what I do for fun

((Photo from wikipedia: Hong Kong Jockey Club))

It's a sign someone is cool when they combine their academic interests with their recreational activities, right? Well, I just spent the last hour parsing this article, and, let me tell you, by the above definition, I'm pretty darn cool. For those of you not particularly interested in reading the whole thing, a summary:

It's an econ paper that addresses rates of return for betting on horseracing in the US, UK, and Australia. Basically, it's been empirically established since the late 1940s that the higher the odds on a horse (the less favored it is to win), the lower the expected payoff for bets placed on that horse (expected = based on how those races actually turned out). So you lose something like 5.5% betting only on favorites and something like 65% betting on 100:1 or worse horses. The question the authors of the paper are addressing is how to explain this discrepancy: do gamblers value risk more than they do actually breaking even, or is it that people don't have a good understanding of what it means to have 100:1 odds against something happening?

Cool part #1: the paper shows that these two explanations are functionally equivalent if you only look at the payoff data for simple bets-to-win. So there has not been a historical consensus because the two models explain the empirical evidence equally well.

Instead, the authors use 'exotic bets'--quinella, exacta, and trifecta bets and their actual rates of return--to serve as a tiebreaker between the two models. This is where my recreational time comes in: at the Hong Kong track, I prefer quinella bets (there's no distinction between second and third place here, so quinella is just picking two of the top three finishers). It's not exciting to pick a favorite, especially a prohibitive favorite, and so quinella betting allows me to bet on a "I have a good feeling about this one" horse to Show (top-3 finish) but with a substantially higher payoff, based solely on the inclusion of the prohibitive favorite to Show as well.

The paper concludes that, psychology of gambling aside, betting patterns more strongly correlate to systematic misunderstanding of very bad odds. This seems to be the more quantitative explanation, anyway, which I find reassuring. Whatever motivates people's individual decisions to gamble, collectively, it's more about our inability to rationally understand a probability of 1/80 than it is about devaluing breaking even in favor of that 1/80 longshot.

Cool part #2: in a footnote, the authors observe that while their data from the US, UK, and Australia shows the same unbalanced payoffs (within confidence intervals), data from HK and Japan does not. Which means I can bet on medium-range to longshot horses (say 3:1 to 20:1) here with more confidence that collective misperception is not dragging my odds downward.

Yep, nothing quite like reading and then summarizing an academic article for fun to start a monday off right!