Wednesday, July 9, 2014

Don't make predictions

Before commenting on predictions and the world cup, it is healthy to stress the obvious: as John Carlin says in his twitter today, it is a bad morning to wake up in Brazil, but it is much worse in Gaza: Israel puts the German massacre in perspective. My apologies to the reader.
Nate Silver, the author of "The Signal and the Noise" had correctly predicted the result of the last US presidential election in all 50 states. That gave him confidence to expand his almost "solo" web page in the New York Times "five thirty eight" into an industrial project where he hired lots of full time experts and was sold to the ESPN network. Now the whole enterprise looks more like an example of over-confidence bias.
Silver and his expanded project are devoting a lot of effort and space to the soccer World Cup in Brazil, with plenty of interesting material. The main message of their predictions, though, was that Brazil were the favourite, much more than what betting markets were predicting. Even yesterday, Silver argued that the probability of Brazil beating Germany was 65% percent, accounting for the injury of Neymar and the suspension of Silva. After the 1-7 defeat of Brazil, the famous statistician has reacted to his own failure with this interesting piece.
It turns out that soccer is more difficult to predict than electoral politics. Political events in general are also difficult to predict: the Great War, the fall of the Berlin Wall, and others, were hardly predicted by anyone some months in advance. To avoid bombings and disasters, better spend your time mobilizing against them  than trying to predict them.

No comments:

Post a Comment