Premier Predictions


In the final chapter of my book, Soccermatics, I test various models for betting on football. Each week, for the next four or five weeks, I will provide a list of football bets on the English Premier League. I will place these bets with various bookmakers with a starting capital of £400.

I have set myself a few rules. First of all, I am going to use mathematical betting strategies. I set up models in advance, based on data from matches and the bookmakers odds, and then use these models to make predictions. The models are all in place before the betting period begins, and although I will tune them slightly I will stick to them throughout. The second rule is that all the models can be built at home by an amateur enthusiast. I’ll qualify that statement a bit. The mathematics and statistics involved is similar to that used in undergraduate economics, physics and engineering courses. So a certain skill level is required (I have had some help from Robin, Daniel and Jan-Erik at OddsCraft). But I have programmed all the models myself in Matlab, which took about 3 or 4 weeks. Anyone with knowledge of Matlab, R, Python or a similar language should be able to implement the steps I take here. The third rule is that all the data used in the models are easily accessible online. There is an amazing amount of match data available, from shot statistics to team ranking indices, and there are a large number of odds comparison sites which include historical records. I only use these sources as input to the models, and not any of any private data collected by clubs and companies.

To find out the full details of the models you will have to buy the book when it comes out in May 2016. But I will say this. My bets will come from a combination of four models.

Odds bias: For this model I have downloaded odds from last year and looked for consistent biases, such as overbetting the favourite or the outsider, as well as patterns in draws etc. I have then fitted a statistical model to pick up these biases.

Euro Club index:  Here I have used the Euro Club Index to make predictions. I have made a small correction to the index, based on my calculation of draw probabilities, but generally take their predictions for each match.

Expected goals: One of the hotest area in football blogging research is expected goals. Here we judge a team not by the actual goals it scored, but the quality of the chances it generated. I use a simple model based on three shooting zones to calculate each teams expected goals. I also mix in statistics based on passing rates. Teams with high passing rates and more expected goals are more successful.

The Expert: I take the weekly betting tips of a media expert and convert them in to result probabilities.

I have been running these strategies for four weeks now and below I give the current profitability, assuming £100 starting capital for each method.
So far 'Odds bias' has provided a steady income, while expected goals came good when Chelsea were beaten by Crystal Palace. There is a lot of variation in the fortunes of different strategies. It will be interesting to see what happens in the long term.

Each week from now on I will combine these four strategies, weighting based on previous weeks' success and produce betting predictions for each match. I'll release my list of bets on Twitter every Saturday morning (or Friday evening if there is a match on Friday). As a twist, my wife is going to provide a 5th model. She represents the average punter and will also place bets, with a starting kitty of £100. This is going to be fun!

Disclaimer: These bets are provided purely for informational reasons related to writing the book. I would not advise any one to follow them or invest in them. This is an experiment designed to test and compare various strategies, not an attempt by me to provide a reliable betting system. Bookmakers have a built in advantage and will invariably earn money. Long term success against a single bookmaker is unlikely.