On 5 June 2018, the RSS Merseyside local group hosted an event based around predictions in sport. The meeting was well attended with 25 present, of whom half were members of the RSS. Attendees had travelled from a variety of places including the University of York and University of Manchester. From the University of Liverpool, we attracted attendees from biostatistics, mathematical sciences, the business school and the Wolfson Centre for Personalised Medicine.
The meeting began with a talk by Tom Hughes, senior physiotherapist at Manchester United. Tom gave a fascinating insight into how data is used by professional sports teams in order to assess risk of injury and to set healthy benchmarks to be aimed at by players returning from injury. It was interesting to see that clubs collect a vast amount of data through regular screening (periodic health examinations) of players, but Tom pointed out that much of this data is relatively underutilised. In addition, he highlighted that many of the tools used to gain this data suffer from very high variability and so offer limited use in terms of prognostic modelling to determine factors that influence risk of injury.
Tom suggested that although this data is probably unable to assess causes of disease it could provide useful prognostic factors related to the risk of disease. He described some of his PhD work aiming to build prognostic models for this purpose. Given that it seems there is a very low evidence base in terms of elite athlete injury prognosis, Tom’s work aims to fill some of this gap and make better use of the huge amounts of data available to the top clubs. Tom’s talk was followed by a lively discussion with a very evident interest in the work he was doing.
Following a break for refreshments, our second talk was given by Professor Ian McHale of the University of Liverpool. Ian has been involved in developing the SAM tool used to predict matches by BBC sport so there was a keen interest in his work. His talk focused on how his group have built models using player ratings to predict the outcome of games. These models were then used to assess the impact of player signings for various top clubs. Interestingly, his models suggested that Manchester United’s signing of Zlatan Ibrahimovic was more beneficial to the club than that of Paul Pogba in the same summer. Whilst the huge transfer fee for Pogba compared to the free transfer of Ibrahimovic might suggest otherwise, many football fans having viewed the 2016-17 season might be tempted to agree!
Ian presented a simulated study of a proposed European super league, based on the player ratings models. Whilst it was disappointing that Liverpool did not come out on top of this league(!), most present agreed that the results table looked pretty realistic. Once again, Ian’s talk was followed by enthusiastic discussion.
Overall, it was a really interesting meeting that highlighted how statistics can be used intelligently to positively impact professional football, both in terms of preventing injury and in terms of identifying potential transfer targets. Sadly, I have to report that this Liverpool fan was unable to persuade Prof McHale to feed Manchester United faulty transfer targets from his model!
Our next meeting will be held on Wednesday 17 October, as part of RSS’s Members’ Week, and will explore Machine Learning topics, with confirmed speakers Professors Chris Williams and Simon Maskell, and also Juhi Gupta. Check out our website for more details: https://sites.google.com/site/rssmerseyside/