Innovation: Academics create machine which predicts match results
Statisticians at the University of Salford have combined their love of stats and football to come up with a machine which predicts match results. Professor Ian McHale explains how the sports analytics machine could impact the game
It’s official, Chelsea 2005-6 is the best English football team of all time. So says SAM, the sports analytics machine created by the University of Salford. As well as putting an end to many arguments (well, maybe not), SAM is apparently able to predict the results of football matches, inform managers about which players they should buy, look at how buying a player could impact the season, as well as retrospectively look at the impact an “unbought” player could have had.
“If Manchester City had bought Paul Pogba in the January transfer window, they were expected to accumulate four extra points during the season which would have guaranteed a Champions League place,” says Ian McHale – one of the team which created SAM.
Premier commission
The University of Salford started research in this area about 10 years ago, when the Premier League contacted them about making an objective ratings system for players. “They wanted a system which wasn’t based on subjective opinion, but on facts,” says McHale. “We came up with an algorithm which calculates the contribution a player makes to the team.”
It became the official ratings system of the Premier League and is still used by the league today, including on its fantasy football website.
“Until that point we didn’t realise there was such a demand for the type of things we can do,” says McHale. “In the past five or six years so much more data has become available, regarding co-ordinates of players running around the pitch and details of every event which happens during the match, so our interest and success has come on the back of this data.”
SAM takes into account multiple factors: recent results, which team is at home, the strength of teams both sides have played, as well as the quality and the form of the players on the pitch.
And it is proving to be more accurate than the bookmakers’ systems. “Bookmakers use a team-based model, using data and then they’ll hand it to a trader who will make subjective adjustments, based on injured players or if the team is not performing so well,” says McHale. “But SAM is a player-based model, it knows exactly who is playing and so there is no need for subjective adjustments. It’s the next generation of predictive models.”
The Salford team has worked with bookmakers to build models for various sports, including golf, cricket, tennis and American football, but McHale says his interests lie with football, not betting.
“The predictions are a side product, it’s the player ratings which are really interesting, because they allow a club to put a player into a team and simulate an entire season to see what would happen,” says McHale. Selling this service to football clubs is a potential commercial opportunity and would lead to clubs being more scientific in their selections,” he adds.
The machine can also predict the trajectory of a player’s career path: “If a player is reliant on speed, like Raheem Sterling, we predict he will peak earlier than a player who doesn’t.”
The team is also planning to study managerial decision making. “We can assess how the decisions managers make affect the outcome of a match, such as when they make substitutions,” says McHale. “For example, manager José Mourinho has a reputation as a great tactician, sometimes making substitutions early on, and we propose to find out whether his reputation is really deserved.”
Using viewing figures from Sky, the team is currently collaborating with the University of Liverpool to try and establish what fans like about football. “A lot of people think the reason people like football is that it’s competitively balanced and there’s lots of uncertainty in outcome,” says McHale. “We want to find out if that’s true, or if it’s down to other factors, like the number of goals, or entertainment value.”
Although the algorithms are geared to football, SAM could be adapted to other sports, particularly cricket. McHale says that if an athlete changed their training regime, SAM could assess its impact on performance.
McHale hopes SAM will be used to help football clubs spend their budgets better, which will ultimately improve the game.
“It annoys me when football clubs spend ridiculous amounts of money, which effectively comes from the fans, without using enough science to judge, rate and ultimately buy players,” he says. SAM isn’t just for the Premier League either: “At Championship level, there is the real possibility of gaining a huge advantage over the opposition by identifying and recruiting players.”
Recreation Assistant (Dry Site)
Party Leader
Cleaning Assistant
Duty Manager
Duty Manager
Team Leader (Harrow School Fitness Club)
Centre Manager (Leisure)
Director of Operations
Fitness Motivator
Recreation Assistant/Lifeguard (NPLQ required)
Membership Manager
Recreation Assistant
Swim Teacher
Swim Teacher
Chief Executive Officer, Mount Batten Centre
Swimming Teacher
Swimming Teacher
Company profile
Featured Supplier
Property & Tenders
Company: Knight Frank
Company: Belvoir Castle
Company: AVISON YOUNG
Company: London Borough of Bexley
Company: Forestry England