A network-based strategy of price correlations for optimal cryptocurrency portfolios

Paper: arXiv 2304.02362

Authors: Ruixue Jing, Luis Enrique Correa Rocha

Abstract

A cryptocurrency is a digital asset maintained by a decentralised system using cryptography. Investors in this emerging digital market are exploring the profitability potential of portfolios in place of single coins. Portfolios are particularly useful given that price forecasting in such a volatile market is challenging. The crypto market is a self-organised complex system where the complex inter-dependencies between the cryptocurrencies may be exploited to understand the market dynamics and build efficient portfolios. In this letter, we use network methods to identify highly decorrelated cryptocurrencies to create diversified portfolios using the Markowitz Portfolio Theory agnostic to future market behaviour. The performance of our network-based portfolios is optimal with 46 coins and superior to benchmarks up to an investment horizon of 14 days, reaching up to 1,066% average expected return within 1 day, with reasonable associated risks. We also show that popular cryptocurrencies are typically not included in the optimal portfolios. Past price correlations reduce risk and may improve the performance of crypto portfolios in comparison to methodologies based exclusively on price auto-correlations. Short-term crypto investments may be competitive to traditional high-risk investments such as the stock market or commodity market but call for caution given the high variability of prices.

Complexity vs Empirical Score

• Math Complexity: 5.5/10
• Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs not modelled; see penalty flags)
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents an interesting application of complex network theory to filter correlations for Markowitz portfolio optimization in the cryptocurrency market. Empirical results show strong short-term performance, though extreme return figures suggest potential sensitivity to market regimes and transaction costs.

Research Flowchart

  flowchart TD
    A[Goal: Build Optimal Crypto Portfolios via Network Decorrelation] --> B[Data: Historical Cryptocurrency Price Series]
    B --> C[Compute Price Correlations & Construct Network Matrix]
    C --> D[Identify Highly Decorrelated Coins using Network Theory]
    D --> E[Apply Markowitz Portfolio Theory for Diversification]
    E --> F[Outcomes: Optimal 46-Coin Portfolios Outperform Benchmarks<br>Up to 1,066% Avg 1-Day Return; Excludes Popular Coins]