Ok guys, its one more day to go before it begins...
I've updated the previos model with the current set of teams and statistics.
As previously, you can run a 1000 tournaments within seconds. This time I also shifted the focus from the supporters to the actual players.
As before, there are no references or images of the actual event since they are trademarked, hence the name Soccer Cup.
I attach a movie I used at a customer event, the model and an updated help file.
I get an error message when trying to attach the model to the post:
The contents of the attachment doesn't match its file type.
What am I doing wrong?
Nice idea to do this. I had an Excel model to do something along these lines for WC-2010.
I have downloaded your model to look at the correlation between raw betting odds and final results, as I used a non-stochastic model straight from the odds before.
This just came in on the wires - it's quite a curious state of affairs :-
Get your calculators out, because today's group matches could get complicated. In Group E , we're in the weird and wonderful scenario where a team who has not picked up a single point at the tournament still has a chance of qualifying while defending champions Germany could be consigned to an early exit. In Group F, Brazil are also on the brink of a similar fate.
I only use the odds as a very basic indicator on how good the teams are. It is not entirely true, since the betting odds already also consider the groups they play in and probably other factors, like popularity and how many bets there are etc.
I re-ran the model with new odds just after Germany lost the first game and their odds went from 4 to 8 and it as a dramatic change.
The odds of Germany beating South Korea however are probably good. A lot better than Sweden beating Mexico.
Thanks - I will be interested to look at what you did.
Did you give any thought to using all the Monte Carlo run results to estimate confidence intervals for each probability of winning the competition?