Soccer Prediction Model

[mcgin]

I stumbled across this paper today.... http://www.swan.ac.uk/economics/dpapers/2003/0311.pdf It's by no means light reading, but I have to say I'm absolutly fascinated by it. It appears to describe a method for modeling soccer matches based on the following attributes:

• Home and Away teams average win ratio over the preceding 2 seasons

• Home and Away results over preceding X home matches

• Home and Away results over preceding X away matches

• Significance indicator of match for home team (Chance of promotion etc.)

• Significance indicator of match for away team (Chance of promotion etc.)

• If home or away team is still in any cups

• Distance between the two teams home grounds

• Average attendance over previous 2 seasons

• League Position over previous 2 seasons

• Head to Head results for the season.

The results claimed in the paper are a 4% return on investment over a season with flat betting (Not sure what the result would be with any sort of wagering scheme). If I'm correct I believe the paper provides enough information to implement the model as any good academic paper should, the only problem is I'm a computer scientist, not an economist so the mention of an Ordered Probit model is a bit perplexing to me :) Now I've been looking for something like this for a while, and am quite keen to see what a model like this is capable of, and with a little help extracting the exact mathematics behind the model, and access to the necessary data, I'm pretty confident that I could at least reproduce the model on a computer and we may be able to exploit it in some way. Anyone fancy having a read and letting me know your thoughts?