Testing football predictions.
An early data analysis project comparing Elo-based football predictions with simpler baselines, with an honest look at what did and didn’t work.
The experiment
An early project using historical English football data to compare match prediction approaches. The analysis explores whether an Elo rating system could outperform simple strategies such as choosing the favourite or the home team.
The implementation
Python scripts process CSV files, calculate baseline strategies, implement an Elo system with a draw factor, and test predictions. Matplotlib is used for visualisation.
What I learnt
As documented in the repository, the project did not produce a profitable betting strategy. Its value was as an early learning exercise in handling real data, comparing models and recognising limitations in statistical methodology and evaluation.