Tuesday, July 19, 2005

It's All Markety and Stuff

Orin Kerr asks if there’s any good reason to believe the betting odds at sites like TradeSports actually tell us anything about whom Bush will nominate to the Supreme Court, given that the decision-maker and his close advisors are presumably not betting on those sites. One commenter replies:
But don't you see? It's a market… and markets, like, have information, and its all markety, and there are these rationale actors, and, um, ...information!
A little unfair, but still funny. Despite my own affinity for things markety, I have to confess my skepticism about betting markets and idea futures as predictors of truth values. I’m reasonably confident that prices of propositional assets – that is, assets that cash out when the truth of a proposition is revealed – at least correlate to true likelihoods. I’m less sanguine about the specific difference in prices reflecting a specific difference in the true probabilities. I say this with full knowledge that some really smart people disagree. Since the reasons for my reservations are still inchoate and possibly ill-founded, I’ll resist airing them.

As for Kerr’s particular objection, however, another reader gives the correct answer:
I would assume that quite a lot of vetting and research occurs before a Supreme Court nominee is selected, and that research is not done by Bush's inner circle but by the inner circle's staff members/drones. Presumably these drones have some insight into who they've been asked to research. They also know how thoroughly and extensively different individuals were researched, and how recently that research was conducted.

The drones would therefore have a good idea who was on the short list. Are the drones placing bets on tradesports on the basis of this information? I don't know, but I have no doubt whatsoever that they've been talking to people who talked to other people who put bets on tradesports.


S said...

I think you're right on, Glen. It's perhaps easier to think about it this way: When people in Vegas are setting their odds, they rarely set them based upon what they think the results will be. Rather, they do their best to ensure that they will make the most money.

For instance, let's say there are 2 teams, Team A and Team B. Team A's fans tend to be really optimistic about their team, while Team B's fans are less so. Oddsmakers are going to set the odds as far in favor of Team B as possible, regardless of what they think the real result will be so that they get as close as possible to the same number of bets on either side. That's why the odds for 2 teams will often be different if it's in one of the team's hometowns (hometown fans are generally more optimistic about their teams' chances than reality would dictate).

Consequentially, while betting sites may provide a partially strong correlation to actual outcomes, it's going to be tainted by the real purpose of betting sites: to make money, not to predict the future.

Hoss said...

Here's an interesting AP article ("interesting", "AP", contracdictory terms?) on Hurricane futures markets.

$imon Mann said...

I have done freelancing works for an online bookmaker and I have future ambitions of working as an oddsmaker in London or Las Vegas. Therefore I find this subject very appealing.

Steve is partly right.

It is true that many oddsmakers aim to balance the bets to make a sure profit (the commission/juice average) from every match-up.

Note, however, that I write 'balance the bets' and not 'maximizing profit'. The reason I won't use the latter label is because seeking balance of the action simply is not synonymous to maximizing profit. Quite the opposite, in fact!

True statement:
When you balance the bets you decrease the risk of going bankrupt.

False statement:
When you balance the bets you increase your overall expected average profit.

If you are a skilled oddsmaker and have the capacity to evaluate match-ups correctly, the way to maximize your expected profit is to set the odds accordingly and hope for as much action on either side as possible. If the action happen to be huge on one side, that's not a bad thing from an expected profit perspective. You may not minimize the risk of going bankrupt when the action gets very unbalanced, but your expected and potential profit will be higher.

To relate it to Glen's IHS lecture on risks:

- The risk averse oddsmaker (who seeks to minimize the risk of going bankrupt) should prefer the 'balance the bets' approach.

- The risk neutral or the risk loving oddsmaker (who seeks to maximize expected profit and/or prefers large potential profit to small risk of going bankrupt) should prefer the 'maximizing profit' (i.e. true odds) approach.

I should add that my view (that balancing bets negates maximizing profit) is a clear minority view in the betting community. It would be interesting to undress this widespread myth further in a future paper on the subject!

S said...

Hm, interesting. I understand that when oddsmakers aim to balance bets that they do sacrifice the possibility of higher gains for surer gains. Isn't the idea, though, that in the long run this is better for the oddsmakers (in terms of preventing them from going bankrupt, which ensure profits in the long term)?

Anonymous said...

It's true that the odds which "balance the bets" may differ from the "true" odds. However, an expected-profit-maximizing bettor can take advantage of this discrepancy. If Team A's fans are too optimistic about their team, driving the odds down, then I can get a positive expected return by betting against Team A. Assuming that at least some (not necessarily all) of the bettors are rational, the "balance the bets" odds should reveal the true odds.

Anonymous said...

Hi there,

Steve (the first commentator) is spot on when it comes to bookmakers. However, the SCOTUS futures markets at TradeSports/InTrade were organized by a betting echange, not a bookmaker. These event-driven futures exchanges give true odds, providing that traders have open access to information ---which was not the case.


Best regards,

Chris. F. Masse