Paper: arXiv 2507.22035
Authors: David Dechant, Eliot Schwander, Lucas van Drooge, Charles Moussa, Diego Garlaschelli, Vedran Dunjko, Jordi Tura
Abstract
Quantum generative adversarial networks (QGANs) have been investigated as a method for generating synthetic data with the goal of augmenting training data sets for neural networks. This is especially relevant for financial time series, since we only ever observe one realization of the process, namely the historical evolution of the market, which is further limited by data availability and the age of the market. However, for classical generative adversarial networks it has been shown that generated data may (often) not exhibit desired properties (also called stylized facts), such as matching a certain distribution or showing specific temporal correlations. Here, we investigate whether quantum correlations in quantum inspired models of QGANs can help in the generation of financial time series. We train QGANs, composed of a quantum generator and a classical discriminator, and investigate two approaches for simulating the quantum generator: a full simulation of the quantum circuits, and an approximate simulation using tensor network methods. We tested how the choice of hyperparameters, such as the circuit depth and bond dimensions, influenced the quality of the generated time series. The QGAN that we trained generate synthetic financial time series that not only match the target distribution but also exhibit the desired temporal correlations, with the quality of each property depending on the hyperparameters and simulation method.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 4.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper employs advanced quantum information concepts, variational quantum algorithms, and tensor network methods (MPS) requiring significant mathematical abstraction, while its validation relies primarily on qualitative comparison of simulated time series against stylized facts rather than on live backtesting or extensive statistical benchmarking against real market data.
Research Flowchart
flowchart TD
A["Research Goal: Can Quantum Correlations<br>Generate Financial Time Series with<br>Temporal Properties?"] --> B["Methodology: Quantum Generative<br>Adversarial Network"]
B --> C{"Simulation Methods"}
C --> D["Full Quantum Circuit Simulation"]
C --> E["Tensor Network Approximation"]
B --> F["Data Input: Historical Financial<br>Equity Time Series"]
F --> G["Training Process: Generator creates<br>synthetic data; Discriminator classifies<br>real vs. synthetic. Hyperparameters<br>varied: Circuit Depth & Bond Dimension."]
G --> H["Key Findings: QGANs successfully<br>generate data matching target<br>distribution and temporal correlations.<br>Quality depends on hyperparameters<br>and simulation method."]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a Wasserstein quantum generative adversarial network (QGAN) that uses a parameterized quantum circuit as a generator (expectation‑value sampler) and a classical convolutional neural network as a discriminator to generate synthetic financial time series.
Source sentence
IMPLEMENTATION For this paper, we trained a Wasserstein QGAN for generating time series based on training data originating in the time series of the S&P 500 index. We use a hybrid approach, with a classical neural network as a discrimi-nator and a parameterized quantum circuit (PQC) as a generator.
Assumes 6
- The expectation values of single‑qubit Pauli‑X and Pauli‑Z operators are interpreted as the log return of the index at each time step. model
Source sentence
The discrete time series spans from 20 to 40 points in time, where the expectation value of single-qubit Pauli-X and Pauli-Z operators is interpreted as the log return of the value of the index at each time step.
- Matrix product state (MPS) simulations can efficiently approximate longer time series by exploiting linear structures. model
Source sentence
While full-state sim-ulations are limited to short time intervals due to their exponential scaling, MPS simulations enable us to model longer time series by exploiting their ability to efficiently replicate linear structures such as financial time series.
- Increasing the bond dimension balances computational cost against simulation accuracy. model
Source sentence
In the MPS simulations, we vary the bond dimension (also referred to as the tensor train rank) to balance compu-tational costs and simulation accuracy.
- Training data are normalized to zero mean and unit variance before model training. data
Source sentence
(i) We normalize our training data to have a mean of 0 and a variance of 1: rt,(i) := rt −µr σr , (12) where rt is the original log returns, and µr and σr esti-mate the mean and standard deviation of all log returns.
Show 2 more
- An inverse Lambert‑W transform is applied to heavy‑tailed log returns to bring them closer to a Gaussian distribution. data
Source sentence
(ii) As learning a heavy-tailed distribution can be chal-lenging due to a limited number of samples in the tails, we implement the inverse Lambert-W transform on the normalized log returns.
- The Wasserstein distance computed over the aggregated time‑series distribution is sufficient for training the QGAN. statistical
Source sentence
This approach leverages the PQC architecture to intrinsically capture temporal correlations in the time series, while the training itself is performed solely on the Wasserstein distance computed over the ag-gregated time series distribution.
Data
- S&P 500 index log returns market-prices public
Source sentence
In this work, we develop a quantum GAN for syn-thetic data generation and examine its ability in gen-erating time series replicating the S&P 500 index, in-cluding its distribution and temporal correlations. An example of a financial time series is the S&P 500 in-dex, which includes 500 of most valuable companies that are listed on US stock exchanges [25].
Costs
Not applicable
Finds 5
- The trained QGAN generates synthetic time series that match the target distribution and exhibit temporal correlations such as volatility clustering. performance
Source sentence
The QGAN that we trained generate synthetic financial time series that not only match the target distribution but also exhibit the desired temporal correlations, with the quality of each property depending on the hyperparameters and simulation method. Our trained QGANs generate time series that match the desired distributions and exhibiting some of the tempo-ral correlations seen in financial time series, such as in
- MPS simulations with higher bond dimensions and more layers enable training of longer time series compared to full‑state simulations. performance
Source sentence
The lat-ter approach allowed us to simulate PQCs with a higher number of layers and qubits, which makes it possible to train the generation of longer time series. Simulating the PQC with full-state simulations and MPS simulations yield different results, with circuit depth and the MPS bond dimension further influencing the performance.
- Fidelity between full‑state and MPS approximations increases with bond dimension, indicating improved accuracy. statistical
Source sentence
As shown in Figure 4, the fidelity increases with χ, indicating improved approxima-tion accuracy.
- Generated series from the MPS‑simulated QGAN shows positive, decreasing absolute autocorrelation (volatility clustering) closer to the S&P 500 behavior. performance figure from elsewhere in the paper
Source sentence
In contrast to the time series generated with the full-state simulation shown in Figure 5, the absolute autocorrelation (Subfigure (d)) that indicates volatility clustering is positive and decreasing for all time lags.
- Circuit depth and MPS bond dimension further influence the quality of generated distribution and temporal‑correlation metrics. performance
Source sentence
Simulating the PQC with full-state simulations and MPS simulations yield different results, with circuit depth and the MPS bond dimension further influencing the performance.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 30 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (99)
- JAX : composable transformations of P ython+ N um P y programs (2018) — link
- S& P 500 \textregistered (2025)
- The Computational Complexity of Linear Optics (2013) — link
- TensorFlow : A System for Large-Scale Machine Learning (2016)
- The Power of Quantum Neural Networks (2021) — link
- Noise Robustness and Experimental Demonstration of a Quantum Generative Adversarial Network for Continuous Distributions (2021) — link
- Wasserstein Generative Adversarial Networks (2017)
- Generating Synthetic Data in Finance : Opportunities , Challenges and Pitfalls (2020) — link
and 91 more
- Data Augmentation Experiments with Style-Based Quantum Generative Adversarial Networks on Trapped-Ion and Superconducting-Qubit Technologies (2024) — link
- Parameterized Quantum Circuits as Universal Generative Models for Continuous Multivariate Distributions (2024) — link
- Parameterized Quantum Circuits as Universal Generative Models for Continuous Multivariate Distributions (2024) — link
- Order p Quantum Wasserstein Distances from Couplings (2025) — link
- A Generative Modeling Approach for Benchmarking and Training Shallow Quantum Circuits (2018) — link
- Tensor Networks for Quantum Computing (2025) — link
- The Pricing of Options and Corporate Liabilities (1973) — link
- Trapped-Ion Quantum Computing: Progress and Challenges (2019)
- Variational Quantum Algorithms (2021)
- Quantum Wasserstein Generative Adversarial Networks (2019)
- Quantum Long Short-Term Memory (2022) — link
- Quantum Computations with Cold Trapped Ions (1995) — link
- Matrix Product States and Projected Entangled Pair States: Concepts , Symmetries, Theorems (2021) — link
- Superconducting Quantum Bits (2008)
- Empirical Properties of Asset Returns: Stylized Facts and Statistical Issues (2001) — link
- Quantum versus Classical Generative Modelling in Finance (2021) — link
- Quantum Generative Adversarial Networks (2018) — link
- Error and Resource Estimates of Variational Quantum Algorithms for Solving Differential Equations Based on Runge-Kutta Methods (2024) — link
- Quantum Computing for High-Energy Physics : State of the Art and Challenges (2024) — link
- Machine Learning in Finance : From Theory to Practice (2020) — link
- Generation of Realistic Synthetic Financial Time-series (2022) — link
- Data Augmentation Using GANs (2019)
- Generative Adversarial Networks in Finance: An Overview (2021) — link
- A Hybrid Quantum Wasserstein GAN with Applications to Option Pricing (2023) — link
- The Lambert Way to Gaussianize Heavy-Tailed Data with the Inverse of Tukey ’s h Transformation as a Special Case (2015) — link
- Generative Adversarial Nets (2014)
- Generative Adversarial Networks (2020) — link
- Quimb: A Python Package for Quantum Information and Many-Body Calculations (2018) — link
- A Review on Generative Adversarial Networks: Algorithms , Theory, and Applications (2021)
- Improved Training of Wasserstein Gans (2017)
- Quantum Computing with Neutral Atoms (2020)
- Anomaly Detection with Variational Quantum Generative Adversarial Networks (2021) — link
- WGAN Financial Time-Series (2025) — link
- Experimental Quantum Generative Adversarial Networks for Image Generation (2021) — link
- A Survey on Superconducting Computing Technology: Circuits, Architectures and Design Tools (2022)
- Quantum Information Processing Using Quantum Dot Spins and Cavity QED (1999) — link
- Hardware-Efficient Variational Quantum Eigensolver for Small Molecules and Quantum Magnets (2017)
- Exploring the Advantages of Quantum Generative Adversarial Networks in Generative Chemistry (2023) — link
- Progressive Growing of GANs for Improved Quality , Stability , and Variation (2018) — link
- A Style-Based Generator Architecture for Generative Adversarial Networks (2019)
- Learning Quantum Data with the Quantum Earth Mover’s Distance (2022) — link
- A Scheme for Efficient Quantum Computation with Linear Optics (2001)
- Quantum Computing for Generative Modeling and Applications (2023) — link
- Barren Plateaus in Variational Quantum Computing (2025)
- Differentiable Learning of Quantum Circuit Born Machine (2018) — link
- Quantum Generative Adversarial Learning (2018) — link
- Community Detection for Correlation Matrices (2015) — link
- Quantum Machine Learning Advantages beyond Hardness of Evaluation (2025) — link
- Contextual Quantum Neural Networks for Stock Price Prediction (2025) — link
- A Survey of Recent Advances in Quantum Generative Adversarial Networks (2023) — link
- Entangling Quantum Generative Adversarial Networks (2022) — link
- Enhancing Financial Time Series Prediction with Quantum-Enhanced Synthetic Data Generation : A Case Study on the S & P 500 Using a Quantum Wasserstein Generative Adversarial Network Approach with a Gradient Penalty (2024) — link
- Tensor- Train Decomposition (2011) — link
- Thermodynamic Limit of Density Matrix Renormalization (1995) — link
- QuantumLeap : Hybrid Quantum Neural Network for Financial Predictions (2022) — link
- Synthetic Data Generation with Hybrid Quantum-Classical Models for the Financial Sector (2024) — link
- Synthetic Data Applications in Finance (2024) — link
- Quantum Computing in the NISQ Era and Beyond (2018) — link
- Understanding Deep Learning (2023)
- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks (2016) — link
- Quantum Computational Finance: Monte Carlo Pricing of Financial Derivatives (2018) — link
- Tensor Networks for Quantum Machine Learning (2023) — link
- A Universal Qudit Quantum Processor with Trapped Ions (2022) — link
- Variational Quantum Generators : Generative Adversarial Quantum Machine Learning for Continuous Distributions (2021) — link
- Trainability Barriers and Opportunities in Quantum Generative Modeling (2024) — link
- Generative Adversarial Networks ( GANs ): Challenges , Solutions , and Future Directions (2022) — link
- Quantum Generative Modelling for Financial Time Series (2022) — link
- Shadow- Frugal Expectation-Value-Sampling Variational Quantum Generative Model (2024)
- Shadow- Frugal Expectation-Value-Sampling Variational Quantum Generative Model (2024) — link
- Effects of Observable Choices in the Performance of Variational Quantum Generative Models - CONFIDENTIAL (2024)
- A Survey on Image Data Augmentation for Deep Learning (2019) — link
- MosaiQ : Quantum Generative Adversarial Networks for Image Generation on NISQ Computers (2023)
- Quantum Generative Adversarial Network for Generating Discrete Distribution (2020) — link
- On the Quantum versus Classical Learnability of Discrete Distributions (2021) — link
- Modeling Financial Time-Series with Generative Adversarial Networks (2019) — link
- Generation of Synthetic Financial Time Series by Diffusion Models — → 2410.18897
- Data Augmentation Using GANs (2019) — link
- Recent Advances for Quantum Neural Networks in Generative Learning (2022) — link
- Hybrid Quantum – Classical Generative Adversarial Network for High-Resolution Image Generation (2023) — link
- Analysis of Financial Time Series
- Density Matrix Renormalization Group and Periodic Boundary Conditions : A Quantum Information Perspective (2004) — link
- Matrix Product States, Projected Entangled Pair States, and Variational Renormalization Group Methods for Quantum Spin Systems (2008) — link
- The Wasserstein Distances (2009) — link
- Qudits and High-Dimensional Quantum Computing (2020) — link
- Quant GANs : Deep Generation of Financial Time Series (2020) — link
- Quantum Reinforcement Learning of Classical Rare Dynamics: Enhancement by Intrinsic Fourier Features (2025) — link
- Quantum Risk Analysis (2019) — link
- Stock Market Prediction Based on Generative Adversarial Network (2019) — link
- Hybrid Quantum–Classical Generative Adversarial Networks for Image Generation via Learning Discrete Distribution (2023) — link
- Quantum Generative Adversarial Networks for Learning and Loading Random Distributions (2019) — link
- Time series (1975)
1 of 99 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.