Donaghy's games. One type of bet drew the most attention, the over-under. Bettors can gamble whether the total points scored in game will be over or under a line set before the game.
The president of one gambling information Web site, Pregame.com, issued a news release that showed the over bet won 57 percent of the time Donaghy refereed an NBA game in the past two years. The previous two years, the over bet won 44 percent of the time Donaghy was an official. The swing was a strong indication of foul play, the news release claimed.
Reese created a model to test the claim.
"It was a fairly naive calculation," he said, "and it seemed to me it was unlikely to be accurate."
Reese studied all NBA referees and the over-under outcomes for games last year and found that Donaghy was not an outlier.
"This kind of data probably has no hope of finding out if an NBA ref is a cheater," he said.
The NBA has not responded to e-mails and a phone call from Reese.
"My recommendation to the NBA, whether it would include me or not, is they ought to do it in the form of a grant to an academic institution," he said. "That way, they could avoid any issues about objectivity and transparency while making a statement about academic pursuit."
A grant also would cost less.
Reese and fellow BYU statistics professor Gil Fellingham did consulting work for the Eagles. They concluded team officials did not have the right data to answer the questions they were asking. The professors also determined the team could collect the right data and that it would be valuable.
Philadelphia owner Jeff Lurie, team president Joe Banner and coach Andy Reid interviewed Reese and offered him a job.
Reese said Lurie had read "Moneyball" the book that introduced statistical analysis in baseball to the mainstream and was convinced "Moneyball" needed to be implemented in the NFL and that the Eagles were positioned to be the pioneers.
Reese declined the offer, and the Eagles hired David Lewin.
"There is substantial interest in the professional sports community," Reese said. "It used to be just the Oakland A's and the Boston Red Sox, but now the Cleveland Cavaliers have someone, and the Seattle Sonics and the St. Louis Cardinals."
Several college football programs have approached Reese about consulting, but discussions are on hold until after the coming season. He approached BYU coach Bronco Mendenhall during his first year as head coach. Mendenhall expressed interest but has not asked for a meeting.
Reese said analyzing the impact of each player on football and basketball games is much tougher than in baseball, which is largely an interaction between pitchers and batters.
"It requires much more data," he said. "You can't evaluate player contributions with the standardly available data."
Reese said the Bayesian hierarchical modeling used in the analysis of over-under bets might result in an academic paper. He has published on a wide range of topics, including a study estimating gestation in bowhead whales.Reese also is working on a national panel studying the reliability of U.S. biological weapons detectors and with the National Center of Atmospheric Research in an attempt to better understand the impact of the upper atmosphere on climates.
- Riverton sees 550-acre LDS Church property...
- Paradigm shift: Fewer Utah juvenile offenders...
- Police investigating man who barricaded...
- About Utah: Selling bikes the new-fashioned way
- Woman killed by mixer at Sandy grocery store...
- Supporters of Oregon occupier honor LaVoy...
- Police ID man who barricaded himself in motor...
- Should Utah have 'blended sentences' for teen...
- Supporters of Oregon occupier honor... 53
- Feds say Orem man duped hundreds of... 28
- Meagan Grunwald won't get a new trial... 25
- Are you willing to take a pledge to... 19
- Woman killed by mixer at Sandy grocery... 16
- Video involving officer whose ear was... 15
- Paradigm shift: Fewer Utah juvenile... 15
- Should scientists be allowed to change... 10