Skip to content

Tanner-C0DE/pl-predictive-model

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

pl-predictive-model

Premier League Heuristic Scoring Engine

This project implements a custom-weighted "Grid Model" to predict English Premier League match outcomes. Unlike "black-box" machine learning models, this engine uses explicit mathematical weights assigned to key performance indicators, providing total transparency into why a specific team is favored.

The Logic: The model calculates a Match Strength Score for each team based on:

Points Per Game (PPG): Base strength of the season.

Momentum (Form): A rolling average of the last 5 matches.

Fatigue (Rest Days): Penalties applied for short turnaround times between games.

Shot Efficiency: A calculated ratio of Shots on Target vs. Total Shots.

Disciplinary Record: Negative weights applied for yellow and red card accumulation.

Home Advantage: A flat +0.45 boost to the home side to account for historical variance.

Key Features: Custom Dashboard: Outputs a side-by-side "Match Analysis" grid comparing the two teams across all factors.

Built-in Backtester: Includes a script to run the model against completed fixtures to calculate historical accuracy.

Zero Warm-up: Uses vectorized pandas operations to calculate stats on the fly.

Sample Output:

Model Accuracy Check: 50.34% Enter home team: Aston Villa Enter away team: Nott'm Forest Enter date (MM/DD/YY): 1/3/26

=== MATCH ANALYSIS: Aston Villa vs Nott'm Forest (1/3/26) === FACTOR | Aston Villa | Nott'm Forest

Strength | 0.411 | 0.189 Form | 0.96 | 0.24 Matchup Gap | 0.442 | -0.442 Rest | 0.4 | 0.4 Efficiency | 0.223 | 0.153 Penalty (Stam/Disc) | -0.35 | -0.3 Home Adv | 0.45 | 0.0

FINAL MODEL SCORE | 2.536 | 0.240

PREDICTION: Aston Villa Wins! <<<

Data was collected from https://www.football-data.co.uk/englandm.php

About

A weighted scoring engine that predicts 2025/26 Premier League outcomes using domain-specific heuristics. This project focuses on "glass-box" logic, where every prediction can be manually traced back to specific team metrics like shot efficiency, rest days, and home advantage.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages