Extending the Dixon and Coles model: an application to women’s football data

A summary of the research:

The article "Extending the Dixon and Coles Model: An Application to Women’s Football Data" by Michels, Ötting, and Karlis explores statistical methods to model goal distributions in women’s football. It addresses limitations in the widely used Dixon and Coles model, which was developed for men’s football and struggles to account for key differences in women’s matches, such as overrepresentation of scores like 2-0 and 3-0. The authors propose an extended model based on the Sarmanov family of distributions, which allows for flexible adjustments in probabilities across a broader range of scores and introduces negative binomial distributions to better handle the observed overdispersion in women’s football data. The study applies these models to match results from four major European women’s leagues (England, Germany, France, and Spain), demonstrating improved fit and predictive accuracy compared to traditional approaches.

The research highlights the need for gender-specific statistical modeling in football to capture unique patterns in women’s matches, such as fewer 0-0 draws and stronger correlations between home and away goals. The proposed models show promise for more accurate prediction and analysis, paving the way for better insights into women’s football dynamics. These advancements can enhance applications like performance evaluation, match predictions, and strategic planning for teams and leagues.

Key Takeaways

1. The proposed model addresses key differences in women’s football, such as overrepresentation of scores like 2-0 and 3-0, and fewer 0-0 draws.

2. The use of negative binomial distributions and the Sarmanov family improves the fit and predictive accuracy of goal distributions compared to traditional models.

3. Gender-specific modeling is essential for capturing the unique statistical patterns in women’s football, offering better tools for analysis and strategy.

Authors: Rouven Michels, Marius Otting, Dimitris Karlis

You can read the entire article here.