Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2512.15738

Hybrid Quantum-Classical Ensemble Learning for S&P 500 Directional Prediction

Weinberg, Abraham Itzhak

Published Jan 1, 2025• 0 citations
arXivHTMLPDF

Abstract

Financial market prediction is a challenging application of machine learning, where even small improvements in directional accuracy can yield substantial value. Most models struggle to exceed 55--57\% accuracy due to high noise, non-stationarity, and market efficiency. We introduce a hybrid ensemble framework combining quantum sentiment analysis, Decision Transformer architecture, and strategic model selection, achieving 60.14\% directional accuracy on S\&P 500 prediction, a 3.10\% improvement over individual models. Our framework addresses three limitations of prior approaches. First, architecture diversity dominates dataset diversity: combining different learning algorithms (LSTM, Decision Transformer, XGBoost, Random Forest, Logistic Regression) on the same data outperforms training identical architectures on multiple datasets (60.14\% vs.\ 52.80\%), confirmed by correlation analysis ($r>0.6$ among same-architecture models). Second, a 4-qubit variational quantum circuit enhances sentiment analysis, providing +0.8\% to +1.5\% gains per model. Third, smart filtering excludes weak predictors (accuracy $<52\%$), improving ensemble performance (Top-7 models: 60.14\% vs.\ all 35 models: 51.2\%). We evaluate on 2020--2023 market data across seven instruments, covering diverse regimes including the COVID-19 crash and inflation-driven correction. McNemar's test confirms statistical significance ($p<0.05$). Preliminary backtesting with confidence-based filtering (6+ model consensus) yields a Sharpe ratio of 1.2 versus buy-and-hold's 0.8, demonstrating practical trading potential.

DOI
10.13140/rg.2.2.34951.46248

Related papers

Evaluating hybrid quantum–classical learning for financial forecasting: a comparative study with classical machine learning models on BSE indices

Rashmi Nair, Pankaj Ramchandra Chandre, Shahin Shoukat Makubhai +3 more

The forecasting of financial time series has gained more significance in decision making within a dynamic economic setting. Over the last few years, there has been a push in exploring both classical machine learning methods and novel quantum machine learning models with a view to enhance predictive accuracy. This paper suggests a quantum-classical regression framework developed that combines the feature selection in the form of a Random Forest in a hybrid approach, dimensionality reduction in the mode of Principal Component Analysis (PCA), and predicting in the form of a Variational Quantum Circuit (VQC). It is tested in various financial indicators, as BSE ESG, Carbon Exchange, Energy, Green Exchange, Oil and Gas and Power, with the help of RMSE and R 2 as the measure of performance. As experimental findings indicate, the classical algorithms like Random Forest and k-Nearest Neighbors always have the best accuracy and stability compared to quantum and hybrid ones. Quantum models are unstable, in contrast to Quantum kNN which is fairly stable. The hybrid model performs moderately with consistent error values but fairly lower explanatory power with significant results mostly on the BSE ESG data. In general, the paper has pointed out that classical approaches are still best in structured financial prediction tasks and hybrid quantum-classical models are potentially effective, but not yet competent. The results give valuable information on the existing limitation and future of the quantum machine learning to financial forecasting applications.

Frontiers in Artificial Intelligence•Sep 2, 2026•0 citations

Quantum vs. Classical Machine Learning: A Benchmark Study for Financial Prediction

Rehan Ahmad, Muhammad Kashif, Nouhaila Innan +1 more

In this paper, we present a reproducible benchmarking framework that systematically compares QML models with architecture-matched classical counterparts across three financial tasks: (i) directional return prediction on U.S. and Turkish equities, (ii) live-trading simulation with Quantum LSTMs versus classical LSTMs on the S\&P 500, and (iii) realized volatility forecasting using Quantum Support Vector Regression. By standardizing data splits, features, and evaluation metrics, our study provides a fair assessment of when current-generation QML models can match or exceed classical methods. Our results reveal that quantum approaches show performance gains when data structure and circuit design are well aligned. In directional classification, hybrid quantum neural networks surpass the parameter-matched ANN by \textbf{+3.8 AUC} and \textbf{+3.4 accuracy points} on \texttt{AAPL} stock and by \textbf{+4.9 AUC} and \textbf{+3.6 accuracy points} on Turkish stock \texttt{KCHOL}. In live trading, the QLSTM achieves higher risk-adjusted returns in \textbf{two of four} S\&P~500 regimes. For volatility forecasting, an angle-encoded QSVR attains the \textbf{lowest QLIKE} on \texttt{KCHOL} and remains within $\sim$0.02-0.04 QLIKE of the best classical kernels on \texttt{S\&P~500} and \texttt{AAPL}. Our benchmarking framework clearly identifies the scenarios where current QML architectures offer tangible improvements and where established classical methods continue to dominate.

cs.LGCross-listedarXiv:2601.03802•Jan 8, 2026•0 citations

Quantum Machine Learning Framework for Predicting Risk and Market Volatility

Sivashanmugam Thangavel, S. Murugan, N. Saranya Devi +3 more

Financial markets undergo rapid oscillations driven by several volatile causes, necessitating expedited and precise models for risk evaluation and market trend forecasting. Classical models encounter computational constraints when examining high-dimensional, nonlinear datasets. This study seeks to use Quantum Machine Learning (QML) methodologies to expedite risk evaluation and improve volatility forecasting for various financial instruments in real-time settings. The unique contribution of this work is the integration of quantum machine learning with financial risk assessment, leveraging quantum parallelism and kernelized feature spaces to improve the accuracy of market volatility prediction. This framework not only enhances computational efficiency compared to classical models but also provides a scalable foundation for real-time financial decision-making. The system employs Kaggle-style datasets including CME futures, Microsoft (MSFT), Amazon (AMZN), HRB, and historical data from the Dow Jones (DJI). Essential metrics including Open, Close, High, Low, Volume, Volatility Index, and Price Returns were retrieved. A hybrid pipeline combining conventional preprocessing with quantum-enhanced classifiers, including variational quantum circuits, was developed using the Qiskit and Pennylane frameworks. The QML-based model had a prediction accuracy of 95.2% in risk categorization and market direction forecasting, exceeding traditional standards. Quantum models provide significant benefits in velocity and scalability for financial analytics, enabling real-time risk assessment and strategic investment analysis.

International Conference on Smart Electronics and Communication (ICOSEC)•Sep 24, 2025•1 citation

Hybrid Supervised-unsupervised Learning Framework for Financial Data Analysis Enhanced by Quantum-inspired Computing

Praneeth Reddy Amudala Puchakayala, Poonam Ghule, Vijay Vaswani +1 more

Financial time-series analysis poses some difficulties, due to the non-stationarity of data, its distribution is heavy-tailed, the regime structure in financial time-series is latent, and there are few labelled training signals available.These models are highly sensitive to unexpected trade-offs and limited out-of-distribution generalisation, and unsupervised models, on their own, are unable to directly inform tradeable decisions.This paper proposes a framework named Quantum-Inspired Hybrid Supervised-Unsupervised Learning (QIHSL) consisting of three tightly-integrated modules: (1) Quantum-Inspired Feature Transformation (QFT) module that maps the financial features into amplitudeand phase-encoded qubit-state analogues, and generates the cross-feature interaction terms inspired by entanglement;(2) Quantum-Inspired K-Means (QI-KMeans) clustering module that uses the quantum fidelity based distance metric to learn the regime (bull, bear, sideways) of the data in an unsupervised manner; and (3) a Quantum-Inspired Support Vector Machine (QI-SVM) regime-gated ensemble for robust and regime-aware return-direction prediction with QI-SVMs that specialise in predicting the return direction in different market regimes.The experiments on a comprehensive benchmark dataset of 4,950 synthetic daily observations from the S&P 500 index, used to capture the statistical properties of equity markets, namely Student-t distributed returns (ν = 5), three-state regime switching and nineteen technical indicators, show that QIHSL achieves an accuracy of 0.4913, a precision of 0.5538, a recall of 0.2588, an F1-score of 0.3527 and an AUC of 0.5029 on the hold-out test set, outperforming the baseline SVM by 5.92 percentage points in precision.It represents the best of all six comparison methods (Table 2) and for all the architectures in the matched-condition comparison (Table 7) is an advantage (not a finding) of precision for two instances of the simulation and does not necessarily mean that they are more likely to forecast than others.The controlled ablation study (Table 3) also shows that a component-level benefit for quantum-inspired feature enrichment can be measured by the QFT module's accuracy of +2.0 pp and F1-score of +11.9 pp over the SVM baseline (A2 vs. A1).Five-fold stratified cross validation gives a stability of AUC = 0.4987 ± 0.0095.An ablation study shows that the QFT module contributes the largest single-step accuracy gain (+2.0 pp) and F1-score gain (+11.9 pp over A1); adding regime gating further improves accuracy (+0.95 pp) but reduces F1 (-0.0786) and AUC (-0.0123) relative to QFT alone.The reported AUC (around 0.5) is at chance level, and no pairwise McNemar comparison is significant at alpha = 0.05 (the smallest p = 0.2433), which is consistent with the almost efficient nature of the synthetic benchmark.Combined, results show no statistically relevant predictive value for the synthetic benchmark.The most interesting aspect of QIHSL is its highest precision (0.5538) of all methods compared, which could have some relevance to conservative trading, but not to proven practical forecast effectiveness.

International journal of intelligent engineering and systems•Oct 16, 2026•0 citations

ADVANCING FINANCIAL PREDICTION THROUGH QUANTUM MACHINE LEARNING

Ananya (Ph.D. Candidate) Patel

The growing complexity, interdependencies, and rapid fluctuations inherent in modern financial markets create substantial challenges for accurate forecasting, portfolio optimization, and risk management. Conventional machine learning techniques, while powerful, often face limitations in capturing nonlinear relationships and processing high-dimensional datasets efficiently. Quantum machine learning (QML) has emerged as a promising paradigm that leverages quantum computing principles to enhance predictive modeling in finance. This study presents a comprehensive investigation into the application of QML methods—including variational quantum circuits, quantum kernel estimation, and quantum-enhanced support vector machines—for financial time-series prediction and asset price classification. We propose a hybrid quantum-classical framework that integrates quantum feature mapping with classical optimizers to improve model expressiveness and convergence. Empirical experiments are conducted using historical stock market data and synthetic datasets to benchmark QML approaches against established classical models such as long short-term memory networks and gradient boosting machines. The results demonstrate that QML techniques can achieve superior prediction accuracy and lower computational latency under certain data regimes, particularly when dealing with small-to-medium-sized datasets and high feature correlations. Additionally, the study examines scalability considerations, hardware constraints of near-term quantum devices, and the interpretability of quantum model outputs in financial decision-making contexts. The findings underscore the transformative potential of quantum machine learning as an innovative tool for advancing predictive analytics in finance and provide practical insights into how financial institutions can begin integrating QML capabilities into their workflows.

International Journal of Intelligent Data and Machine Learning•Feb 1, 2025•2 citations

FD4QC: Application of Classical and Quantum-Hybrid Machine Learning for Financial Fraud Detection A Technical Report

Matteo Cardaioli, Luca Marangoni, Martini, Giada +5 more

The increasing complexity and volume of financial transactions pose significant challenges to traditional fraud detection systems. This technical report investigates and compares the efficacy of classical, quantum, and quantum-hybrid machine learning models for the binary classification of fraudulent financial activities. As of our methodology, first, we develop a comprehensive behavioural feature engineering framework to transform raw transactional data into a rich, descriptive feature set. Second, we implement and evaluate a range of models on the IBM Anti-Money Laundering (AML) dataset. The classical baseline models include Logistic Regression, Decision Tree, Random Forest, and XGBoost. These are compared against three hybrid classic quantum algorithms architectures: a Quantum Support Vector Machine (QSVM), a Variational Quantum Classifier (VQC), and a Hybrid Quantum Neural Network (HQNN). Furthermore, we propose Fraud Detection for Quantum Computing (FD4QC), a practical, API-driven system architecture designed for real-world deployment, featuring a classical-first, quantum-enhanced philosophy with robust fallback mechanisms. Our results demonstrate that classical tree-based models, particularly \textit{Random Forest}, significantly outperform the quantum counterparts in the current setup, achieving high accuracy (\(97.34\%\)) and F-measure (\(86.95\%\)). Among the quantum models, \textbf{QSVM} shows the most promise, delivering high precision (\(77.15\%\)) and a low false-positive rate (\(1.36\%\)), albeit with lower recall and significant computational overhead. This report provides a benchmark for a real-world financial application, highlights the current limitations of quantum machine learning in this domain, and outlines promising directions for future research.

arXiv:2507.19402•Jul 25, 2025•0 citations
Financial market prediction is a challenging application of machine learning, where even small improvements in directional accuracy can yield substantial value. Most models struggle to exceed 55--57% accuracy due to high noise, non-stationarity, and market efficiency. We introduce a hybrid ensemble framework combining quantum sentiment analysis, Decision Transformer architecture, and strategic model selection, achieving 60.14% directional accuracy on S&P 500 prediction, a 3.10% improvement over individual models. Our framework addresses three limitations of prior approaches. First, architecture diversity dominates dataset diversity: combining different learning algorithms (LSTM, Decision Transformer, XGBoost, Random Forest, Logistic Regression) on the same data outperforms training identical architectures on multiple datasets (60.14% vs.\ 52.80%), confirmed by correlation analysis (r>0.6r>0.6r>0.6 among same-architecture models). Second, a 4-qubit variational quantum circuit enhances sentiment analysis, providing +0.8% to +1.5% gains per model. Third, smart filtering excludes weak predictors (accuracy <52%<52\%<52%), improving ensemble performance (Top-7 models: 60.14% vs.\ all 35 models: 51.2%). We evaluate on 2020--2023 market data across seven instruments, covering diverse regimes including the COVID-19 crash and inflation-driven correction. McNemar's test confirms statistical significance (p<0.05p<0.05p<0.05). Preliminary backtesting with confidence-based filtering (6+ model consensus) yields a Sharpe ratio of 1.2 versus buy-and-hold's 0.8, demonstrating practical trading potential.
The forecasting of financial time series has gained more significance in decision making within a dynamic economic setting. Over the last few years, there has been a push in exploring both classical machine learning methods and novel quantum machine learning models with a view to enhance predictive accuracy. This paper suggests a quantum-classical regression framework developed that combines the feature selection in the form of a Random Forest in a hybrid approach, dimensionality reduction in the mode of Principal Component Analysis (PCA), and predicting in the form of a Variational Quantum Circuit (VQC). It is tested in various financial indicators, as BSE ESG, Carbon Exchange, Energy, Green Exchange, Oil and Gas and Power, with the help of RMSE and R 2 as the measure of performance. As experimental findings indicate, the classical algorithms like Random Forest and k-Nearest Neighbors always have the best accuracy and stability compared to quantum and hybrid ones. Quantum models are unstable, in contrast to Quantum kNN which is fairly stable. The hybrid model performs moderately with consistent error values but fairly lower explanatory power with significant results mostly on the BSE ESG data. In general, the paper has pointed out that classical approaches are still best in structured financial prediction tasks and hybrid quantum-classical models are potentially effective, but not yet competent. The results give valuable information on the existing limitation and future of the quantum machine learning to financial forecasting applications.
In this paper, we present a reproducible benchmarking framework that systematically compares QML models with architecture-matched classical counterparts across three financial tasks: (i) directional return prediction on U.S. and Turkish equities, (ii) live-trading simulation with Quantum LSTMs versus classical LSTMs on the S\&P 500, and (iii) realized volatility forecasting using Quantum Support Vector Regression. By standardizing data splits, features, and evaluation metrics, our study provides a fair assessment of when current-generation QML models can match or exceed classical methods. Our results reveal that quantum approaches show performance gains when data structure and circuit design are well aligned. In directional classification, hybrid quantum neural networks surpass the parameter-matched ANN by \textbf{+3.8 AUC} and \textbf{+3.4 accuracy points} on \texttt{AAPL} stock and by \textbf{+4.9 AUC} and \textbf{+3.6 accuracy points} on Turkish stock \texttt{KCHOL}. In live trading, the QLSTM achieves higher risk-adjusted returns in \textbf{two of four} S\&P~500 regimes. For volatility forecasting, an angle-encoded QSVR attains the \textbf{lowest QLIKE} on \texttt{KCHOL} and remains within $\sim$0.02-0.04 QLIKE of the best classical kernels on \texttt{S\&P~500} and \texttt{AAPL}. Our benchmarking framework clearly identifies the scenarios where current QML architectures offer tangible improvements and where established classical methods continue to dominate.
Financial markets undergo rapid oscillations driven by several volatile causes, necessitating expedited and precise models for risk evaluation and market trend forecasting. Classical models encounter computational constraints when examining high-dimensional, nonlinear datasets. This study seeks to use Quantum Machine Learning (QML) methodologies to expedite risk evaluation and improve volatility forecasting for various financial instruments in real-time settings. The unique contribution of this work is the integration of quantum machine learning with financial risk assessment, leveraging quantum parallelism and kernelized feature spaces to improve the accuracy of market volatility prediction. This framework not only enhances computational efficiency compared to classical models but also provides a scalable foundation for real-time financial decision-making. The system employs Kaggle-style datasets including CME futures, Microsoft (MSFT), Amazon (AMZN), HRB, and historical data from the Dow Jones (DJI). Essential metrics including Open, Close, High, Low, Volume, Volatility Index, and Price Returns were retrieved. A hybrid pipeline combining conventional preprocessing with quantum-enhanced classifiers, including variational quantum circuits, was developed using the Qiskit and Pennylane frameworks. The QML-based model had a prediction accuracy of 95.2% in risk categorization and market direction forecasting, exceeding traditional standards. Quantum models provide significant benefits in velocity and scalability for financial analytics, enabling real-time risk assessment and strategic investment analysis.
Financial time-series analysis poses some difficulties, due to the non-stationarity of data, its distribution is heavy-tailed, the regime structure in financial time-series is latent, and there are few labelled training signals available.These models are highly sensitive to unexpected trade-offs and limited out-of-distribution generalisation, and unsupervised models, on their own, are unable to directly inform tradeable decisions.This paper proposes a framework named Quantum-Inspired Hybrid Supervised-Unsupervised Learning (QIHSL) consisting of three tightly-integrated modules: (1) Quantum-Inspired Feature Transformation (QFT) module that maps the financial features into amplitudeand phase-encoded qubit-state analogues, and generates the cross-feature interaction terms inspired by entanglement;(2) Quantum-Inspired K-Means (QI-KMeans) clustering module that uses the quantum fidelity based distance metric to learn the regime (bull, bear, sideways) of the data in an unsupervised manner; and (3) a Quantum-Inspired Support Vector Machine (QI-SVM) regime-gated ensemble for robust and regime-aware return-direction prediction with QI-SVMs that specialise in predicting the return direction in different market regimes.The experiments on a comprehensive benchmark dataset of 4,950 synthetic daily observations from the S&P 500 index, used to capture the statistical properties of equity markets, namely Student-t distributed returns (ν = 5), three-state regime switching and nineteen technical indicators, show that QIHSL achieves an accuracy of 0.4913, a precision of 0.5538, a recall of 0.2588, an F1-score of 0.3527 and an AUC of 0.5029 on the hold-out test set, outperforming the baseline SVM by 5.92 percentage points in precision.It represents the best of all six comparison methods (Table 2) and for all the architectures in the matched-condition comparison (Table 7) is an advantage (not a finding) of precision for two instances of the simulation and does not necessarily mean that they are more likely to forecast than others.The controlled ablation study (Table 3) also shows that a component-level benefit for quantum-inspired feature enrichment can be measured by the QFT module's accuracy of +2.0 pp and F1-score of +11.9 pp over the SVM baseline (A2 vs. A1).Five-fold stratified cross validation gives a stability of AUC = 0.4987 ± 0.0095.An ablation study shows that the QFT module contributes the largest single-step accuracy gain (+2.0 pp) and F1-score gain (+11.9 pp over A1); adding regime gating further improves accuracy (+0.95 pp) but reduces F1 (-0.0786) and AUC (-0.0123) relative to QFT alone.The reported AUC (around 0.5) is at chance level, and no pairwise McNemar comparison is significant at alpha = 0.05 (the smallest p = 0.2433), which is consistent with the almost efficient nature of the synthetic benchmark.Combined, results show no statistically relevant predictive value for the synthetic benchmark.The most interesting aspect of QIHSL is its highest precision (0.5538) of all methods compared, which could have some relevance to conservative trading, but not to proven practical forecast effectiveness.
The growing complexity, interdependencies, and rapid fluctuations inherent in modern financial markets create substantial challenges for accurate forecasting, portfolio optimization, and risk management. Conventional machine learning techniques, while powerful, often face limitations in capturing nonlinear relationships and processing high-dimensional datasets efficiently. Quantum machine learning (QML) has emerged as a promising paradigm that leverages quantum computing principles to enhance predictive modeling in finance. This study presents a comprehensive investigation into the application of QML methods—including variational quantum circuits, quantum kernel estimation, and quantum-enhanced support vector machines—for financial time-series prediction and asset price classification. We propose a hybrid quantum-classical framework that integrates quantum feature mapping with classical optimizers to improve model expressiveness and convergence. Empirical experiments are conducted using historical stock market data and synthetic datasets to benchmark QML approaches against established classical models such as long short-term memory networks and gradient boosting machines. The results demonstrate that QML techniques can achieve superior prediction accuracy and lower computational latency under certain data regimes, particularly when dealing with small-to-medium-sized datasets and high feature correlations. Additionally, the study examines scalability considerations, hardware constraints of near-term quantum devices, and the interpretability of quantum model outputs in financial decision-making contexts. The findings underscore the transformative potential of quantum machine learning as an innovative tool for advancing predictive analytics in finance and provide practical insights into how financial institutions can begin integrating QML capabilities into their workflows.
The increasing complexity and volume of financial transactions pose significant challenges to traditional fraud detection systems. This technical report investigates and compares the efficacy of classical, quantum, and quantum-hybrid machine learning models for the binary classification of fraudulent financial activities. As of our methodology, first, we develop a comprehensive behavioural feature engineering framework to transform raw transactional data into a rich, descriptive feature set. Second, we implement and evaluate a range of models on the IBM Anti-Money Laundering (AML) dataset. The classical baseline models include Logistic Regression, Decision Tree, Random Forest, and XGBoost. These are compared against three hybrid classic quantum algorithms architectures: a Quantum Support Vector Machine (QSVM), a Variational Quantum Classifier (VQC), and a Hybrid Quantum Neural Network (HQNN). Furthermore, we propose Fraud Detection for Quantum Computing (FD4QC), a practical, API-driven system architecture designed for real-world deployment, featuring a classical-first, quantum-enhanced philosophy with robust fallback mechanisms. Our results demonstrate that classical tree-based models, particularly \textit{Random Forest}, significantly outperform the quantum counterparts in the current setup, achieving high accuracy (97.34%97.34\%97.34%) and F-measure (86.95%86.95\%86.95%). Among the quantum models, \textbf{QSVM} shows the most promise, delivering high precision (77.15%77.15\%77.15%) and a low false-positive rate (1.36%1.36\%1.36%), albeit with lower recall and significant computational overhead. This report provides a benchmark for a real-world financial application, highlights the current limitations of quantum machine learning in this domain, and outlines promising directions for future research.