2

The Goldman Sachs model was already discussed here. I wonder whether there are other statistical forecasts publicly available, or even better, models including raw data. Specifically, I thought that the Goldman Sachs model didn't work so well for the 2010 cup (see the original report on page 7). I also heard that actual or estimated transfer values of players have high predictive power, but they are not included in the Goldman Sachs model.

kjetil b halvorsen
  • 77,844
Julian Schuessler
  • 2,375
  • 14
  • 17

2 Answers2

2

I like this approach: http://www.r-bloggers.com/home-victory-for-brazil-in-the-2014-fifa-world-cup/ (more formal description in the working paper here: http://econpapers.repec.org/paper/innwpaper/2014-17.htm)

The idea is to use a Bradley-Terry model where the team ability parameters are found using simulations of whole tournaments in order to match bookmaker probabilties of each team winning the tournament.

I suspect even better results could be achieved by using the implied probabilities of the Betfair market.

M. Berk
  • 2,575
  • This is by far the most comprehensive piece of research I have seen so far, thanks. I also found an earlier paper interesting where they compare their approach to other forecasts methods (Leitner, Christoph, Achim Zeileis, and Kurt Hornik. "Forecasting sports tournaments by ratings of (prob)abilities: A comparison for the EURO 2008." International Journal of Forecasting 26.3 (2010): 471-481.). – Julian Schuessler Jun 11 '14 at 15:51
1

I haven't tried it personally but there is this application, by a company name Bayesia, that is relevant to your question: http://worldcup.bayesialab.com/.

Zhubarb
  • 8,269