Inferring Multi-Period Optimal Portfolios via Detrending Moving Average Cluster Entropy

TL;DR

This paper introduces a novel detrended cluster entropy method for estimating multi-period portfolio weights, demonstrating improved robustness and stability over traditional approaches using high-frequency market data.

Contribution

The paper proposes a new entropy-based approach to portfolio optimization that overcomes limitations of traditional mean-variance methods.

Findings

Reliable estimation of portfolio weights from real market data

High diversity, robustness, and stability of the proposed portfolio

Effective across various temporal horizons

Abstract

Despite half a century of research, there is still no general agreement about the optimal approach to build a robust multi-period portfolio. We address this question by proposing the detrended cluster entropy approach to estimate the portfolio weights of high-frequency market indices. The information measure produces reliable estimates of the portfolio weights gathered from the real-world market data at varying temporal horizons. The portfolio exhibits a high level of diversity, robustness and stability as it is not affected by the drawbacks of traditional mean-variance approaches.