1 Million Pound Football Challenge

A while ago I downloaded some of the historical data from the Betfair site but never actually got round to doing anything with it….until now!

I’ve started to do some simple analysis of the data to see if there is anything I can make from it and this is the first in a series of hopefully useful articles.

The first data that I have analysed is some of the football data from 2005. All the data downloaded from Betfair is in raw form so it needed a bit of a clean up beforehand. Of the cleaned data, I managed to analyse about 1000 matches.

This test compared the Betfair Starting Price (SP) of football matches against the actual odds or probabilities. The Betfair Starting Price I measured as the last price matched before kick off at Betfair and the actual odds is the calculated probability of a win at that price based on the sample size and final result. The difference between the two I then normalised.

The graph below shows the results:

The red trendline shows the normalised Betfair SP price. For a perfect market, i.e. one where the Betfair starting price is the exact true price, the red trendline will be a horizontal line on the y axis at position 1 (y=1). When the red trendline is above the y=1 line then it is a layers market, below the y=1 line a backers market.

The conclusion from the graph is that the Betfair SP for the football matches analysed is very close to a perfect market. There are variations, but these are generally of a small order of just a few percent. In addition, you have to take in account commission at Betfair, which means in the long run the chance is that if you are just betting at a certain price the commission will result in a loss with time.

However, there is value in the market for the punter in that the price is a true probability of the event, and to be profitable, in theory only (easier said than done) requires selective betting (either backing or laying) of events.