The Curious Case of Brexit Polls and Bookies
There is something odd going on in the forecasting process around Brexit.
Polls are showing the contest between “Leave” and “Remain” about 50/50, a statistical tie. It’s definitely too close to call. Betting venues are showing the odds of Remain winning at about 75% with Leave about 25%.
How can we explain this huge discrepancy?
There’s an old saying that “A little learning is a dangerous thing.” It means it’s OK to be ignorant of something as long as you know your own ignorance and act accordingly. It’s also OK if you’re a true expert and use your expertise.
The danger is when you think you’re an expert, but you really aren’t — or, at least you don’t know as much as you think you do. That’s when you get into trouble. That’s a good summary of how a lot of presumed experts are misinterpreting the U.K. referendum betting odds.
The first thing to understand is that betting venues such as Ladbrokes try as hard as possible not to take a view on the referendum. They are middlemen trying to balance interest between two sides. The odds you see are not bettors versus bookies. They are bettors versus bettors with bookies just trying to balance the money flows in the same way a market maker in stocks balances interest between buyers and sellers by moving the price up or down.
The next point is that the odds do not reflect a balanced number of bets between Remain and Leave. In fact, there are far more bets for Leave than Remain. It’s just that the Remain bets are larger on average than the Leave bets. This probably reflects the fact that Remain bettors are bankers and elites, while Leave bettors are everyday U.K. subjects. If all bets were weighted equally, the odds would be heavily in favor of Leave.
Another problem is to understand what it means when bettors give a 75% probability to Remain. It does not mean that 75% of the voters will vote Remain according to bettors. All it means is that 75% of bettors believe Remain will get at least 50.01% of the vote. In other words, Remain bettors might see a close race, but a preponderance believe that they can foresee the outcome. That’s a very risky bet and probably reflects a lot of cognitive biases (confirmation bias, risk aversion, herding, denial, etc) rather than cool objectivity.
Finally, and most importantly, a betting pool is only a good reflection of the outcome if it’s a good representation of the voting pool. A poll tells you what the voters think. Betting odds tell you what bettors think the voters think. That’s two-steps removed from the actual vote with exponentially more ways to be wrong.
If the bettors do not resemble the voters, and if bets are not equally weighted (like votes) then betting odds can skew wildly away from the outcome at the voting booth. A good example of this was the 2015 election for Parliament when the betting odds showed an 85% chance of a “hung Parliament” when, in fact, the Tories emerged with a resounding victory.
Fans of betting odds often turn to a theory called “the wisdom of crowds” which was proposed by a book of the same name written by James Surowiecki. The idea is that the average opinion of a large group of everyday citizens can produce more accurate estimates than those of individual experts. This can be true in some instances, but not all.
Generally, the easier the problem, the more likely that the crowd will produce a better estimate than the expert. Classic cases include contests like “Guess the Weight of the Cow” or “Guess How Many Jelly Beans are in the Jar.” These are simple problems. The average of a large group of guess filters out noise and comes close to an accurate estimate. There are also no barriers to entry in these problems; no one has to pay money to play the game.
This wisdom of crowds theory does not transfer well to betting odds on Brexit where you have to put up some real money to bet (and be able to afford to lose it) and where there’s nothing simple about the arguments pro and con. Just requiring a money bet filters out those who cannot afford to bet or cannot afford to lose, which describes many eligible voters. Also, much betting is done online or with mobile apps, which also filters out some older voters who may be less agile with smartphones and apps.
I spent ten years building market prediction engines for the CIA, so I’ve studied these dynamics closely. All things considered, it seems likely that the polls have this right and that the outcome really is too close to call. The betting odds reflect a lot of biases and filters that cause them to diverge from the actual view of the electorate.
Why does this matter to you if you’re an investor? Because markets are putting great weight on the betting odds. In recent days we have seen rallies in stocks and sterling, and declines in gold. Markets are re-pricing risk based on a Remain victory as reflected in the betting odds. This may even be a feedback loop in which markets are set by wealthier bankers and institutions who themselves favor Remain and are among those placing bets. If markets are priced for Remain when the actual odds are 50/50, then investing in assets that will benefit from a Leave victory is an excellent risk/reward trade.
P.S. “A charmingly mordant take on the stock news of the day, accentuated by philosophical maunderings…” That’s how one leading financial magazine described the free daily email edition of The Daily Reckoning. You’ll find cutting-edge analysis from the complex worlds of finance, politics and culture. Presented in an entertaining style few can match. Click here now to sign up for FREE.