The Premier League started this weekend bringing the English soccer season into full swing. For those that don't know, the Football Association; presumably under pressure from the sport's governing body, Coca Cola, have decided to rename the remaining leagues. Last year's League Divisions 1, 2 and 3, formerly known as Divisions 2, 3 and 4, have cleverly been renamed: The Championship, League One and League Two.
Anyway confusing name changes aside, the start of the new season requires me to dust off that old favourite project, the Football Predictor. Now it doesn't really start to come into its own for about 10 weeks when it has some form to work with, so I have plans to enhance it in the meantime.
This year I aim to add a module that predicts the score of a fixture. Some of the statistics for each team's score patterns can be quite remarkable - these are a snapshot after about 30 matches last season:
- Arsenal to win 2 - 1 at home (40 % of home matches)
- Leicester to draw 1 - 1 at home (31.25 % of home matches)
- Manchester United to win 2 - 1 away (26.67 % of away matches)
The prediction process is two fold. Firstly making predictions (obviously) but secondly analysing how the information is interpreted to see if it can improve the quality of the predictions. Last year I added a system where the project also rated the 'chance of success' of the prediction. This worked okay, but I had to stop the project from trying to have a 100% success rate. At one point I did end up where it would have only made 2 or 3 predictions in the season and these were all Arsenal to win