Researchers show financial modelling can work on today’s noisy quantum computers
HSBC has worked with quantum middleware developer Haiqu and a group of academic researchers on a method for running financial models on commercially available quantum hardware. The joint research, published in Physical Review Research, focuses on how to encode real-world probability distributions into quantum circuits efficiently.
The work comes as banks and financial institutions look at quantum computing both as a future security challenge and as a potential tool for better market simulation. One major concern is post-quantum cryptography, but quantum systems may also help with more advanced financial modelling.
The research team included specialists from HSBC, Haiqu, Czech Technical University, the University of Zurich, the Akhiezer Institute for Theoretical Physics, Karazin Kharkiv National University, and Athena Research Center in Greece. Their focus was Lévy distributions, which are widely used to model extreme movements in stock markets around the world.
The researchers said their approach could improve the modelling of market behaviour by capturing heavy tails, skewness, and volatility clustering more accurately.
Haiqu said quantum computing has potential uses in derivative pricing, portfolio optimisation, fraud detection, and machine learning, but those applications depend on loading realistic financial distributions into a quantum system. That data-loading step is widely seen as one of the biggest obstacles in quantum algorithm design, especially for risk modelling and simulation.
The company added that traditional methods can require an exponential number of quantum operations as the number of qubits increases, which makes them difficult to run on current noisy hardware with limited circuit depth. To solve this, Haiqu developed compact quantum circuits with linear scaling instead of exponential scaling.
Mykola Maksymenko, co-founder and CTO of Haiqu, said one of the main practical problems is getting realistic financial data onto today’s quantum hardware. He said the new work offers a scalable way around that limitation and moves quantum finance workflows closer to real execution.
The researchers used matrix product state methods to build shallow quantum circuits that can encode smooth functions, including probability distributions, directly into quantum states.
On a 25-qubit IBM quantum computer, the team said the accuracy was good enough to pass quantitative statistical tests, even on current noisy devices. They added that the distribution-loading results could matter for financial risk analysis, risk management, and decision-making based on time-series financial data.
The team also tested a sampling-based workflow on 64-qubit hardware, which they said supports the feasibility of the approach at larger scales. They reported similar behaviour in simulations up to 156 qubits, suggesting the method could extend to much larger problem sizes.
Philip Intallura, group head of quantum technologies at HSBC, said efficiently preparing complex probability distributions is a key step in many quantum algorithms. He said the work shows how this can be done with much shallower quantum circuits, bringing practical financial risk modelling closer to reality.
