Data Analytics in Football - Defining the Playing Styles of Singapore PL Teams

From August 2022 to May 2023, I had the opportunity to experience the life of a researcher as a participant of NTU’s URECA1 program. I’ll write a full recap of my journey as an undergraduate researcher after the next academic year, but in many ways, this was the perfect project for me. It allowed me to work on a field that I have been passionate about for many years - sports analytics in football.

Working with Assoc. Prof John Komar from NIE2, we built an Expected Goals model from the last three seasons of Data in the Singapore Premier League, and applied it on top of research conducted by another URECA student to identify which styles of play are effective in Singapore, and which teams use them the most.

The full paper and my findings are below if you’re interested in taking a look at it! If there’s interest in an in-depth runthrough of the code that I wrote for it, I’ll be happy to write that up as well.

Expected Goal Model

Input Variables

  1. X Coordinate of the Shot
  2. Y Coodinate of the Shot
  3. Euclidian Distance from the Center of the Goal
  4. Angle of the Shot from the Center of the Goal

Coefficients

Variable Value
Intercept 0.2204
X Coordinate -0.0281
Y Coordinate -0.0062
Distance -0.0998
Angle -0.0081

Try clicking on different parts of the pitch below to explore how expected goals vary based on shot location:

Additional Plots

I wasn’t able to add all of these visualisations into the final paper, so I am taking this opportunity to share them here. I’ve always been impressed by the way publications like The Economist present their data, and I wanted to try to replicate that. While making these graphs, I realized that data visualization is really a labor of love, with so much fine-tuning needed in Python to get a graph looking just the way you need it to.

Expected Goals Heat Map

Expected Goals from Left Footed Shots vs Right Footed Shots

Left Footed Shots

Right Footed Shots

Expected Goal Difference vs Actual Goal Difference

2018-2019

2019-2020

2020-2021

Paper

If you’re interested in reading the full paper detailing the methodology and findings of the xG model and decision tree outputs, you can find it here.


[1] - Undergraduate Research Experience on Campus

[2] - National Institute of Education