Sports betting and the limits of rationality
Earlier this year, I wrote most of a book about the psychology and mathematics of sports gambling called Your Parlay Sucks. The book never quite came together, and is probably too weird to ever get published, but it has some interesting bits, so I figured I'd share them here.
Why did I get interested enough in sports betting to write a whole book about it? I think it's because I'm fascinated by the limits of rationality. Philosophers, economists and social scientists would like to treat humans as though they are capable of making rational decisions. That conflicts with the real world, where even pretty smart people make irrational choices. I certainly have.
Sports betting is a sort of rationality lab. You and I might have different values or beliefs. What's crazy to me might be normal to you, or vice versa, but we should both be able to agree that placing bets that are guaranteed to lose money is irrational.
This paper, "Intuitive Biases in Choice versus Estimation", is a wonderful illustration of cognitive bias and irrationality in the realm of sports betting.
The researchers had people bet on NFL football against the point spread. If you're not familiar, the idea behind the point spread is to attract an equal amount of action on both sides of the bet by handicapping one of the teams. If you bet on the favorite, they need to win by at least the amount of the spread for the bet to win. The other side side wins if the team loses by less than the spread, or wins the game outright.
Gamblers can bet either side, and if there are more bets on one side than the other, the sportsbook can change the spread to attract equal action. So there's a potential for the wisdom of crowds to kick in, the invisible hand of the market moving the line towards the best estimate possible.
Of course, that depends on gamblers being rational. A rational gambler has to be willing to take either side of a bet (or not bet at all), depending on the spread. If the spread is biased towards the underdog, they should be willing to take the underdog. If it's biased towards the favorite, they should take the favorite. And if the line is perfectly fair, they shouldn't bet at all.
As I showed a while back with ensemble learning, the wisdom of crowds only works if the errors that people make are uncorrelated with each other. If most people are wrong about a particular thing, the "wisdom of crowds" will be wrong, too.
This study found that people are consistently irrational when it comes to point spreads. They will tend to bet the favorite, even though both sides should have an equal chance of winning. It's probably easier to focus on which team is better, and assume that the better team is more likely to win against the point spread as well. It's harder to imagine the underdog losing the game but winning the bet, or pulling an upset and winning outright.
The study took things further and adjusted the lines to be biased against the favorite team, so that taking the favorite would be guaranteed to lose more than 50% of the time. They even told the gamblers that they did this. And the gamblers still overwhelmingly picked the favorites. The researchers continued the study for the whole season. Even after weeks and weeks of steadily losing, being told over and over that the lines are unfair, the gamblers still preferred to take the favorites. They never learned. The participants got to keep their winnings, so they had an incentive to be right. And they still couldn't do it.
Sportsbooks have a lot of ways of trick people into taking extra bad bets, as I will show. But they don't really need to. People will consistently take bad bets even if they should know they're bad bets.