There are lots of football data about.
StatsBomb is previewing the English Premier League season over the next fortnight. Their preview started with Manchester City (link).
On 30 July (link), BT Sport announced that it has given ‘big data’ the task of “comprehensively plotting out the events of … another roller coaster EPL football season”.
The announcement included the news that BT has “worked with data scientists, real-time sports analysts Squawka and Opta to predict the placements of the league’s 20 clubs, the top goal scorers, assists and more using historical data”. A sixty-one page account of this analysis is available online (link).
There is an account of the methodology BT used to develop the analysis (link). The approach includes:
- Attacking strength (goals)
- Defensive solidarity (goals conceded)
- The probability of a score line for all games
- Assist makers in each game
- Simulation of random events (including injuries and transfers)
- Data visualisation
This model will become dynamic during the season when BT combines with Google Cloud to ingest real-time data (link).
Werner Dubitzky and his colleagues (2019) have looked at a variety of approaches to football match prediction (link). Robert Rein and Daniel Memmert (2016) examined the use of big data and tactical analysis in football (link). Their papers provide some helpful background to Statsbomb’s and BT’s look at the 2019-2020 English Premier League season.
It will be very interesting to learn how approaches to machine learning affect the coaching and performance process in the EPL. Hugo Sarmento and his colleagues (2018) concluded (link):
It is evident that the complexity engendered during performance in competitive soccer requires an integrated approach that considers multiple aspects. A challenge for researchers is to align these new measures with the needs of the coaches through a more integrated relationship between coaches and researchers, to produce practical and usable information that improves player performance and coach activity.