Monday 24 September 2007

Predicting Football Results

My work involves data analysis and prediction of outcomes for a variety of customers including telecommunications: churn, financial services: up-sell and cross-sell, retailers: spend, spend, spend. So I think a lot about how you can apply data analysis to a number of different problems. In particular if there are large amounts of data then this makes it more intersting.

I have been familiar with the rec.sport.soccer.statistics.foundation for quite some time as they are regularly referenced by the one true daily tea-time football email - The Fiver. They collect an enormous amount of football statistical information and make it available on the internet. Sometimes this is just the final results for historical league and cup competitions, however for the English premiership they have the results for all games from 1992. This was when the First Division was renamed as the premiership.

I had been persuaded by some friends to join the Guardian Pick the Score competition and discovered that I was particularly bad at this. So I have decided to apply some of my knowledge of data analysis to apply it to this problem.

I haven't really done any research into this yet - I'm convinced that others will have done this already with varying degrees of success. how ever I think it should be fun so I plan to document my progress here on the blog.

No comments: