Replication Data and Code for: Machine Learning-Based Hedge Ratios for Cryptocurrency Futures under Regime-Dependent Tail Risk

This repository contains the dataset and replication code for the manuscript: “Machine Learning-Based Hedge Ratios for Cryptocurrency Futures under Regime-Dependent Tail Risk” (submitted to Research in International Business and Finance).

The dataset includes daily BTC and ETH spot and futures prices sourced from Yahoo Finance (BTC: September 2019 – May 2025; ETH: February 2021 – May 2025), along with computed hedge ratios and hedging effectiveness metrics for three models: Ordinary Least Squares (OLS), Dynamic Conditional Correlation GARCH (DCC-GARCH), and Random Forest (RF).

Files included:
– btc_merged.xlsx: BTC spot and futures returns with hedge ratios (OLS, DCC-GARCH, RF)
– btc_hr_results.xlsx: BTC hedging effectiveness results (HE, VaR, CVaR reduction)
– btc_all_models.xlsx: BTC full model comparison across all regimes
– eth_merged.xlsx: ETH spot and futures returns with hedge ratios
– eth_hr_results.xlsx: ETH hedging effectiveness results
– eth_all_models.xlsx: ETH full model comparison across all regimes
– table1_summary_stats.xlsx: Descriptive statistics (Table 1 in manuscript)
– crypto_hedging.ipynb: Jupyter notebook containing all analysis code

All data are publicly available from Yahoo Finance (BTC=F, ETH=F).