Bayes Forecast is a Spanish/Brazilian company working since 1991 to provide solutions that are able to analyze the massive amounts of information available to our customers and generate axplanatory models in order to make automatic forecasts of the key business variables.
They have recently developed a forecast for the World Cup at
BayesForecast for World Cup Group C
In the above table England is predicted to beat USA, with 58.9% chance in the first round.
revised: In the final, BayesForecast predicts Brazil beating Spain
Here is the excerpt from a technical paper which describes how the model was derived.
2 Sport strength model
The sport strength of a team in a certain moment of time represents its aptitude to score goals. Hence, the score of a match depends on the strengths of the teams confronted. The following text is applied to football, the prevailing sport in most of European and South American countries. However, adapting this model to other sports like basketball, tennis, golf and so on would be easy.
2.1 Model assumptions
The model proposed is based on the next assumptions:
- It takes into account only the sport results obtained by teams at the past.
- A team's strength is unique for all the tournaments and rounds it takes part.
- Strength is dynamic and adaptative, as it depends on the past results and gives a greater stress to more recent scores.
- The strength system of a set$CDIR/htmake.pl $DIR_NROOT.html of teams is a zero-sum game.
- Strength has a metrics: it represents the expected difference of goals that a team should obtain in a match played in a neutral field against a zero-strength opponent.
- The model has two general parameters, representing the field advantage of a team when playing at home and the innovation the score introduces in the strengths of the confronted teams. Both are constant.
- A match score modifies the strengths of the teams in a linear way: each additional goal makes strength increase in the same quantity.
Former assumptions constitute a set of limitations and restrictions leading the following criticisms to the model:
- It doesn't take into account other objective factors that may affect the strength of a team, such as: the players participating in the match, the annual club budget, the number of spectators attending the stadium, meteorology, etc.