Strike Brexit up as another failure of prediction markets. But empirical studies show that betting markets are, on average, good predictors. In this post, I will hypothesise a specific category of instances in which markets systemically fail as predictors, with reference to my own bets.
Over the past year, I’ve found my intuitions consistently running against mainstream pundit opinion. Rather than just rant on Twitter, I decided to test my views on the market, embracing contrarianism for fun and profit.
I agree with Nassim Taleb that the only way to identify a genuine belief is to have skin in the game. Otherwise, predictions might be strategic, an attempt to create momentum for a preferred candidate. Or they might be currying favour with bosses and very serious people (VSPs) by saying what they want to hear. Or they could be a cheap effort to look smart despite having no real knowledge or expertise. Finally, they might be nothing more than pure entertainment. Most punditry falls into the last two of these categories.
Over the past year, I have decided to bet on four events that have now passed:
- Corbyn’s leadership win
- Turnbull’s leadership win (against markets at the time)
- Trump’s nomination
All of these were against markets, most were against pundit consensus (except Turnbull), but all were correct. In addition, I consistently predicted a Hillary Clinton win by a far more modest margin than pundits and popular prediction models suggested early on. I would have bet on this, if there had have been a convenient way to bet on winning margins.
I have two outstanding bets:
- Australian Labor Party election victory
- Trump general election victory.
No matter what happens on the remaining bets, I will still have made a large profit. I made no other bets during this time. The only other formal bet I have ever made was a late bet in favour of Gillard in Rudd’s final leadership challenge. That lost, but it was a long-odds bet.
Prior to embracing skin in the game, I made two other big contrarian calls. I predicted that the Liberal Party would be competitive in the 2010 election if they replaced Turnbull with Abbott. This was because he would be best placed to mount a brutal cost of living scare campaign on carbon pricing. That opinion ran against the wisdom not only of betting markets but virtually all the press gallery and senior political insiders. It turned out bang on.
My second non-financial call was the absurdity of statements like “Labor can’t win for three elections/a decade” etc after the 2013 election loss. In less than three years, Labor is already competitive. It surprises me that otherwise intelligent people continue to fall for such short-term emotive reactions against historical evidence. Two weeks is a long time in politics. It only took Labor eight years to win after the Whitlam dismissal, which was a defeat vastly worse defeat of a vastly more reviled government than Rudd-Gillard-Rudd.
My intention is not to boast, but to show that my opinions here are not just idle words. They are my genuine beliefs and I have had success based on these beliefs.
The common denominator in my predictions (except for Turnbull’s latest win) is that they ran against the “respectable” thinking. They clashed with the “serious” people.
When folks come up and ask me the secret of my success, I answer “well, you see, I look at the polls”. Actually no one asks me this, but that’s a good line when they do.
Seriously, though, my method is to identify when pundit consensus and “serious opinion” contradict long-term poll results. If the markets also contract the polling, it suggests what I will call a “pundit premium” is at work.
My hypothesis is that prediction markets are distorted by interrelating factors:
- a status quo bias against radical change
- a “respectable” policy bias (or an anti-populist bias)
- a “preferred outcome of elites” bias, and
- a herd bias.
The pundit premium is due to perverse incentives facing investing agents. If an investment goes pear-shaped, there is always the herd defence: “but all the serious, expert people said the same – clearly it cannot be my fault!” But if a contrarian investment is off, there is no such recourse. You’re just a crank. The investment was outlandish and irresponsible. Your bank will fire you and your clients will drop you, shunning you at cocktail parties.
This creates a powerful incentive to follow the ‘respectable’ line. In effect, you pay a “pundit premium”, which buys insurance against reputational damage. This will distort market prices compared to true probability.
On only this specific category of instances is it advisable to go against the markets. These instances are usually rare, but they are becoming more common. This is because we have entered a period of populist revolt against elites.
All over the world, unexpected departures from VSP consensus are becoming run of the mill. Some of these departures are good, some bad, and some down-right ugly.
But for all the bad, there is at least a silver lining. It is twilight of the pundits.