The FIFA World Cup is one of the most prestigious tournament all over the world and hence there is major interest, among fans and experts alike, in forecasting the winner of this tournament. To investigate this issue, a class of linear mixed-effects models for quoted winning odds from various bookmakers is explored. Based on this ``prospective" data reflecting the expectations of the bookmakers (as opposed to past performances used in many other forecasting methods) different models for the ``true" odds of winning the tournament can be established, capturing both team-specific effects (along with effects for the team's tournament group and continental confederation) and bookmaker-specific variations. A selection among various model specifications yields a model with a fixed team effect plus a random bookmaker-specific deviation. It forecasts team Spain with a probability of 17.86% as the winner of the tournament; the second best team is Brazil with a winning probability of 15.27%. In addition to the forecast of the winning probability, information about the groups of the preliminaries and the different continental confederations can be obtained from the model.
|Publication status||Published - 1 Jun 2010|
|Name||Research Report Series / Department of Statistics and Mathematics|
- Research Report Series / Department of Statistics and Mathematics