Predicting 2-way football results by means of data mining

Abstract

In the last decade, has been found an increase in the number of bookmakers, particularly in the online market (ebusiness). It is possible deducing that this activity is profitable for them and consequently damaging to their users. Nowadays, football is considered one of the most popular sports. Regarding the betting world it was acquired an outstanding position, which moves millions of euros during the period of a single football match. The lack of profitability of football betting website users has been stressed as a problem. In accordance with the stated arises here, an opportunity to explore. This lack gave origin to this research proposal, which is going to address the possibility of existing a way to support the users on their online bets, in order to improve their results and profitability. A football match could be analysed from the perspective of several types of statistical data, which do not have a direct influence on the final match result. This research work has the aim of helping to improve the performance of online football bets, by providing users statistical data that may be important to take into account, at the time of doing their own bets. In this work it was possible introduce data mining models which are able to predict 2-way results (home team win/draw or visitor team win) with 96,2 % of sensitivity and a good level of accuracy (74.8%). These models are prepared to be the base of an Intelligent System.