Below are the probabilities estimated from a 9,999 iteration run of the Isthmus Partners World Cup Simulator, using the Model Weightings 100% and without any results hardcoded for any game. Brazil is most likely to win the World Cup with an 18.8%, followed by Spain with an 18.2%, and third Italy with a 9.5% chance. Ghana, Honduras, Japan, Korea DPR, New Zealand and South Africa have neglible chances of winning.
Posts
Great simulator guys. I've built my own simulator using rankings based on ESPN's Soccer Power Index, but I get some different results, for example I have Brazil winning about 30% of the time. My simulator is written in Javascript and can be found at:
ReplyDeletehttp://www.doyledevelopment.com/world_cup_simulator.html
Javier said...
ReplyDeleteMatt, thanks for the comment and for the link to your simulator, it looks quide good.
I like the ability in your simulator to be able to run one simulation at a time and see all the results of one World Cup simulation all the way to the final.
Many thanks
I would like to see how this correlates with current odds found here: http://www.worldcupodds.eu/outright-betting
ReplyDeleteThey have Spain on top, for example. You guys have England at 8.7% and Argentina at 5.3%. I am not sure what the deviation is, but probably this is quite large. However, the bookies are giving them equal odds at the moment.
This comment has been removed by the author.
ReplyDeleteSukhdev said...
ReplyDeleteAnonymous, we've been asked this a lot, and there is some correlation, though we are not using the model for speculative purposes.
There is likely to be some variation here, but standard deviation probably isn't the best measure here, as the results are not normally distributed. The results are largely skewed by the group stages. If you look at the results for the knockout rounds only, the standard deviation will more likely fit though.
On how much variation, I would say the model is unlikely to provide a closer fit with more iteration. It is more likely that the results need calibrating through the user inputs and the model and user input weightings.
Javier said......
ReplyDeleteDear Anonymous, the model is very sensitive to the FIFA rankings, and in the March rankings England was ahead of Argentina. With the May rankings that would change as Argentina is now ahead of England.
Interesting to see that in our model the direct England vs Argentina distribution of results are weighed heavily in England's favour, which I suspect also reflects the impact of historical head-to-head results in the estimation of distributions (in spite of our recollection of 1986, England beat Argentina in 1962, 1966, 2002 and the 1998 game was a tie).
You can adjust the model by modifying the user input range and providing a weight to user inputs. Use the head-to-head statistics to get to distributions you are comfortable with.