Super Bowl Bettors and the Folly of the Crowd

Photo credit: Pascal

Photo credit: Pascal

Two days before Super Bowl XLVIII the Broncos are still the favorites to win, and despite oddsmakers spotting the Seahawks about 3 points, 90% of early bettors have put their money on the Broncos anyway.

So if crowds are wise, why bother to watch the game at all?1 If it weren’t for the commercials, the food, and the parties, why waste our time? A study by Joseph Simmons2 at Wharton demonstrates why the game this Sunday may not be a foregone conclusion.

“The wisdom-of-crowds hypothesis derives from mathematical principles,” Simmons explains. “If a crowd’s judgment comprises signal-plus- noise, averaging judgments will cancel out the noise and extract the signal. Two conditions are necessary for the production of crowd wisdom. First, and most obviously, at least some members of the crowd must possess relevant knowledge. For example, a crowd comprised entirely of people who know nothing at all about major league baseball would err considerably if its members were asked to predict the 2012 on-base percentage of Nick Markakis.”

When it comes to the Super Bowl, the signal is dwarfed by the noise. “It’s the only line that we make for the entire year that is geared towards the betting public,” said Jay Kornegay, the director of the sports book at LVH — Las Vegas Hotel and Casino. “Most of the betting public are just your average Joes that are just fans and doing it for entertainment. They will overrun any type of sharp money that we will take on the Super Bowl.”

But even so, the crowd is still wise if the considerable noise cancels itself out, which also means that crowd members cannot exhibit biases in the same direction. “If all of the judges in a crowd make the same mistake”, Simmons says, “then averaging responses will obviously not negate the error.”

But Simmons found that sports betting crowds are extremely biased.

“When predicting against point spreads, bettors’ initial inclination—their intuition—is to believe that the superior team (the favorite) will win against the spread. Moreover, bettors are usually quite confident in their intuition to choose the favorite…[Even when] point spreads quite accurately balance the favorite and the underdog, gamblers are substantially more likely to bet on favorites, and they seem to lend insufficient weight to point spreads when assessing which team is going to win against the spread…[In fact,} one experiment found that the majority—the crowd—predicted favorites in over 90% of the games…, even though favorites and underdogs were equally likely to win against the spread. [Of course], when favorites and underdogs are equally likely to win, betting on favorites more than underdogs does not constitute evidence that crowds are unwise, any more than would a systematic tendency to bet “tails” on a series of fair coin flips….Although systematically betting on “tails” is at worst merely peculiar when the coin is fair, it is distinctly unwise if the coin systematically, and detectably, favors heads.”

In other words, when both teams are equally likely to win against the point spread (taking into consideration the points that oddsmakers spot the underdog), crowds may exhibit a bias for the favorite, yet still be wise. In order to prove that sports betting crowds tend to be unwise crowds, bettors would exhibit a bias even when they knew they couldn't trust the points the oddsmakers spot the underdog.

To test this, Simmons conducted another experiment where NFL fans wagered over $20,000 on NFL football games over the course of an entire football season. In this study, however, Simmons told the crowd that instead of using the lines set by the oddsmakers in Las Vegas, we will instead use a line that we have rigged in favor of the underdog. Contrary to the wisdom-of-crowds hypothesis, people were more likely to pick “favorites” than “underdogs”, even when they were told that the point spreads used to calculate the results had been rigged in favor of the underdog.

Of course, I’m writing this before the players ever take the field. The Broncos may actually win by more than 3-points. An unwise crowd can still be correct. And truth be told, although I am rooting for Peyton to win another Super Bowl, when I wake up Monday morning I won’t really care which team team wears the world champion3 crown. I would care, of course, if the Saints were playing instead of the Seahawks. Or if I actually gambled on sports.

In fact, this is probably a perfect example of how quickly we see others' biases while remaining blind to our own.

  1. Simmons cites examples of how (1) “students’ average estimate of the temperature of a classroom was only 0.4 degrees from accuracy, a result that was better than 80% of the individuals’ judgments” (2) “The average guess [of the number of jelly beans in a jar] was 871, very close to the true number (850) and better than 98% of the individual guesses” (3) Fair attendees’ average guess of the weight of an ox “was 1,197, just 1 pound away from the 1,198-pound ox’s true weight!”
  2. In order to make this post easier to read, I incorrectly attribute this work solely to Simmons. It should be noted however, that Leif Nelson, Jeff Galak, and Shane Frederick.also contributed to this work
  3. I find it humorous that we call the winner of the Super Bowl the world champion, since only teams from the Unites States who happen to belong to the NFL are allowed to participate. Since the NFL has deeper pockets than any other league, they can attract the best players to a bigger stage. So we may be correct to assume a team in any other league would lose to an NFL team. So why play the game at all? Why not crown the Broncos the world champion now? The crowd has already spoken.