More thoughts on the…

More thoughts on the election:

Futures Markets: TradeSports is one of the best known futures markets, and it uses real money. Players buy and sell contracts for certain situations (e.g. Bush wins popular vote, Redskins win the Super Bowl, etc.) and the contracts pay out $0 if they don’t happen or $100 if they do. Before election day, the TradeSports contract for Bush and Kerry were trading close to $50. Once the bogus exit polls hit the Internet, the Kerry contract jumped to $70+. As the night went on, and it was apparent how bad those early numbers were, the contract trades started going in the opposite direction. The question then becomes; how good are these futures markets at predicting the future if they’re just as subject to polls as the networks are? Why not just stick with the polls we already have?

A lot has been written about Idea futures markets and their predictive value. One web site by Robin Hanson has links to a lot of such articles, including many he has written. Robin is also a player in a game that I mentioned earlier, the Foresight Exchange, which is similar to TradeSports but using fake money. While I have an account with FX, I don’t play it as actively as others do. However, I stay on the mailing list because of the interesting discussions there. The predictive value of ideas futures has generally been considered quite good according to those in the field, but on the discussion list today I found some folks sounding more skeptical, especially in relation to their original idea that futures markets based on real money would be better predictors that fake money ones. Here’s how the discussion went (names expunged):

Player A: “Tradesports, IEM, Betfair give Kerry a 71 to 74% chance to win.” [This was at 5:30pm Eastern time. “IEM” is the Iowa Electronic Markets and Betfair is another (mostly sports) futures market.]

Player B: “Bush is even _lower_ on iowa–0.26 or so. Looks like W will really have to pull a rabbit out of the hat at this point. Having Jeb yank Kathy Harris’s chain won’t work this time.” [Also around 5:30 pm. This morning I replied to this, “Anyone see a rabbit around here?”]

Player A: “Tradesports now gives Bush a 62% of winning.” [10:02pm]

Player C: “Well, I’m starting to wonder if maybe real money isn’t enough in itself to make a good market. IMHO, FX has been better than TS for the past few months. Looks like FX has some advantages. I guess player skill and better transparency.” [11:15pm. First sign of skepticism. Now some of the heavyweights are having their notions about using real cash challenged. Frankly, if there is predictive value to futures markets, I too would’ve thought that putting your own money on the line would make it a better predictor.]

Player A: [at 7:00am]

Well the betting markets made a big reversal last night, from up to almost 75% in favor of Kerry, now down to a 5% chance for Kerry (even that looks too high, so I finally made my first election bet).

Of course we should expect this sort of reversal at least 1/4 of the time, so one can’t be too stunned. But this does at least raise a small suspicion that theses markets were too volatile due to over-confidence. However, one case won’t really show this – we have to look at statistics over many events to see if there’s a trend.

In the end the important question is comparative – are there any other institutions that on average do better? So far direction comparisons between markets and other institutions in the field have favored markets. And real and play money have come out about the same. But the jury is still out.

Later his morning, another player has summed things up this way, noting that perhaps the markets didn’t really add any new information to the mix.

It seemed to me that the money markets were poll driven. When the early exit polls seemed to say Kerry had the advantage the money markets moved swiftly for Kerry, and perhaps moved faster and further than they should have. Perhaps people hoped to turn a quick reaction to the polls into a tidy profit.

Later in the evening the money markets were declaring a Bush win while the networks were still waffling. I suppose that the networks also knew Bush had likely won but refused to say so, out of fear of looking bad. The networks have a vested interest in a close election–if they say Bush has won everyone will shut off the TV, go to bed, and stop watching the paid advertisements. Their reluctance to call it is understandable given the egg on their face last time, and the economics of it. The market’s only incentive was to call it straight up, and did so.

Two general comments:

— if the money market is poll driven, meaning, if movement in the market can overall be explained by the publication of polls, then there is no further research being done by market participants and the money market can only be WORSE then the polls, given that it simply adds market noise to them.

— the market WAS good at cutting through the reluctance to declare a winner and present a clearer summary of the situation than the television anchors were willing to, despite apparently being based entirely on information provided by television anchors. So the market seems to be pretty good at presenting a clear summary of the information that has been presented; but it is not clear in this case that the markets had any *additional* information to go on.

Perhaps if the markets were much larger people would invest in independent research–until that point, is it reasonable to think that the markets can do better than the polls, if theoretically they are driven by the polling data?

Yes, the jury is still out, and this has really given the futures market folks a lot of data to chew on. The concept does sound interesting to me, but when you do this with real money, you have to add the “quick buck” factor in. Many people will play the market, not so much to get the cash benefit of being right, but just gaming it based on gut feel and trying to get some quick cash out of the whole thing. I think that adds “noise” to the information trying to be expressed by the market and you don’t really get a good picture. Then again, these are humans we’re talking about. One more poster said this:

But was it really rational to accept this poll data at face value? In retrospect, the data was wrong. And many commentators during the day pointed to inaccuracies and bias in raw exit poll data, such as http://www.mysterypollster.com/main/2004/11/exit_polls_what.html . And historically the midday exit poll data has not been that good, in 2000 and 2002 it was pretty far off. Yet the markets seemed to move based on this data.

Maybe the markets have learned their lesson and will be more cautious in the future, but institutionally it should not be necessary to learn in this way. This historical data should have been enough to teach the lesson.

But the markets are only human.

Filed under: Uncategorized

Like this post? Subscribe to my RSS feed and get loads more!