Cryptocoins Analytics

Cryptocoins Analytics is a Python and R project for the analysis and forecasting of financial time series and cryptocurrency price trends.

Project Structure

This repository is organized as it follows:

Analysis: collection of scripts designed to:

• Study correlation patterns among cryptocurrencies and generate representations in the form of correlograms.
• Implement the Toda-Yamamoto procedure to test for Granger-causality between correlated cryptocoins.
• Train and test SOTA machine learning models to forecast cryptocoin price series (namely GRU, LSTM, CatBoost, LightGBM and XGBoost).

Data: pre-built datasets adopted in the above-mentioned analyses, spanning 33 months from 20-02-2020 to 26-02-2023.

Data

The data sources used to gather information about cryptocurrency trends are CoinMarketCap and Binance. The two pre-built datasets (coinmarketcap.csv and binance.csv) are available in a compressed .zip format.

Getting Started

The Python version used in this project is 3.9. The R version is 3.6. A list of the external Python libraries/dependencies can be found in the file requirements.txt.

Authors

Pasquale De Rosa, University of Neuchâtel, pasquale.derosa@unine.ch.
Pascal Felber, University of Neuchâtel, pascal.felber@unine.ch.
Valerio Schiavoni, University of Neuchâtel, valerio.schiavoni@unine.ch.

References

Pasquale De Rosa, Pascal Felber and Valerio Schiavoni. 2023. Practical Forecasting of Cryptocoins Timeseries using Correlation Patterns. In: Proceedings of the 17th ACM International Conference on Distributed and Event-based Systems. DEBS 2023. https://doi.org/10.1145/3583678.3596888.

Pasquale De Rosa and Valerio Schiavoni. 2022. Understanding Cryptocoins Trends Correlations. In: Distributed Applications and Interoperable Systems. DAIS 2022. https://doi.org/10.1007/978-3-031-16092-9_3.

License

MIT