Match Outcome Simulation

This project simulates the possible outcomes of a football match using a Poisson-based model. By leveraging expected goals (xG) data, the simulation generates probabilities for different scorelines and visualizes them through a heatmap. The project also includes image processing to enhance the presentation of the results.

Features

• Extracts match data from a CSV file.
• Uses a Poisson distribution to simulate match results.
• Generates a probability matrix for different scorelines.
• Visualizes the results with a color-coded heatmap.
• Highlights the most probable outcome.
• Saves and processes the heatmap image.

How It Works

1. Simulation Function (simulation(url)):

   • Extracts match_id from the provided URL.
   • Reads the corresponding CSV file containing team data.
   • Simulates 100,000 match results based on expected goals (xG) using a Poisson distribution.
   • Creates a probability matrix for possible scorelines (limited to a maximum of 5 goals per team).
   • Visualizes the probability matrix using a heatmap.
   • Highlights the most probable outcome with a rectangle.
   • Saves the generated image with adjusted cropping and rotation.
   • Returns the probability of home win, draw, away win, and expected points for each team.

2. Image Import Function (import_img(url, ax=None)):

   • Extracts match_id from the provided URL.
   • Finds and loads the saved heatmap image.
   • Displays the image either in a new figure or within a specified subplot.

Example Output

• Home team win probability: 45.2%
• Away team win probability: 30.5%
• Draw probability: 24.3%

Additionally, a heatmap is generated where the most probable scoreline is highlighted.

Example Heatmaps

Here are some example heatmaps generated by the simulation:

1. Copa de la Liga Profesional 2024 Round 12 - Talleres 1 v 0 Velez Sarsfield

   

2. Copa de la Liga Profesional 2024 Round 12 - Boca Juniors 2 v 1 San Lorenzo

   

3. Copa de la Liga Profesional 2024 Round 11 - Argentinos Juniors 3 v 0 Rosario Central

   

Potential Improvements

• Integrate real-time match data.
• Enhance visualization aesthetics.
• Optimize performance for large datasets.
• Allow user-defined simulation parameters.