Comparing Classical-Quantum Portfolio Optimization with Enhanced Constraints

TL;DR

This paper extends quantum annealer-based portfolio optimization by incorporating real-world constraints and compares its performance with classical algorithms, finding classical methods currently outperform quantum approaches.

Contribution

It introduces new constraints for quantum portfolio optimization, including fundamental analysis and sector investment bands, and evaluates quantum versus classical solution quality.

Findings

Classical algorithms outperform quantum annealer solutions with added constraints.

Enhanced constraints increase problem complexity, challenging quantum approaches.

Quantum solutions show potential but are not yet superior to classical methods.

Abstract

One of the problems frequently mentioned as a candidate for quantum advantage is that of selecting a portfolio of financial assets to maximize returns while minimizing risk. In this paper we formulate several real-world constraints for use in a Quantum Annealer (QA), extending the scenarios in which the algorithm can be implemented. Specifically, we show how to add fundamental analysis to the portfolio optimization problem, adding in asset-specific and global constraints based on chosen balance sheet metrics. We also expand on previous work in improving the constraint to enforce investment bands in sectors and limiting the number of assets to invest in, creating a robust and flexible solution amenable to QA. Importantly, we analyze the current state-of-the-art algorithms for solving such a problem using D-Wave's Quantum Processor and compare the quality of the solutions obtained to commercially-available optimization software. We explore a variety of traditional and new constraints that make the problem computationally harder to solve and show that even with these additional constraints, classical algorithms outperform current hybrid solutions in the static portfolio optimization model.