OpenFPL-Scout-AI

Image credits: Generated with GPT-4o

OpenFPL-Scout-AI is an AI-powered Fantasy Premier League Scout that uses ensemble machine learning (Linear Regression, XGBoost, CatBoost) to predict player points and optimize FPL team selection. It features a beautiful web interface for visualizing your optimal team in a football pitch layout.

🚀 Live Demo & API Access

Web Interface: https://openfpl-scout-ai-186049008266.europe-west1.run.app

🔥 API Access via RapidAPI: Subscribe on RapidAPI Marketplace

• Free tier: 10 requests/hour
• Professional support and documentation
• Easy integration with RapidAPI headers

Features

• 🎯 AI-Powered Predictions: Ensemble ML models (Linear Regression, XGBoost, CatBoost)
• ⚽ Interactive Web UI: Beautiful pitch visualization with player cards
• 📊 Real-time Data: Live fixture and match data integration
• 🚀 Fast Performance: Async player predictions and caching
• 🏆 Smart Team Selection: Automated optimal team selection by position
• 👑 Captain Assignment: Intelligent captain/vice-captain selection
• 📱 Mobile Responsive: Works perfectly on all devices
• 📸 Screenshot Feature: Download your team lineup as PNG
• 🎨 Professional Design: FPL-themed UI with gradient backgrounds
• 🔌 RapidAPI Integration: Professional API marketplace access

Installation

Docker:

docker build -t openfpl-scout-ai .
docker run -d -p 8000:8000 --name openfpl-api openfpl-scout-ai

Usage

Web Interface

Visit the live demo or http://localhost:8000 for local development:

• Visual Team Display: See your optimal team laid out on a football pitch
• Player Cards: Detailed cards showing player stats, fixtures, and expected points
• Gameweek Selection: Navigate between different gameweeks
• Screenshot Export: Download your team lineup as a high-quality image
• Interactive Elements: Click on player cards for detailed information

API Access via RapidAPI

🔥 Primary API Access: Subscribe on RapidAPI Marketplace

Base URL: https://openfpl-api.p.rapidapi.com

Authentication:

X-RapidAPI-Key: YOUR_RAPIDAPI_KEY
X-RapidAPI-Host: openfpl-api.p.rapidapi.com

Quick Example:

const options = {
    method: 'GET',
    headers: {
        'X-RapidAPI-Key': 'YOUR_RAPIDAPI_KEY',
        'X-RapidAPI-Host': 'openfpl-api.p.rapidapi.com'
    }
};

fetch('https://openfpl-api.p.rapidapi.com/api/gw/scout?gameweek=7', options)
    .then(response => response.json())
    .then(data => console.log(data));

API Documentation

• RapidAPI Docs: https://rapidapi.com/elcaiseri-elcaiseri-default/api/openfpl-api

Main Endpoints

• GET /api/health — Health check
• GET /api/gameweeks — Available gameweeks with saved data
• GET /api/gw/scout — Get optimal FPL team for specific gameweek
• GET /api/gw/playerpoints — Get filtered player point predictions
• GET /api — API information and metadata

Sample /api/gw/scout response:

{
  "scout_team": [
    {
      "element_type": "Goalkeeper",
      "web_name": "Alisson",
      "team_name": "Liverpool",
      "expected_points": 5.2,
      "role": "",
      "now_cost": 55,
      "selected_by_percent": 15.5
    },
    {
      "element_type": "Defender",
      "web_name": "Alexander-Arnold",
      "team_name": "Liverpool",
      "expected_points": 8.1,
      "role": "captain",
      "now_cost": 70,
      "selected_by_percent": 45.2
    }
  ],
  "gameweek": 7,
  "version": "4.0.0",
  "credits": "OpenFPL-Scout AI - Team Predictions | Developed by Kassem @elcaiseri, 2025"
}

Screenshots

The web interface provides a beautiful visualization of your optimal FPL team:

Features of the UI:

• Football Pitch Layout: Players arranged in realistic formation
• Color-Coded Positions: Goalkeepers (Orange), Defenders (Blue), Midfielders (Light Blue), Forwards (Green)
• Captain Badges: Golden 'C' for captain, Silver 'VC' for vice-captain
• Fixture Information: Opponent teams and home/away indicators
• Expected Points: AI-predicted points for each player
• Team Statistics: Total expected points and player count
• Responsive Design: Works on desktop, tablet, and mobile devices

Model Overview

Model	Version	Description
Linear Regression	v4.0	Baseline linear model
XGBoost	v4.0	Gradient boosting ensemble
CatBoost	v3.0	Categorical boosting model

• Ensemble predictions for accuracy
• Feature importance analysis
• Optimized for FPL player performance

API Integration

Integrates with Football Data API for:

• Live fixtures and matchups
• Home/away status
• Gameweek info

For RapidAPI Users: All data is pre-processed and cached for optimal performance.

Code Structure

• main.py: FastAPI app and endpoints
• src/scout.py: FPLScout class (predictions, team selection)
• src/models.py: Pydantic response models
• src/utils.py: Config and helpers
• src/logger.py: Logging

What's New

• 🔌 RapidAPI Marketplace: Now available on RapidAPI with professional support
• 📈 Enhanced API: New endpoints for gameweeks and player filtering
• 🌐 Live Deployment: Available on Google Cloud Platform
• 🎨 Beautiful Web Interface: Interactive team visualization with football pitch layout
• 📸 Screenshot Feature: Export your team lineup as high-quality PNG images
• 📱 Mobile Responsive: Perfect experience on all devices
• 2024/2025 Season: Models updated with latest data
• CatBoost Integration: Improved ML pipeline (Issue #1)
• RESTful API: FastAPI endpoints for team selection and predictions
• Rebranding: Now OpenFPL-Scout-AI
• Refactored Code: Improved modularity and maintainability
• AI-Powered Predictions: Advanced ensemble models
• Async Processing: Fast parallel predictions
• Live Data: Real-time match integration
• Docker Support: Easy deployment

Contributing

Contributions welcome! Ideas for improvement:

• Enhanced algorithms and selection logic
• Additional UI features and visualizations
• Player injury/form tracking
• Better documentation
• Mobile app development

Fork, branch, and submit a pull request.

License

MIT License — see LICENSE for details.

API Support & Contact

For API Support:

• RapidAPI Marketplace: https://rapidapi.com/elcaiseri-elcaiseri-default/api/openfpl-api
• Email: support@openfpl.kassem.dev

General Questions:

• Email: iqasem4444@gmail.com