Improved Financial Forecasting via Quantum Machine Learning

TL;DR

This work demonstrates two complementary quantum-ML approaches to financial forecasting: (1) DPP-enhanced Random Forests for churn prediction that improve precision and business impact at higher training cost, and (2) quantum neural networks with orthogonal and compound layers for credit risk that maintain or improve predictive performance with far fewer parameters. Classical implementations on churn data show a ~5.9% precision gain and measurable bottom-line improvements, while quantum circuits for DPP sampling and hardware experiments reveal current hardware limitations but potential gains with error mitigation. On the credit risk side, quantum-layer architectures achieve similar Gini scores to strong classical baselines but with dramatically reduced parameter counts, and hardware experiments with mitigation bring observed performance closer to noiseless expectations. Collectively, the results illustrate that quantum ideas can yield practical benefits today as quantum-inspired classical solutions and are poised to deliver larger gains as quantum hardware matures.

Abstract

Quantum algorithms have the potential to enhance machine learning across a variety of domains and applications. In this work, we show how quantum machine learning can be used to improve financial forecasting. First, we use classical and quantum Determinantal Point Processes to enhance Random Forest models for churn prediction, improving precision by almost 6%. Second, we design quantum neural network architectures with orthogonal and compound layers for credit risk assessment, which match classical performance with significantly fewer parameters. Our results demonstrate that leveraging quantum ideas can effectively enhance the performance of machine learning, both today as quantum-inspired classical ML solutions, and even more in the future, with the advent of better quantum hardware.

Figures (19)

• Figure 2: Precision-recall curve for the test set. Using DPP with the Random Forest algorithm shows an improvement of $5.9\%$.
• Figure 3: Classical benchmark (BM) vs DPP-RF solution: money withdrawn per month by the flagged 500 customers, comparing the benchmark model (blue line) to the DPP-RF one (orange line). On the y axis, we have monetary values (not shown). The green line represents the total amount of money withdrawn by all customers in each month. The purple line is the sum of the 500 largest withdrawals, which is the maximum value that the model could capture. The red line represents the withdrawals captured by randomly flagging 500 observations. The y-axis units are omitted for confidentiality.
• Figure 4: Classical benchmark vs DPP-RF solution - percentage of total withdrawals captured per month, that is, relative to the green line in Fig. \ref{['fig:bench1']}. On average over the 11 test months, the BM model captures 61.42% of the total, whilst the DPP-RF model captures 62.77% --- an improvement of 1.35%.
• Figure 5: Classical benchmark vs DPP-RF solution - the percentage of maximum money possible to be captured (given $\text{n\_flags} = 500$ customers flagged every month), that is, relative to the purple line in Fig. \ref{['fig:bench1']}. On average over the 11 test months, the BM model captures 69.18% of the total, whilst the DPP-RF model captures 70.72% --- an improvement of 1.54%.
• Figure 6: Clifford Loader circuit $\mathcal{C}(x)$ for $x \in \mathbb{R}^8$.