alan-turing-institute/AIrsenal

Machine learning Fantasy Premier League team

What it solves

AIrsenal helps Fantasy Premier League (FPL) players optimize their team selection and transfer strategy using machine learning to maximize points.

How it works

The system operates as a pipeline that integrates data collection, prediction, and optimization:

  1. Data Management: It builds and updates a local SQLite database with historical data from the last three seasons and real-time data from the FPL API.
  2. Points Prediction: It uses a combination of a team-level model (to predict match scorelines) and a player-level model (to predict goal involvements), alongside historical heuristics, to estimate expected points for players.
  3. Optimization: An optimizer calculates the best transfer strategy and team lineup over a specified number of future gameweeks, including the strategic use of special chips like Wildcards and Free Hits.
  4. Execution: The tool can automatically apply recommended transfers and lineups directly to the FPL website via the user's credentials.

Who it’s for

Fantasy Premier League players who want a data-driven, automated approach to managing their squads.

Highlights

  • End-to-end pipeline: Automates everything from database setup to applying transfers.
  • Predictive modeling: Combines team and player-level ML models for point forecasting.
  • Multi-week optimization: Plans transfers and captaincy across multiple future gameweeks.
  • Direct FPL integration: Ability to push team changes directly to the official FPL platform.

Related

  • Project
    AgileRL/AgileRL

    AgileRL is a Python library for deep reinforcement learning that adds *RLOps*: evolutionary hyper‑parameter optimisation, distributed training, and a cloud platform (Arena) for running experiments at scale. It ships many algorithms (PPO, DQN, TD3, MADDPG, bandits, LLM fine‑tuning, etc.) and lets you train a single agent or a population that evolves its hyper‑parameters automatically, promising up to 10× faster HPO than traditional methods.

  • Dispatch
    PipelineRL: Optimizing LLM Reinforcement Learning via Inflight Weight Updates

    Hugging Face and ServiceNow Research have open-sourced PipelineRL, an experimental RL implementation that uses inflight weight updates to eliminate the trade-off between inference throughput and on-policy data collection.

  • Dispatch
    Prime Agent self-improving RLM harness release

    Prime Agent is an open‑source, self‑improving coding harness built on Recursive Language Models and a continual‑state harness, achieving state‑of‑the‑art scores on ARC‑AGI‑3 and competitive performance on long‑context benchmarks.

  • Project
    agentuniverse-ai/agentUniverse

    agentUniverse is an Apache‑2.0 Python framework for building and orchestrating LLM‑powered multi‑agent applications. It offers ready‑made collaboration patterns (PEER, DOE), supports dozens of LLM providers, includes tools for domain knowledge injection, observability via OpenTelemetry, and a visual workflow UI. The library is used in real‑world financial products at Ant Group and provides extensive docs, sample apps, and a community on GitHub/Discord.

  • Project
    plasma-umass/scalene

    Scalene is a fast, line‑level Python profiler that measures CPU, GPU (NVIDIA) and memory usage, separates Python from native code, visualizes stitched stacks, and can ask LLMs (OpenAI, Azure, Bedrock, Ollama, etc.) for optimization suggestions directly from the UI.