README.md

OpenFPL

The accurate openly available forecasting method for Fantasy Premier League

---

Want to support this initiative?

---

Get started

1. Plug

With Python preinstalled, run: pip install -r plug.txt

2. Play

Open and run play.ipynb for OpenFPL predictions on sample data

Custom data

To use OpenFPL on custom data, you need to construct samples based on data from FPL and Understat APIs (see data/samples.csv and paper for inspiration):

• FPL API
• Understat API

Historical FPL and Understat data can be accessed by help of FPL Historical Dataset

Head-to-head evaluation with state-of-the-art commercial method

Method	RMSEZeros*	RMSEBlanks*	RMSETickers*	RMSEHaulers*
OpenFPL	0.818	1.291	1.517	5.142
FPL Review Massive Data Model	0.689	1.189	1.594	5.172

^* Zeros: Non-playing and 0 FPL points, Blanks: ≤ 2 FPL points, Tickers: 3 or 4 FPL points, Haulers: ≥ 5 FPL points

Resources

• Scientific paper - OpenFPL: An open-source forecasting method rivaling state-of-the-art Fantasy Premier League services
• Model search framework - K-Best Search

Citation

Should you find the work helpful in your research, please cite the following:

@article{groos2025openfpl,
  title={OpenFPL: An open-source forecasting method rivaling state-of-the-art Fantasy Premier League services},
  author={Groos, Daniel},
  year={2025},
  publisher={arXiv}
}