Economist Claims to Have Evidence of Point Shaving
MICHELE NORRIS, host:
From NPR News, this is ALL THINGS CONSIDERED. I'm Michele Norris.
ROBERT SIEGEL, host:
And I'm Robert Siegel with an item now about betting on college basketball games, and a claim that point shaving is common. We're not talking about the kind of betting that millions of March Madness bracket fillers are currently engaged in, picking the winners in the NCAA Men's Basketball Tournament. We're taking about betting on a single game in which a bookmaker offers a point spread.
You don't bet on whether Hawksley will beat Darwin. Rather if Hawksley is favored and there's a 9 and a half point spread, you bet on whether Hawksley can win the game by 10 points or more. Over the decades, there have been occasional point shaving scandals. That's when a player on Hawksley muffs a couple of shots, not so badly as to throw the game, but to win by, say, five points, anything less than the point spread. For such illicit cooperation, gamblers have been willing to pay handsomely.
Well, how common is that? The NCAA says it's rare. But economist Justin Wolfers of the Wharton School of the University of Pennsylvania has written a paper saying that it has to be more common. He joins me now from Philadelphia. Justin Wolfers, you've used the methods of a field called forensic economics. What have you done?
Mr. JUSTIN WOLFERS (Economist, Wharton School, University of Pennsylvania): The question you start by asking yourself is, if there is corruption occurring in sports, what sort of footprints would it leave behind in the data? And so here if teams are point shaving, who would point shave? Presumably a strong favorite like Hawksley. And so you'd look for, for instance, games with a 10 point spread, how often did a team win by 11 or 12? Not often enough, versus winning by, say, eight or nine. And it turns out it looks like a little bit too often.
SIEGEL: Now, when you say you looked at some scores, you looked at a database of thousands and thousands of scores.
Mr. WOLFERS: This is 44,000 games, pretty much every major NCAA Division 1 game over the last 16 years.
SIEGEL: Now, could one argue, let's say, bookmakers, just simply overstate this spread when a very successful team is playing? Because if not, there will just be very little gambling on a game that featured, say, a very successful team, Duke, for example, so that there would be a natural impulse to offer a bigger spread than they're really going to win by?
Mr. WOLFERS: Yeah. It turns out it's not so much of an issue. And here's the test for that, if you thought that people were consistently overbidding the Dukes of this world, then teams that maybe should have been a 12-point favorite will turn out to be a 14-point favorite. So relative to the betting spread, you would expect them to not win the game often enough, or to not be involved in large blowouts, winning by 20 points often enough, or so on.
In fact, what we find is that too often 10-point favorites win by eight or nine, and too often 15-point favorites win by 13 or 14. And it's that sensitivity of game outcomes to whatever it is that's happening in Las Vegas that that story can't quite reconcile.
SIEGEL: Now, a spokesman for the National Collegiate Athletic Association has said that while the problem is there, it's not a crisis. A recent poll of theirs asked players if they knew of a teammate taking money to play poorly, and the players who answered yes were 1.5 percent of the sample. Could you reconcile that number with the number of instances of point shaving you infer from the data you've studied?
Mr. WOLFERS: Yeah, I actually think we're pretty close to agreeing. If you want to catch crime, usually, we think the best way to do it isn't simply to ask people, did you commit a crime? We might think that would typically lead to an understatement. But even when we do that, which is what the NCAA does, they find, maybe, one or two percent of players who are willing to admit to having been involved in point shaving at some point in time.
I find that probably around five percent of all strong favorites are involved with point shaving, suggesting maybe 500 games have involved some sort of point shaving. So to me those numbers look awfully similar.
SIEGEL: I understand how one could look at data and say, here was all the income for income earners in America. Here's what the tax rates should be, and here's how much the tax collections were, and you would do the arithmetic and say a lot of people are cheating on their taxes.
Mr. WOLFERS: Yeah.
SIEGEL: But this is something different. This is drawing an inference from probability that suggests corruption.
Mr. WOLFERS: Most of these inferences we draw from probability, but that's precisely the point. Given that it's so difficult to figure it out from watching the videotape, what we might want to do instead is look systematically for footprints in the data.
SIEGEL: But you might actually be finding teams that choke, say, and when they're big favorites they let down their guard.
Mr. WOLFERS: That's absolutely a possibility. But again, one needs to think about what the evidence here is. Too often we see a team just fail to cover the spread. Too rarely, justly able to beat the spread, and exactly the right amount of times the team wins the game. So that's exactly the sort of pattern that you'd expect to see if a team was point shaving. If they were simply letting down their guard, you might expect them to do slightly worse all the way through the possible distribution of game outcomes.
SIEGEL: With so many scores that you've examined, could you look and find a particular school that has such a bad record against the point spread when it's heavily favored that that particular team draws attention to you?
Mr. WOLFERS: Believe me, I've tried that exercise. The difficulty is, because this is an inference based on, as you say, probabilities, one needs mountains of data. And for any given team I don't have mountains of data. You know, forensic economics can never be particularly accurate. It can tell you that there's a problem here, but in this case I can't be accurate enough to tell you where the problem is or where to look. I'm just telling you that perhaps we should be look.
SIEGEL: Well, Justin Wolfers, thank you very much for talking with us about it today.
Mr. WOLFERS: Thanks, Robert.
SIEGEL: Justin Wolfers is an economist at the Wharton School at the University of Pennsylvania. He has been investigating betting on college basketball.
NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.