With the Premier League season about to begin its busy Christmas schedule, Prozone’s analysts have published the latest results of their EPL Performance Model.
A means of measuring underlying performance and stripping away the effects of random variation, the aim of the model is to provide an objective assessment of team performance detached from the simplistic indication given by league position.
How Does It Work?
Using shot data to understand the probability of scoring from any type and location of attempt, it is possible to estimate the number of goals a team would be expected to score and concede on average in any given fixture.
For example, take a hypothetical match in which there are only three shots. Team A has one penalty and Team B two penalties. In such a situation we’d expect Team A to score 0.77 goals on average and Team B 1.54 given that the league-wide penalty success rate is 77%).
Taking a long-term view, we would expect teams to match these probabilities as they are a strong indicator of underlying performance and ability. In the short-term, however, luck and random variation enables teams to perform above or below expectation. In the long run, any over or underachievement can be attributed to skill.
This season, the statistical analysts in our Performance Analysis Centre have been calculating how many points each team in the Premier League would expect to have based on their shooting performance.
The model also incorporates the caveat of Opponent Difficulty (Pts), a figure which indicates the difficulty of each team’s fixture list based on where their opponents lie in the table and adjusting for home and away results. By taking the expected points-per-game values of each team’s opponents, we also get a ranking of how difficult opponents have been based on their underlying performance data.
What Does It Tell Us?
Arranged according to Expected Position, the model shows us what the Premier League table would look like if the influence of luck and random variation was eliminated.
When we put everything together, the Premier League Performance Model provides a strong indicator of underlying performance and a measure of the extent to which individual teams have performed above or below expectations over the course of the first 15 games of the season.
As we can see, with a Residual Points total of +10, Arsenal are the team exceeding expectations by the greatest margin after 15 games. Conversely, with -8, Sunderland is the team playing the furthest below expectations so far this season.
An interesting case is that of Norwich City. Expected to sit 18th but currently 14th and with +1.8 Residual Points, the Canaries’ underlying performance is significantly better than results might suggest. Having faced the toughest set of fixtures so far according to the model, Chris Hughton’s side might expect results to pick up over the coming months.
The pioneer in sports performance analysis, Prozone has been empowering data with meaning since 1998. Working with sport’s elite, we provide expert technology and consultancy to create world-leading performance insights.
Analysing objective data from more than 10,000 match every season, we help 250 clubs across five continents to achieve genuine on-field advantage.
ROGAN TAYLOR: WHY DID SKY POKE BT WITH A BIG STICK?
ROGAN TAYLOR: YOU CAN BET YOUR BOOTS
ROGAN TAYLOR: THE GREATEST NUMBER NINE OF ALL TIME
ROGAN TAYLOR: SEEDS OF DISCONTENT
KEIR RADNEDGE: FIFAPRO PILES PRESSURE ON KAFALA SYSTEM
ROGAN TAYLOR: BRING OUT THE BEST CHINA
ROGAN TAYLOR: THE BUBBLES DIDN’T FADE AND DIE, AFTER ALL