Past performance of the systems

To give a bit of background about the systems I’m using this season, below shows the results that the systems have obtained in the second half of last season in the TFA forum and for backtesting back to 2006/07 season. I have also used data before this (back to 2002) when building the systems but I don't plan to include these results in any of my data going forward as the data is backfitted too much in these early years.

My rating algorithm needs at least 6 games to be played before any system bets can be generated and therefore, the first week when bets may be possible is the middle of September. Things then slowly build up from there and by mid October, all of the systems will be on the go for all of the leagues.

I concentrate my efforts on all of the 5 top leagues in England as well as the SPL in Scotland. I have looked at doing the top European Leagues and also the lower leagues but the ratings tend to not work as well I have found. Therefore, I'm happy to concentrate my efforts on these 6 leagues. I will exclude all cup games also from my analysis although I do adjust the ratings based on cup games.

The performance of all the systems before this season are shown below in a variety of tables. Each time I add a system to my portfolio, I will show the historical results so people can see how it has performed in the past.



Like most statistical models, the more results the system has, the greater the chance that the system may perform like that in future. Hence, I’m very wary of any system that has a handful of bets as there can be no guarantee that the system even has an edge yet.

I also like to look at the returns of the system by calendar month to help understand the swings that may be possible on a monthly basis. Below is the performance of all systems by month.