When the longshot bias disappears

In horse racing it's pretty well known that pari-mutual pools attract a disproportionate amount of money on long-shots. Horses going off at long prices tend to be over bet, while short money horses tend to be under-bet. Research has found this time and again, over decades of studies.

The bias by itself isn't enough to overcome the track take except in the case of very short priced horses, odds on favorites are sufficiently under-bet so that you could actually make a long term profit by just flat betting every odds on favorite. The profit margin would be very slim though.

What that means is that gambler's are willing to pay for a shot at a big payoff. Not really surprising when you think about it. There's not a lot of thrill in cashing a $2 ticket for $2.20. But cashing a $2 ticket for $40 gives you a pretty good kick.

I was reading Why are Gambling Markets Organized so Differently from Financial Markets? by Steven Levitt of Freakonomics fame when I ran across this about NFL betting markets

Bettors exhibit a systematic bias towards favorites

That observation by itself isn't new to me, but he points out in a footnote the difference between this bias and the horseracing favorites/longshot bias. It's just not something I've really thought about before. In NFL betting, payoffs are even money, so there's not really an incentive from a big potential payoff to bet on underdogs, they might be longshots to win, but not with corresponding big payoffs. Of course, after point spread adjustments the underdogs aren't always even longshots to win, which is part of the broader point of the article.