Where football and maths come together

For many people betting is part of the total football experience. Getting to the bookies on a Friday or logging into one of the many betting websites has become a routine, as is the constant disappointment when you can already throw your betslip away after the first game. Predictions are made based on gut feelings, hope or justrandompresumptions.

Of course, to be well prepared one could scour through websites and newspapers, re-watch entire matches or visit the training grounds to find out more about the tactics. Luckily there are mathematicians bringing this to the next level by creating algorithms to predict the probability of outcomes.

Football prediction APIs

Obviously, we all know that in football anything can happen. A red card or an early goal can derail the entire match and a ball an inch to the left can be the difference between winning or losing a final. But taking away these outliers, there’s a lot that can be predicted based on historical data. Think of the amount of goals being scored in a game, the chance of a team winning or whether both teams will score. With complex algorithms mathematicians have worked out a way to use data to quite narrowly predict these outcomes.

So if you own a website with betting tips or if you’ve created your own fantasy football league, such an API can be very useful. But how would that look for you, you might ask? Well, we are happy to explain this for you!

What does it entail?
A prediction API will give you all the necessary data to help your visitors predict the outcome of both games and leagues. This comes with not just the prediction of the algorithm, but also the quality of previous predictions. If you look at the prediction Football API by Sportmonks for instance, they supply you with a hit ratio, log loss, predictability of the league and predictive power of the league.

  • Hit ratio: the hit ratio is the percentage of correct predictions of the outcome over the last 100 matches of the league. With the perfect score obviously being 1.00 (100%);
  • Log loss: the log loss is basically the score of the prediction. If the algorithm sets a probability of 80% and the outcome is different, this has a negative effect on the log loss. This number is calculated on the last 100 matches as well.
  • Predictability: this is as straightforward as it gets, the predictability of a league is described in words, ranging from poor to high predictability.
  • Predictive power: the predictive power means if the league is getting more predictable, based on the log loss.