ESPN’s™ Luck Index

Jan 1, 0001

In 2018 ESPN commissioned me to develop and run a predictive model of football scores to ascertain the impact of luck on the Premier League.

The project had four objectives:

  1. Consultation: Engage in an in-depth consultation period with ESPN to arrive at factors that can be deemed as good or bad luck (e.g., deflected goals, erroneous decisions, etc).

  2. Data coding: Train student coders to view Premier League footage for the luck factors and enter this into a database.

  3. Data analysis: For each game identified as having a luck incident, model the expected outcome from certain parameters associated with the game and teams.

  4. League table: Redraw Premier League table and derive a luck index from this analysis.

A Bayesian hierarchal model was developed to facilitate these objectives. Data were collected via a combination of: (a) data coding for freely available Premier league footage; and (b) established constants for home advantage, team strength, red cards, and penalty conversion. The collected data were analyzed using programming language in the R STAN package.

The 2019 luck index is avalaible here: https://www.espn.co.uk/football/english-premier-league/story/3991359/luck-index-2019-20-arsenal-the-unluckiest-team-in-the-premier-league

It was awarded a Campaign Media Award in 2020: https://www.campaignmediaawards.com/finalists/theres-no-luck-to-this-campaign-/

Assistant Professor of Psychological and Behavioural Science

I’m a social psychologist interested in the development of perfectionism who also happens to enjoy R programming.

Next