Sports Economics is not just about The Wages of Wins. To make that point, I am starting a new feature that spotlights the empirical work of other sports economists, similar to our discussion of the work of Beck Taylor and Justin Trogdon in the book. My first review in this forum will examine the issue of corruption in NCAA men’s basketball.
Justin Wolfers (The Wharton School, University of Pennsylvania) recently published “Point Shaving: Corruption in NCAA Basketball” in the American Economic Review [May, 2006; pp. 279-283]. As the title indicates, the study Wolfers offers examines the prevalence of point shaving in NCAA men’s basketball games.
Point shaving is where players/coaches purposefully keep the point difference between their team and their opponent below the “point spread”. For example on November 29th 2006, Wake Forest played Air Force. The point spread for that game was (-11), with Air Force the favorite. So for one to win a bet on Air Force, the Falcons would have to not only win the game, but also win by more than 11 points (which they did).
Here is where corruption can enter into the picture. Wolfers argues that players could “be bribed not to cover the point spread” (italics in original). Such bribes would be directed towards the favorites, not the underdogs. Bribing underdogs forces these players to commit to losing a game. Giving money to the favorites, though, allows these players to not only win the game but also collect extra money.
Now which favorites should a gambler approach? Wolfers argues that approaching the favorite in a game with a small point spread would be difficult. Players in these contests are going to struggle just to win the game. But in a game with a very large point spread, the favorites should be able to win with greater ease and hence – if point shaving is an issue – we should see these favorites fail to cover the spread more often than one expect just from chance.
Wolfers gathered data on 44,120 NCAA games from 1989 to 2005 and first finds that all favorites beat the point spread in 50.01% of the games. Now that is some pretty good odds making! Yet for strong favored teams — favored to win by 12 points or more — only 48.37% of the 9,244 games covered the spread. Using kernel density estimates of the winning margins, Wolfers shows that the observations are consistent with strong favorites shaving points.
Wolfers does address some factors that would lead to skepticism of his results. One is the change in effort among teams that are winning by large margins at the end of a NCAA basketball game. Wolfers examination of the data suggests this is not an issue.
In concluding the paper Wolfers states that “these data suggest that point shaving may be quite widespread” (italics in the original). From his estimates it is suggested that 1% of all games, or 500 games in his sample, involve gambling related corruption.
Wolfers does note that although one can see in the data a systematic pattern, it is quite difficult to identify specific culprits. Still, this study does give us something to think about next time ESPN broadcasts a game between one of the NCAA’s top teams and a school difficult to find on the map.
– Stacey
For those interested in reading the entire article:
Wages of Wins Journal 
Martin
December 3, 2006
I haven’t read the article yet, but one thing that was not mentioned in your post is the fact that the impact players most likely to be solicited, are also those most likely to be rested at the end of blowouts. If a team is favored by 12 points, and up by 10 points or more with less than a minute, they are not going to be in the game. The benchwarmers will be in the game and in the position to influence the final score.
Jalen
December 3, 2006
Crappy research. Big favorites are bad plays in all sports, this is NOT evidence of point shaving.
Average gamblers like playing the better team and so the line is inflated. Winning gamblers in most sports play almost all underdogs.
Jalen
December 3, 2006
I want to add: you should be ashamed to support this research.
Do you think it’s a good idea to denigrate a healthy portion of college basketball players as criminals just because some sports economist finds some data showing big favorites don’t cover the spread that well and decides that the cause MUST be point shaving???
I mean there is no possible other explanation so let’s call them criminals?
FYI all Home underdogs in the NFL have covered 52.5% over the past fifteen years. Rampant point shaving in the NFL by away favorites says Justin.
Home underdogs getting a touchdown or more have covered 56%. Send in the FBI and arrest all these NFL players says Justin.
This is shameful research and should not have been published.
Jason
December 3, 2006
While I agree that there are other possible explanations, your reaction reads much like someone who had decided apriori to fly off the handles and much less like someone who actually read the paper. The conclusion was that point shaving *may* (author’s emphasis) occur. [If you did not read the arcticle, Stacey was kind enough to present that same statement with same emphasis and same notation of emphasis, which makes your reaction a bit at odd with the actual paper.] Wolfers supports that *possibility* (my emphasis) with data on results and an independent study by the NCAA indicating that players are indeed approached by gamblers.
“Be ashamed” for presenting this? That seems like an awful harsh charge. Would it not be more interesting to indicate what might be responsible for these data than to toss insults?
Jalen
December 4, 2006
No Jason I did read the paper. Did you?
“These data suggest that point shaving may be
quite widespread, with an indicative, albeit
rough, estimate suggesting that around 6 percent
of strong favorites have been willing to
manipulate their performance.
This estimate derives from
analyzing the extent to which observed patterns
in the data are consistent with the incentives for
corruption derived from spread betting; other
forms of manipulation may not leave this particular
set of footprints in the data, and so this is
a lower bound estimate of the extent of corruption.”
And how about this:
“A shortcoming of the economic approach to
identifying corruption is that it relies on recognizing
systematic patterns emerging over large
samples, making it difficult to pinpoint specific
culprits.”
You would think he was about to say the shortcoming was that he had no proof that the cause of the big favorites reduced cover rate was point shaving and that he was making wild and slanderous speculation. No the shortcoming according to him was he couldn’t identify the specific teams doing the point shaving.
And here’s the last paragraph:
“The advantage of the economic approach is
that it yields a clear understanding of the incentives
driving corrupt behavior, allowing policy
conclusions that extend beyond the usual platitudes
that “increased education, prevention, and
awareness programs” are required (NCAA,
2004, p. 5). The key incentive driving point
shaving is that bet pay-offs are discontinuous at
a point—the spread—that is (or should be) essentially
irrelevant to the players. Were gamblers
restricted to bets for which the pay-off was
a linear function of the winning margin, their
incentive to offer bribes would be sharply reduced.
Similarly, restricting wagers to betting
on which team wins the game sharply reduces
the incentive of basketball players to accept any
such bribes. This conclusion largely repeats a
finding that is now quite well understood in the
labor literature and extends across a range of
contexts—that highly nonlinear pay-off structures
can yield rather perverse incentives and,
hence, undesirable behaviors.”
Let me repeat that so you get it:
“The advantage of the economic approach is
that it yields a clear understanding of the incentives
driving corrupt behavior”
That’s not someone who is admitting this is all fanciful conjecture, no this is some smarky sports economist who has made the wrong conclusion and is standing by it 100%. He has identified “corrupt behavior” through his “economic approach.”
If I was the NCAA I would sue him for libel and slander since he no doubt talks about this in conferences. Make him prove that 6% of big favorite are point shaving incidents.
Think about this for a moment: how does the point shaving work? One player, who as others have pointed out isn’t even in the game at the end of a blowout? The whole team? Does the team live on the edge of losing the whole game? Do they build a big lead and mysteriously miss the last fifteen shots and free throws of the game to get under the spread?
This is repulsive research where a guy finds some data in the pattern, easily explainable by betting markets if he had taken the time to look at other sports, and makes the wrong conclusion.
Betting on every Randy Johnson pitching game this year was a negative and he was usually a huge favorite. Arrest him says Justin.
The underdogs had another huge week in the NFL this week. Arrest the players and coaches says Justin.
The paper is titled “Point Shaving: Corruption in NCAA Basketball” and the header on the concluding page is “Exposing Cheating and Corruption.”
You tell me where this guy expresses doubts about the conclusion he is making. Tell me why the guy shouldn’t be punished for accusing college basketball players of being criminals without a shred of evidence that would stand up for someone who knows sports gambling.
RSaunders
December 4, 2006
Jalen, it is common that one can predict something on average in a population and yet not be able to predict which individuals in particular exhibit the behavior. For example, suppose we “know” from a study of alcoholism that 20% of people who start drinking before age 15 will become alcoholics as adults, the ability to predict this on average for a population does not imply that we can predict which 1 out of 5 people that will be. And this inability does not, by itself, invalidate the observation that 20% of early drinkers become alcoholics.
RSaunders
December 4, 2006
I think all this does is define a possibility space for point-shaving (i.e., no more than about 1% of games could be shaved games).
I’m curious: why would the author not have been able to determine “team” effects or “conference” effects from this data? Getting coach information onto the data would probably be difficult, but team and conference would still be possible. Still, I think he should be able to identify whether some teams were more likely to be in that 1%.
Also, I’m not sure the 12-point rule is that strong of an indicator of how strong a favorite a team is because not all >12-point spreads are the same. Being favored by more than 12 against a team of comparable skill may give the favorite less power to manipulate their advantage. That is, I think the standard deviation of the difference in observed scores may be wider when playing teams of comparable ability (e.g., teams in the same conference; teams from conferences of similar ability such as Big 10 v. ACC).
Also, are there potential ceiling effects when favored by extreme amounts? I wonder if there are not limits when you’re favored by more than 30 or 35 points that make is hard to exceed. What fraction of that 1% of potentially shaved games are accounted for by, say, games where a team is favored by 35 points or more? Is the betting line as close to 50-50 in these extreme cases as it is in the population overall? I bet not.
Jason
December 4, 2006
(Yawn)
Yes, Jalen, I read the paper.
Now, do you actually have a point beyond hyperbolic exaggeration of positions that Wolfers didn’t actually come close to taking and assinine charges of libel and slander?
Jason
December 4, 2006
There’s also the potential factor that knowledge of the point spread in absence of any external pressure can influence behavior. In a losing effort where the margin is *close* to the actual spread, there may be some psychological incentive for the underdog to try to ‘beat the spread’ that isn’t as pronounced for the favorite. I’m not convinced that the measure of distribution of wins above and below the spread that Wolfers uses indicating that there are more blowouts than would regularly be expect would address this as once a game gets out of hand relative to the spread, this incentive for the ‘moral victory’ from the underdog vanishes.
barry
March 26, 2007
Get NBA league pass. Tivo the endings to all the games.
A perfect example of suspicious behavior is in a matchup between the LA Clippers and the Washington Wizards on Saturday, March 24th 2007 when the over/under line for the entire game was 207.
The Wizards were coming off a game the previous night where they combined with the Golden State Warriors to score an incredible 263 points.
The first half line with for the matchup between the Wizards and Clippers was 100, (keep in mind the entire line for the game was 207 and the public was heavily on the over for first half) the officials let both teams play rough with one another for the entire first half rarely calling a foul and neither team getting into the foul penalty the entire first half. Then comes the second half where the line was set at 106 and the public bet heavily on the under, probably factoring in that the refs were not calling fouls and the Wizards must have been tired from play a 263 point fast pace game the previous night. Both teams were in the penalty by the 8th minute in both the third and fourth quarter. The two teams had approximately 10 personal fouls called in the first half. Then came the second half when the two teams combined for 42 personal fouls. The two teams attempted around 15 free throws in the first half and then shot around 54 free throws in the second half. (These figures are not concrete because I can’t find a site that shows fouls and free throw attempts by halves or quarters, if any of you do you can surely correct and verify this)
With 3 minutes to go and the Clippers leading by 10, the two teams had scored only 88 combined points. Needing 19 points to cover the over/under for the second half and 25 points for the entire game’s over/under which was set at 207, the two teams inexplicately started fouling each other and putting one another to the free throw line. The two teams eventually scored a combined 39 points worth of free throws and wide open layups.
I was at the game and also had it Tivoed to see if I was on camera. During the ending, a lot of the crowd, who probably bet on the game, starting yelling, “FIXED!” And when I watched the recording on my television again, it was not on there. But in person, the yelling of FIXED! would have been impossible not to be heard. Also there is a shot of Gilbert Arenas winking at a ref while a Clipper was shooting a free throw! I know winks don’t mean guilt but when two teams score 39 points in 3 minutes in a game that could hardly be described as close. And fouling down by 7 with a full minute left on the clock is suspicious to say the least.
If any of you happened to Tivo or see that game as I did, please post your opinion.
barry
March 26, 2007
When I say wide open layups, I mean so wide open that some of the crowd at Staples Center were looking at one another confused. The wide open layups by the Wizards were followed by intentional fouls by the Wizards immediately afterwards even though there was plenty of time on the clock. Why would you foul with so much time left when you can simply defend, hope the other team misses, and still have plenty of time on the clock to stage a legitimate comeback. By fouling after a layup, you are earning two points and then immediately giving it back for no apparent reason…
Frank
March 17, 2011
Jalen, respectfully, you’re a sucker. “Winning gamblers in most sports play almost all underdogs.” Spoken like a born loser. Lay the points! I.e., in 2008 and most of 2009 the Gators covered….WINNING!!
Know what you don’t know.