DYNAMIC INTEGRATION OF BITCOIN INTO TRADITIONALPORTFOLIOS: A WALK-FORWARD OPTIMIZATION APPROACHUNDER RISK CONSTRAINTS

Abstract

This study examines the integration of Bitcoin—an asset characterized by extreme volatility and an asymmetric return distribution—into traditional investment portfolios from a dynamic perspective. Utilizing daily data spanning the 2016–2026 period, out-of-sample tests were conducted on a traditional benchmark portfolio through a dynamic walk-forward algorithm. To capture the fat-tailed nature of cryptocurrencies, a comparative analysis was performed between the traditional Mean-Variance (Markowitz) model and Conditional Value-at-Risk (CVaR) optimization, while a dynamic risk-free rate (T-Bill) was integrated into the model to reflect macroeconomic regime shifts. Empirical findings indicate that while unconstrained optimization models enhance the Sharpe ratio, they simultaneously drive the portfolio toward corner solutions, thereby generating unacceptable maximum drawdown risks exceeding 37%. Conversely, in scenarios where the Bitcoin weight is constrained to 5% and 10% in line with rational risk management principles, both models demonstrate similar performance, maintaining maximum drawdown close to benchmark levels while substantially increasing the Sharpe ratio. This reveals that the primary mechanism ensuring portfolio resilience is not the chosen mathematical model, but rather the disciplined constraint mechanisms themselves. Ultimately, a dynamically constrained, low-weight crypto-asset alloca-tion offers an optimal framework for investors seeking to safely capitalize on asymmetric return potentials without materially compromising the portfolio's risk isolation.