Betting Markets

Surprisingly Accurate

The betting markets for President aren’t legal in the US. There used to be an experimental market called the Iowa Prediction Markets. It was a great market for academics, primarily economists, to watch how the prediction market did when it came to actual events.

I never followed the Iowa markets particularly closely. However, in 2009 when they passed Obamacare I did. The market predicted the passage far before the vote. I was a bit shocked since Obamacare was such a terrible law. I didn’t see how legislators in their right mind could vote for such a terrible government program. Turns out, I was wrong in my analysis on the rationality of legislators, and 100% correct about how Obamacare would perform. I was also dismissive of the Iowa prediction market because so little money was involved.

I had made my living in futures markets, which are kind of prediction markets since there is speculation involved. But, futures markets are big, with lots of liquidity and big money. The risk transfer and true hedging that occur in those markets far overwhelm the amount of true speculation there is.

I have watched the prediction markets with interest this election cycle. Right now, they all basically have Trump at about 58-59%, and Kamala down around 40%. PredictIt is slightly different since it only has a market for Kamala. It’s at .47 but if you look at the 90 day trend, it’s not good for Kamala.

Predictably, hard-left publications are decrying the platforms that allow people to place a wager on events. They’d be silent if it was different. The news media is so predictable and has no credibility anymore.

However, actual PhD-level academic research says the hard-left publications are wrong in their analysis. Here is a link to a 2012 paper. It is not the only one. Here is the summary of that paper. I might add, I know nothing about these professors but given their backgrounds, my assumption is they lean left.

Prediction markets--markets used to forecast future events--have been used to accurately forecast the outcome of political contests, sporting events, and, occasionally, economic outcomes. This chapter summarizes the latest research on prediction markets in order to further their utilization by economic forecasters. We show that prediction markets have a number of attractive features: they quickly incorporate new information, are largely efficient, and impervious to manipulation. Moreover, markets generally exhibit lower statistical errors than professional forecasters and polls. Finally, we show how markets can be used to both uncover the economic model behind forecasts, as well as test existing economic models.

Markets work when their market structure isn’t rigged or has another problem with it. They are better allocators of capital than centralized authority. They are better for guiding decision making.

When you read the study, I think this paragraph is helpful to understand why markets work.

Three inter-related facets lead to prediction markets’ ability to produce accurate, reliable forecasts. First, the market mechanism is essentially an algorithm for aggregating information. Second, as superior information will produce monetary rewards, there is a financial incentive for truthful revelation. Third, and finally, the existence of a market provides longerterm incentives for specialization in discovering novel information and trading on it. While these facets are inherent in any market, other forecasting mechanisms, such as polling, or employing professional forecasters, lacks one or more of them. For example, polling lacks incentives for truthful revelation, and professional forecasters may have other motivations than simply forecast accuracy

Interestingly, I read Talking Points Memo’s reason they don’t use prediction markets.

My analogy about scales is certainly imperfect in a number of ways, just as polls are imperfect. Indeed, it isn’t even really a question of which is better. The most important thing to understand about the relationship between polls and political betting markets is that the latter is largely downstream of the former. Most bets in political betting markets are driven by people looking at polls and betting accordingly. So by definition they can’t be better. Because the bets are derived from the polls.

I might suggest TPM review this short ditty from eons ago on YouTube.

I thought the same back in 2009. I was wrong.

As we go forward, it will be interesting to watch the prediction markets. If Trump has a humongous lead on the morning of the election, does it guarantee a victory? No! It doesn’t. As new information emerges it will get priced in. How people read the same cues differently will feed the market. I might see one county report numbers and think it’s great for Trump but the same data could be interpreted by someone else as being great for Harris.

That’s why there are markets.

I don’t think the prediction markets are good indicators of fraud. They only price the information as they receive it. Hence, if there is fraud, they simply price it in. The same thing happens with insider trading on Wall Street. It gets priced in as the market learns about it. However, the closer you get to the election information does get priced in so forecasting errors are far less.

This doesn’t happen with pollsters. They miss and often miss by quite a large margin.

Could there be insider trading in prediction markets? Maybe but I think it’s a lot harder. You’d have to have access to all kinds of non-public information and the information in politics becomes public so fast it would be hard to be in front of it.

Many pundits are saying the current odds in the Presidential race are being manipulated by some whale placing bets on Trump. The academic studies say this is false. Prediction markets are not easily manipulated. As a matter of fact, the night before the election prediction markets are incredibly accurate when it comes to politics.

Anyone like me who has traded for a living inherently has experienced this at some point. You are bullish and try to push the market higher and even in thin markets, your bids will be pushed back to whatever the market thinks is equilibrium. I have seen markets where traders push markets to the limit on the close. They get the settlement they want for overnight money, but as soon as the market can open it goes right back to where it started.

Here is what the study concluded:

Attempts by party bosses to manipulate the price of their candidates in turn of the century gambling markets were largely unsuccessful, as were more recent attempts by candidates themselves (Rhode and Strumpf, 2008; Wolfers and Leigh, 2002). More recent, and systematic, attempts at manipulation yielded only brief transitory effects on prices. Rhode and Strumpf (2008) placed random $500 bets (the largest allowed) on the IEM and found that prices quickly returned to pre-manipulation levels. Camerer (1998) placed large bets in pari-mutuel horse racing markets (which he cancelled moments before the race) to see if this would create a bandwagon effect of follow-on bets. It did not. Finally, evidence from experimental prediction markets, run in the lab, show similar results. In a first experiment where some participants were incentivized to try to manipulate prices, there was little evidence that these participants were successful (Hanson, Oprea and Porter, 2006). A second experiment based incentives on whether or not observers of the market price could be manipulated. Similarly, there was little evidence that manipulators affected the beliefs of observers (Hanson et al., 2011). Indeed, manipulators may increase the accuracy of prediction markets by providing more liquidity (Hanson and Oprea, 2009).

I would not ignore the prediction markets and I wish we had more of them for individual down-ballot races. I think the House races would be especially volatile given the lack of real good information, and prediction markets might make more information come to light.

(An additional thought)

Driving home from dinner tonight, I was thinking about how the prediction market almost 100% of the time predicted the actual election the day before voting day. But, when is the “voting day” now with early voting, absentee voting, mail in voting, and then the actual day?