Just five more days before the first world cup games and predictions abound as to who will win the title.
Yet, if you want to stand out from the crowd and have your prediction noticed, you need big names.
First let’s look at Goldman Sachs. In a paper by Jan Hatzius and Jari Stehn which generates more than 340,000 on google (“goldman sachs world cup predictions”) the prediction is that Brazil is going to win against Argentina in the final by 3-1.
Next is Stephen Hawking whose “formula” generates 140.000 results on google (search “stephen hawking world cup prediction”).
Professor Hawking, who admits being “slightly more qualified than Paul the Octopus” and sponsored by the bookmaker PaddyPower, recommends England play a 4-3-3 formation, with a European referee and a 3:00pm kick-off.
Finally, we have Jose Mourinho, who, in a video for Yahoo, forecasts an England – Portugal final with England taking the cup. His judgement is motivated by personal interest (“I want to be able to have a holiday in Portugal”) rather than his deep insight of the game and the form of the players going into the tournament.
The first two models rely on statistical analysis going back decades to amass enough data points (who plays whom and the final result). The problem with this approach is that most of the games taken into account in the statistical analysis are irrelevant. The 23 players of each world cup delegation have nothing to do with the players who played in the games in the 60’s, 70’s, 80’s and 90’s. For instance, all the games between Germany and England taking place more than 4 years ago have no impact on a possible England vs Germany next month. Neither Gary Lineker nor Franz Beckenbauer will be playing. It wouldn’t occur to you to take your amateur league in Uzbekistan as data to predict Colombia’s world cup performance but that is precisely what is going on. Colombia 1994 or 1998 is alien team to today’s Colombia.
So, forget the predictions, but do remember the brands: Goldman Sachs, PaddyPower and Yahoo.