Do we read the betting data correctly?

Feb - 12 2014 | By

When you open a betting forum you will inevitably come across at least ten topics in which experienced punters are trying to produce a mathematical model predicting the winner in a football match. In this post I will try to express my opinion on how this is simply not possible and how football is not suitable as a sport for creating mathematical models.

The main argument I have against statistical and mathematical models for betting predictions on football matches is related to the very nature of the game. This is a sport in which often a single goal decides the match and with pure statistic it is quite difficult to predict whether any particular shot will be realized as a goal, will be saved by the goalkeeper or will hit the crossbar.

Of course, it’s easy to make statistics on what number of goal attempts are needed to score a goal (in the Premier League statistics show that around 7.5 goal attempts are needed for a goal) , but unfortunately this is not enough to make a really worthwhile betting prediction.

Whichever other item to add to this mathematical model, you again will come to the moment that it is a pure luck will а shot become a goal or not.

The same goes for tennis. We often have matches in which both players have equal number of scored points in the match, but at the end one of them win by 6-4, 6-3. How mathematics or statistics can predict which of them will be the winner?

That’s why I believe the only option in which such statistical models can be successful is when they are applied in sports, where we have many points scored. These are mostly handball and basketball. In these sports the statistics is crucial for each betting prediction and the large number of points scored reduce the element of chance in modeling the stakes.