After a rigorous judging process, the following presentations and interactive posters have been selected to be showcased at the 2018 OptaPro Analytics Forum, which takes place on Wednesday 7th February.
Ben Torvaney, Counting across borders
Football varies from competition-to-competition in both style and substance. Intuitively, we know that scoring goals in the Norwegian Eliteserien is less impressive than scoring in the Premier League; however, this is rarely quantified.
In public analytics, little work has been done to effectively quantify these differences. Understanding the effect of different factors on player output is critical to accurately evaluating player performance.
Ben’s presentation will allow us to better understand and quantify how these differences can help lighten the load on decision makers, and allow human judgement and intuition to be deployed more effectively.
Freddie Wilson, Identifying a defence’s vulnerabilities
Using clustering techniques, Freddie’s presentation applies Opta data to investigate how certain teams and styles concede chances. The analysis will focus on the particular clusters that represent high frequency or danger for the defence.
John-Mark Sisman, An analytical approach to exploring pressing styles
The project seeks to develop a scheme for analysing teams' pressing styles and habits using tracking and event data. Teams are profiled according to how their out-of-possession movement compresses space and disrupts passing lanes, considering different areas of the pitch and as their opponents' possessions develop.
The use of these analyses in identifying pressing triggers and traps, and the implications for tactical decision-making, are also discussed.
Marek Kwiatkowski, Beyond the Baseline: individual goalscoring, shotstopping and managerial skill
Marek’s presentation will showcase a novel framework for assessing finishing and goalkeeping ability, and is provided in the context of the expected goals framework.
While team-level applications of xG are well understood, there is nothing comparably settled when it comes to player-level applications. However, with the right approach it is in fact possible to distill estimates of player skill even from relatively small samples.
Marek’s presentation will acknowledge additional context by considering players, managers and competitions simultaneously to demonstrate a thorough approach to quantifying individual contributions to scoring.
Martin Eastwood, Analysing footballers’ decisions in and around the penalty box
At the 2017 OptaPro Analytics Forum, Martin presented this poster. His analysis integrated both event and tracking data, exploring players’ decision-making in the final third.
One year on, Martin has developed and progressed this work, and will present his findings at the 2018 OptaPro Forum.
Ray Hamill, Alex Rathke and Ferdia O’Hanrahan, Developing optimal defensive strategies for corner situations
The three analysts aim to develop and present a method for identifying opposition corner strategies.
The analysis, which applies both Opta event data and Chryonhego tracking data, will look to break down various aspects of these strategies, such as marking styles, players on posts, delivery types and delivery areas.
Alongside the six presentations, five interactive posters will be displayed at the 2018 OptaPro Analytics Forum.
Ben Woolcock, Visualising the effects of game state on defensive positioning
At least publicly, football analytics has yet to fully determine how to use the data on defensive actions captured to analyse individual clubs’ tactics.
Ben’s poster will focus on an in-depth study exploring and visualising where ‘defenders’ engage with the opposition, and how this can change depending on factors such as opposition style, match venue and game-state.
David Perdomo Meza and Bobby Gardiner, Finding robust style and performance insight
Using a methodology inspired from the theory of Topic Extraction from Natural Language Processing (NLP), the focus of this poster will be on its easy-to-interpret results and applications deriving from aggregated event data that are relevant to the professional football industry.
This poster will showcase a host of ‘topics’. In this context, ‘topics’ are the underlying styles of teams and players which determine the features and actions that team/player will perform with greater frequency.
Andrew Duncan, Yiannis Papastathopoulos, Kostas Zygalakis and Gian Marco Campagnolo, Predicting playing styles at the 2018 World Cup
This study will look to predict key positional and passing characteristics at the 2018 World Cup.
Using Bayesian spatial statistical analysis, the poster will aim to identify the role of key players in moves which lead to a shot on goal as well as in defensive performance, allowing national team coaches to know months in advance the expected playing styles and key role's passing and positional features of World Cup 2018.
Kevin Shank, Evaluating throw-ins: clustering in passing sequences
The goal of this study is to gain tactical insights on making optimal decisions from throw-in situations. Using data from the English Premier League, passing sequences after throw-ins will be clustered and analysed to determine which sequences result in better scoring opportunities.
The results of this project will be visualised to discover optimal sequences across different areas of the pitch.
Mark Carey and Mladen Sormaz, Quantifying player profiles and playing styles
In recent seasons across the top European Leagues, the role of the full-back has evolved. For such a specialised position, there are now a wide variety of full-back profiles, each with differing strengths and weaknesses. Using Principal Components Analysis and other clustering analyses, we aim to identify common dimensions (skillsets) present in every type of full-back, allowing us compare players across teams and leagues.
This method provides valuable application to both player recruitment and tactical analysis of opposition teams.