Story URL: http://news.medill.northwestern.edu/chicago/news.aspx?id=199799
Story Retrieval Date: 12/21/2014 12:52:31 AM CST
Now more than ever, if you don’t have a pocket protector, you don’t understand football. A motley crew of professional gamblers, college professors and renegade sports reporters are using “Moneyball”-style analytics to better understand the Super Bowl teams and players, often with surprising results.
The New York Giants touted running game, with its two stars Ahmad Bradshaw and Brandon Jacobs, may actually be worse than the New England Patriots much maligned one. Neither the best team in the AFC or NFC may have made it past their first playoff game. And at the bottom line, Patriots have a 61 percent chance of winning.
Those are at least the calculations of Football Outsiders, the website that revolutionized statistical methods in football, and of its founder (and now ESPN commentator) Aaron Shatz.
“Everyone knows about the Giants running game in terms of big names,” said Shatz “But the Patriots running game is much more efficient.”
With Jacobs and Bradshaw both averaging more yards per carry than the Patriots lead running back, BenJarvis Green-Ellis, it may be tough to fathom how the Shatz could conclude Green-Ellis as the best back of the three. But while yards per carry provides an easy, general quantification of how good a Brandon Jacobs is, it does not take into account game situations, consistency or the quality of the defense. But Shatz developed a statistic that does.
A player’s defense-adjusted value over average, known as DVOA, compares players with the league average. Accounting for all of the things that yards per carry does not, DVOA spits out the percentage better or worse than the average player at a position. It is not a statistic that is easy enough to calculate on the back of a napkin or explain in the pages of a newspaper. But it is a more complete metric of player performance. Bradshaw and Jacobs scored 4.5 percent and -1.6 percent respectively. Green-Ellis scored a 6 percent. DVOAs are available on the Football Outsiders website.
Using DVOA, Football Outsiders can model how a game would likely play out. They calculated 60.9 percent odds that the Patriots would win on Sunday. The odds are in New England’s favor, but not overwhelmingly so.
“It will not be a historic upset if the Giants win,” said Shatz.
That historic upset, he said, came in the Giants 2008 Super Bowl win versus the Patriots.
In a sense, the upset has already come. If the best team in the each conference made the Superbowl, the Superbowl would be very different.
“Green Bay or San Fransisco against the Patriots or Steelers,” he said.
But the football season is designed to allow worse teams into the biggest sporting spectacular of the year. Assign, at random, different levels of talent for each NFL player and simulate the regular and post season 1,000 times. The team with the best overall players and coaching, Shatz said, only will only win the Super Bowl a quarter of the time.
“Teams get hot,” he said, “Players get healthier and teams try different styles of offence and defense.”
That streakiness is one of the findings of a 2011 study by a three-professor, multi-university team, who found that Las Vegas was setting over/under lines – lines where bettors gamble on the point totals rather than the winner – by giving to much weight to the previous weeks play.
Tracy Rishell, an associate professor of management at Susquehanna University warned against getting too carried away with even just the postseason play of favorite teams."
"The playoffs aren't a large enough sample size," she said.
Predicting 60 percent of games correctly is astoundingly high – gamblers need to win less than 53 percent of their games to make a profit. That is why analytics are so popular among bettors, said Elihu Feustel, professional gambler, one-time casino consultant and co-author of “Managing Risk: Attacking Vegas and Wall Street.”
Feustel, of South Bend, Ind., devotes multiple chapters of his book to betting on football. He does not, however, like football.
“I think it’s boring,” he said. He is still willing to make money off of it.
Feustel noted that overestimating the value of last week’s play is only one way to make gambling mistakes. The public, he said, will bet on the Super Bowl by picking the underdog solely for the sake of a higher payout. This is a terrible reason to pick the Giants, he said, noting that the odds the Giants would win will be lower than the payout merits.
Feustel uses offensive and defensive points per game weighted by the strength of a team’s opponent’s schedule and takes into account teams turnover differential. He has the same basic prediction as Las Vegas: New England by three.
In his book, Feustel mentions that many games could be determined by which team was most injured. Assigning one point for each player listed as out for a game, 3/4 for each player listed as doubtful, 1/2 for those listed as questionable and 1/4 for those listed as probable, the team with fewer injury points wins 53 percent of the time. A team with a more than three injury point advantage wins 55 percent of the time. At press time, the Giants list two players on their injury as probable and the Patriots list one as questionable. Both teams have 1/2 injury points. Knowing how to use injuries is a useful tool for gamblers in general, said Feustel. It does not look like it will be useful for this Super Bowl.
But football, he said, is a tough sport to bet on to begin with. With so many people betting, and with the point spreads changing based on incoming bets, the point spreads in football narrow very rapidly. The window of getting a good line for a game is very slim.
“I am most interested in tennis because it has the smallest betting market, he said. “Tennis is much more beatable than any American sport.”
Baseball may be better known than football for the use of analytics. The Oscar-nominated movie based on a best-selling book, "Moneyball," traced the Oakland A’s road to success using advanced statistics , but no plans have been made for “Football Outsiders: The Movie.” But football analytics have their own vibrant research community in universities across the country.
Vince Genarro, executive director of the Sports Analytics institute at Manhattanville University in New York, said football analytics time has come.
“Baseball came first because it’s so much more difficult to analyze individual contribution from a player in football,” he said. “In football, there is so much interdependency.
Genarro is famous in the analytics community for his work back in the 1970s, when he first used the cost/benefit evaluations he had learned at PepsiCo to determine the values of baseball players based on the potential revenues they could bring a team. He said that he came to analytics when he realized that while PepsiCo would spend weeks determining whether to build a $40 million plant – 1/10 of 1 percent of their total revenues, baseball teams were determining player contracts by a few seconds of thought and gut feeling.
He said he believed that statistics he pioneered, such as wins over replacement player, are the future of every sport and that coaches such as the Patriots Bill Belicheck are showing the potential for data driven methods to generate wins.
“The reason analytics exist is because it puts conventional wisdom to the test,” he said. “And researchers have volumes of data and computing power that didn’t exist even 20 years ago.”
He does not, however, think that advanced statistics will ever be easy enough to compute or comprehend for a wide audience of sports fans.
“Belichek had that situation a few years ago where he went for it on a fourth down.” Genarro said, recalling a 2009 Patriots game against the Indianapolis Colts and a decision that ultimately failed. “He had the odds in his favor, and the fans still were angry at him."