Risk-Based Portfolio Optimization in the Cryptocurrency World

Abstract

This study explores the performance of seven state-of-the-art risk-based portfolio optimization strategies from the perspective of a cryptocurrency investor. Analyzing the inverse volatility, minimum variance, l2-norm constrained minimum variance, l2-norm constrained maximum decorrelation, maximum diversification and risk parity portfolio, we find that most strategies systematically outperform individual cryptocurrencies and the equally-weighted benchmark portfolio. Further, a bull and bear market performance comparison as well as tail, extreme risk, and diversification analyses reveal that these strategies provide significant downside risk reduction. The results are robust to using different estimation windows, rebalancing periods and covariance estimation methodologies. Finally, our empirical results indicate that the maximum decorrelation portfolio is the worst strategy in terms of risk-adjusted return, while the long-only minimum variance portfolio is the best performing strategy.