A Noise-Aware Quantum Algorithm for Credit Valuation Adjustments on Real Quantum Hardware

Credit Valuation Adjustment (CVA) is a critical financial metric that accounts for counterparty credit risk. Because it requires repeated expectation estimation, it is computationally expensive to calculate using classical Monte Carlo methods. While quantum amplitude estimation (QAE) theoretically offers a quadratic speedup, it remains unclear whether this advantage persists when using realistic financial models on current, noisy quantum hardware. This paper investigates the practical viability of an end-to-end quantum workflow for CVA.

The authors develop a comprehensive quantum pipeline that includes market calibration, discretisation, and oracle construction for a correlated two-asset portfolio. To address the hardware limitations, they introduce Contrast-Aware Bayesian Iterative Quantum Amplitude Estimation (CABIQAE). Unlike standard methods that assume ideal conditions, CABIQAE incorporates experimentally calibrated Grover-contrast loss—a measure of how hardware noise degrades the signal—directly into the Bayesian inference and circuit-depth selection process. The researchers validate this workflow using a hardware-replay methodology on IBM Quantum hardware.

The study demonstrates that CABIQAE is better suited to the limitations of near-term quantum devices than noise-agnostic alternatives. By accounting for the loss of Grover contrast as circuit depth increases, the algorithm achieves more reliable estimates and significantly reduces the classical post-processing runtime compared to standard Bayesian amplitude estimation (BAE) baselines. The authors provide a detailed error decomposition, showing that while the quantum approach is feasible, the overall accuracy remains heavily dependent on the resolution of the discretisation grid and the depth of the quantum circuits.

This work bridges the gap between theoretical quantum finance algorithms and real-world hardware implementation. By treating amplitude estimation as a practical bottleneck rather than an ideal primitive, the authors provide a realistic assessment of where quantum advantage can currently be applied in finance. It highlights that for near-term quantum devices, algorithmic design must be tightly coupled with the specific noise characteristics of the hardware.