Sports modelling research

Dr. Alun Owen discusses the agenda for the Mathsport International Conference, to be held in June 2015, and highlights some of the most interesting aspects of the schedule.

Categories: All Sports, Events, Professional, Statistical models

As you may know from my previous postings, I lecture in statistics in Higher Education and have a research interest in the development of predictive models in sport, including football and horse racing. Well, this time of year is conference season in my line of work, and so this month I thought I’d highlight a conference I’m co-organising at Loughborough University which has a few talks which may of interest to some who subscribe to Sports Trading Network.

I’m also presenting at this conference on some work I’ve been engaged with looking at predicting the probability of scoring each penalty in penalty shoot-out competitions in major international football competitions. This conference is MathSport International 2015 which will be hosted at Loughborough University (UK), from the 29th June – 1st July 2015. This is the 5th conference in Europe in the biennial series that brings together Mathematics and Sport and the website for the conference is here.

There are a range of talks planned and includes four invited keynote speakers. The first of our invited speakers is Dr Ian McHale (Reader in Statistics at the University of Manchester), who has research interests which include statistics in sports and the analysis of various issues related to gambling and gambling markets. Dr McHale has also published papers on a wide variety of topics in these fields, including Kelly betting, player ratings in football, a modified Duckworth-Lewis method in cricket, time-varying ratings models and forecasting. His talk will describe a model for predicting the results of international football matches which allows for team strengths to vary over time, and reveals which team might claim to be the “all-time greatest”! He will also report on factors which his research suggests affects medal share in the Olympic Games.

Secondly, Mr David Kendix (Actuary and statistician for the International Cricket Council) will consider the potential but also the practical limitations of building models to predict sporting outcomes. In particular he will consider the extent to which techniques used previously in financial modelling can now be applied to sport. His talk will compare financial models with sports models, looking at the use of past data (results, times, distances) and the relative roles played by expert judgement, as well as factors such as motivation, confidence and team spirit.

Thirdly, we have Professor Tony Purnell (Head of Technology for British Cycling and founder of Pi Research developing race car analysis tools), who will argue that sports in general lend themselves well to mathematical modelling, whether using statistical methods or physics modelling approaches. His talk will consider how human beings are not so easily modelled and so the output of a mathematician’s or engineer’s work is more likely to be used to improve a coach’s or an athlete’s intuition rather than being of direct value.

Finally, Professor Sigrid Knust (Professor of Computer Science at the University of Osnabrück) will discuss different methods to construct a schedule for a sports league which are based on an area of mathematics called Graph Theory. Professor Knust will consider the basic problem of finding a schedule for a single or double round robin tournament, in which every team plays against each other team exactly once (or twice), and every team plays one game per round. Additionally, side constraints will also be highlighted e.g. the avoidance of breaks (consecutive home or away games of a team) and fairness issues (like opponent strengths, carry-over effects).

Further details of these four keynote talks can be found here.

In addition to the above keynotes, we also have over 70 further contributed talks, many of which include papers in the conference proceedings, for those attending the conference to take away with them. A full (draft) list of talks can be found here  whilst a detailed (draft) schedule of when these talks are planned can be found here.

I’ve picked out a few of these talks that I thought might be of interest to you in what follows below, but please note some of the scheduled timings are subject to change so for the latest information please do refer to the above web links.

The morning of Monday 29th June (Day 1 of the conference) starts with a session on prediction models. This includes two talks which look at predictive modelling in t20 Big Bash cricket, basketball, and two further talks on predictive modelling in football.

Tuesday 30th June (Day 2) sees a further session on prediction models in tennis that I’m looking forward to as much as the Monday session. This includes a talk comparing the performance of a range of different predictive models that have been published, another looking at the use of tennis rankings to predict performance and a third on regression models in tennis and volleyball with applications to gambling strategies.

That first Tuesday session is followed after coffee by another which includes three further talks, all on models in football. These include modelling the time to the first goal in football, the use of COM-Poisson models, with the final talk by Dimitris Karlis (who developed the R package called bivpos which fits a bivariate Poisson model to football and other applications) on robust modelling in football. Again I hope to be at this session.

The afternoon of the Tuesday sees the first of several sessions by Simon Gleave, Head of Analysis at Infostrada Sports. In this session Simon will give an introduction to the Infostrada Sports database, which will be followed up the following day with a longer one-hour interactive session on what the Infostrada Podium data tool can offer.

Wednesday 1st July (Day 3) kicks off with a session devoted to betting. This includes talks on applications to roulette, optimizing market margins in football, betting in PGA golf, and finally a talk which has the interesting title of “A Safe bet – or is it?” by Dominic Cortis. Dominic has proposed a potentially fascinating talk which looks at the key requirements for odds set by bookmakers.

The final set of talks which I will highlight includes that in the afternoon on the Wednesday. This comprises three talks on prediction in sport, including two which look at predicting athletic performance and a third which looks at prediction in hockey. Again, as I said earlier please note some of the above scheduled timings are subject to change so for the latest information please do refer to the above web links. The conference starts on the Sunday evening with an informal gathering for drinks, networking and catching up with old friends, in the bar at Burleigh Court on the Loughborough University campus.

The Monday evening then sees the Sports Quiz, which this year is being sponsored by Sports Trading Network who are providing drinks for all those coming along and also some quite significant prizes for the winning teams. The Tuesday evening includes a tour of the very extensive sports training facilities across the Loughborough University campus, followed by the conference dinner which is being held at Trent Bridge cricket ground in Nottingham (transport is provided).

I would be delighted if you were able to join us at the conference and details of how to register can be found here, but please note that the deadline to register is 12th June! We can offer a one-day rate but this is not too much cheaper than the full conference rate as we need to cover our fixed costs, but if you wish to discuss options such as this can you email in the first instance? If you are able to register and join us then do please introduce yourself to me! I hope you found this information useful? Do please get in touch if you wish to discuss anything with me, especially if it relates to modelling football or horse racing, or trading and I hope to meet you at the conference?