Analyzing the impact of various statistics on the outcome of football matches

Abstract

Association Football, played in over 200 countries by 250 million+ professional players and followed by 1.3 billion people, is the world’s most popular sport in terms of participants and fan-following . Football has intertwined itself into the lives of millions of people around the world and is a way out of poverty for millions of youth in South America Africa and other such poor nations. Europe is the major market for football in terms of the overall revenue and quality of football at the highest professional level as the five biggest leagues namely England’s Premier League, France’s Ligue 1, the German Bundesliga, Italy’s Serie A and La Liga in Spain dominate the club football scene. European football market is estimated to have generated €30 billion in revenues in 2016-17 . In such a game with high stakes, analytics is playing a crucial role in improving the players’ fitness, athleticism, and overall skill whereas the managers make better decisions and the owners cut better deals. Clubs have realized the potential of analytics as evident from Arsenal’s £2 million buyout of statDNA, a US-based analytics company. Arsenal’s manager Arsene Wenger and Borussia Dortmund’s ex-manager Thomas Tuchel have publicly referred to ‘xG’ (expected goals), a key measure popular in sports betting and analytics field of how often a team was statistically likely to score goals . Most Premier League football stadiums are equipped with a set of 8-10 digital cameras that track every player on the pitch. Ten data points are collected every second for each of the 22 players on the pitch, generating more than 1.4 million data points per game . The managers and performance analysts use these data to analyse the performance of the team through every pass, tackle, shot, etc and gain insights about every moment on and off the ball. The team statistics that are collected during the game such as total shots, passes, tackles, corners, etc. could serve as an important tool in predicting the outcome of the match. The betting industry uses many such parameters in their models to make their prediction about a match and a similar analysis of such in-game data is carried out in this project to make predictions about the outcomes of matches. Objective is to analyse the in-game statistics data such as total shots, passes, tackles, cards received etc to build a model to predict the outcomes of the games namely a win, a loss, or a draw for the home/away team. The number of goals scored by a team (Home/Away) could also be predicted using the data mentioned above.