A Credibility-Based CVaR Model for Cryptocurrency Portfolio Optimization under Partial Information

Abstract

The cryptocurrency market presents profitable yet highly uncertain investment opportunities, marked by rapid expansion and strong volatility. Conventional risk management approaches, built on probabilistic models and past data, often fail to capture the distinctive behavior and unpredictability of this market. To address these issues, this study introduces a portfolio optimization framework designed specifically for cryptocurrencies, based on a credibilistic Conditional Value at Risk (CVaR) model. CVaR is applied as the main risk metric because it targets downside risk and extreme losses, making it suitable for handling sharp downturns common in volatile digital assets. The model integrates credibility theory and trapezoidal fuzzy variables, enabling better representation of uncertainty and market instability. In contrast with traditional probability-based methods, this approach allows for more flexible and accurate risk control. The framework also accounts for real-world restrictions such as cardinality and floor–ceiling limits, which encourage diversification and reflect costs and regulatory requirements. Results from empirical testing confirm the ability of the model to build diversified portfolios that maintain a balance between risk and return. Overall, this work advances portfolio optimization for digital assets by applying advanced methods that provide investors with a practical and effective tool for managing risk in a complex and rapidly changing financial environment.