Sell the rumour, buy the fact

As a Liverpool fan, I watched the latest football transfer window like a hawk. Of particular concern were the endless rumours of our playmaker Philippe Coutinho moving to FC Barcelona. But in the last few days before it shut, my attention shifted from individual transfers to whether rumours are generally over-hyped, in turn leading people to place too much likelihood on them really occurring.

Away from investment markets, a small group of my friends share suggestions for mispriced bets. Recently, one friend suggested that the odds for Coutinho moving to Barcelona overstated the risk given Liverpool’s insistence that he wouldn’t leave. The basic idea was that newspapers and websites attract readers through interesting rumours, but attract far fewer by talking down stories. The availability of lots of stories supporting players’ moving clubs could in turn lead people to over-estimate the likelihood of this happening.

The availability bias causes people to tend to over-estimate events when they’re more easily recalled. Sustained exposure to transfer rumours, but limited counter-information, could have this effect. Google Trends confirms that search on players’ names tend to peak when they are the subject of transfer rumours.

To investigate whether the rumour mill leads to an availability bias and mispricing, I analysed a third party list of seventeen potential transfers* where the betting odds implied that the players had a greater than 50% chance of moving clubs.

Only five of those potential transfers were completed and somebody betting on each of them would have lost over 60% of their initial stake. Of course, that could just be that all odds offered on transfers are very poor or that it’s a very uncompetitive market and so almost any betting pattern leads to large losses. So it’s also necessary to investigate the pay-off if the strategy is to bet against each of the proposed moves instead.

In this instance, I calculated the implied likelihood of each transfer not being completed (for example, 33% for Coutinho). I then worked out the odds that implied (2/1 for Coutinho). Finally I then halved those odds, to make them less attractive and reflect that betting firms aim to make a profit. In the table below I rounded the odds to make them easier to show**.

Interestingly, this strategy would have led to a profit of over 50% of the initial stake. Overall, the evidence supports my friend’s suggestion that transfer rumours can lead to unreasonably short odds on players moving clubs, as the availability of all those stories causes people to over-estimate their likelihood.

So in this instance, the availability bias contradicts the famous investment adage of “buy the rumour, sell the fact”, as those interested in transfer outcomes would appear to have been better off selling the rumour.

* There were 18 on the list, but two related to the same player so I excluded one

** The sum of implied likelihoods  for each player moving or not is greater than one. This is because it’s assumed betting firms aim to make a profit on all bets.