DEEP LEARNING APPROACHES TO STOCHASTIC VOLATILITY MODEL CALIBRATION: A COMPARATIVE ANALYSIS OF NEURAL SDES AND TRADITIONAL METHODS
Volume 2, Issue 4, Pp 13-24, 2025
DOI: https://doi.org/10.61784/jtfe3061
Author(s)
YiFan Zhang*, Wei Chen, Oliver Schmidt
Affiliation(s)
Department of Industrial & Enterprise Systems Engineering, University of Illinois Urbana-Champaign, Illinois, USA.
Corresponding Author
YiFan Zhang
ABSTRACT
The calibration of stochastic volatility models remains a computationally demanding challenge in quantitative finance, where traditional optimization algorithms often encounter difficulties with numerical stability, convergence speed, and local minima entrapment. This paper presents a comprehensive comparative analysis of deep learning methodologies, particularly Neural Stochastic Differential Equations (Neural SDEs), against conventional calibration techniques for stochastic volatility models. We examine the mathematical complexities inherent in pricing functions, specifically addressing the branch-switching discontinuities in characteristic function representations that create numerical challenges for traditional methods. Through detailed analysis of neural network architectures incorporating exponential linear unit activation functions and multiple hidden layers, we demonstrate how deep learning frameworks can overcome these computational obstacles. Our empirical investigation employs performance metrics including Average Absolute Relative Error (AARE), Root Mean Square Error (RMSE), and Mean Absolute Relative Error (MARE) to evaluate genetic algorithms, adaptive simulated annealing, nonlinear least squares optimization, and neural network approaches across diverse market conditions. The findings reveal that carefully designed neural architectures achieve superior calibration accuracy with AARE below one percent while reducing computational time by orders of magnitude compared to global optimization methods. Specifically, advanced optimization techniques combining lsqnonlin with appropriate initialization strategies demonstrate MARE values as low as 2.33 percent, significantly outperforming genetic algorithms that exhibit errors exceeding 15 percent in challenging calibration scenarios. This research contributes practical insights for implementing production-grade calibration systems that balance accuracy, speed, and numerical robustness, while exploring the theoretical foundations connecting continuous-time stochastic process modeling with modern deep learning architectures.
KEYWORDS
Stochastic volatility models; Neural networks; Heston model calibration; Characteristic function; Branch switching; Deep learning; Optimization algorithms; Exponential linear units; Model calibration; Computational finance
CITE THIS PAPER
YiFan Zhang, Wei Chen, Oliver Schmidt. Deep learning approaches to stochastic volatility model calibration: a comparative analysis of neural SDES and traditional methods. Journal of Trends in Financial and Economics. 2025, 2(4): 13-24. DOI: https://doi.org/10.61784/jtfe3061.
REFERENCES
[1] Nafiu A, Balogun S O, Oko-Odion C, et al. Risk management strategies: Navigating volatility in complex financial market environments. World Journal of Advanced Research and Reviews, 2025, 25(1): 236-250.
[2] Ma Z, Chen X, Sun T, et al. Blockchain-based zero-trust supply chain security integrated with deep reinforcement learning for inventory optimization. Future Internet, 2024, 16(5): 163.
[3] Sun T, Yang J, Li J, et al. Enhancing auto insurance risk evaluation with transformer and SHAP. IEEE Access, 2024.
[4] Patawari A, Das P. From traditional to computationally efficient scientific computing algorithms in option pricing: Current progresses with future directions. Archives of Computational Methods in Engineering, 2025: 1-40.
[5] Li Q, Jiang H, Lee C H. Phase‐Space Generalized Brillouin Zone For Spatially Inhomogeneous Non‐Hermitian Systems. Advanced Science, 2025: e08047.
[6] Zhang Y, Zhang H. Enhancing robot path planning through a twin-reinforced chimp optimization algorithm and evolutionary programming algorithm. IEEE Access, 2025, 12: 170057-170078.
[7] Gilli M, Maringer D, Schumann E. Numerical methods and optimization in finance. Academic Press, 2019.
[8] Devan M, Thirunavukkarasu K, Shanmugam L. Algorithmic trading strategies: real-time data analytics with machine learning. Journal of Knowledge Learning and Science Technology, 2023, 2(3): 522-546.
[9] Khalili H. Spatial competition of learning agents in agricultural procurement markets. Doctoral dissertation, Universitats-und Landesbibliothek Bonn, 2019.
[10] Nikolenko S I. Synthetic data for deep learning. Cham, Switzerland: Springer, 2021, 174.
[11] Ge Y, Wang Y, Liu J, et al. GAN-Enhanced Implied Volatility Surface Reconstruction for Option Pricing Error Mitigation. IEEE Access, 2025.
[12] Hirsa A. Computational methods in finance. Chapman and Hall/CRC, 2024.
[13] Sabbar Y. A Review of Modern Stochastic Modeling: SDE/SPDE Numerics, Data-Driven Identification, and Generative Methods with Applications in Biology and Epidemiology. arXiv preprint arXiv:2508.11004, 2025.
[14] Shlezinger N, Whang J, Eldar Y C, et al. Model-based deep learning. Proceedings of the IEEE, 2023, 111(5): 465-499.
[15] Rohilla A. Exploring Volatility: Evolution, Advancements, Trends, and Applications. 2024.
[16] Alòs E, León J A. An intuitive introduction to fractional and rough volatilities. Mathematics, 2021, 9(9): 994.
[17] Bajaj P, Kaur J. Recent Developments and Applicability of the Black and Scholes Model in Option Pricing: A Literature Review. Mudra: Journal of Finance & Accounting, 2020, 7(2).
[18] Kristensen S. High-Performing Computational Methods for American Option Pricing and Implied Volatility Surface Construction. 2025.
[19] Cao W, Mai N T, Liu W. Adaptive knowledge assessment via symmetric hierarchical Bayesian neural networks with graph symmetry-aware concept dependencies. Symmetry, 2025, 17(8): 1332.
[20] Mai N T, Cao W, Liu W. Interpretable knowledge tracing via transformer-Bayesian hybrid networks: Learning temporal dependencies and causal structures in educational data. Applied Sciences, 2025, 15(17): 9605.
[21] Liu S, Borovykh A, Grzelak L A, et al. A neural network-based framework for financial model calibration. Journal of Mathematics in Industry, 2019, 9(1): 9.
[22] Chen S, Liu Y, Zhang Q, et al. Multi‐Distance Spatial‐Temporal Graph Neural Network for Anomaly Detection in Blockchain Transactions. Advanced Intelligent Systems, 2025: 2400898.
[23] Zhang Q, Chen S, Liu W. Balanced Knowledge Transfer in MTTL-ClinicalBERT: A Symmetrical Multi-Task Learning Framework for Clinical Text Classification. Symmetry, 2025, 17(6): 823.
[24] Ren S, Jin J, Niu G, et al. ARCS: Adaptive Reinforcement Learning Framework for Automated Cybersecurity Incident Response Strategy Optimization. Applied Sciences, 2025, 15(2): 951.
[25] Moradinia Z, Vandierendonck H, Murphy A. Navigating speed–accuracy trade-offs for multi-physics simulations. Meccanica, 2025, 60(6): 1583-1599.
[26] Mai N T, Cao W, Wang Y. The global belonging support framework: Enhancing equity and access for international graduate students. Journal of International Students, 2025, 15(9): 141-160.
[27] Liu Y, Ren S, Wang X, et al. Temporal logical attention network for log-based anomaly detection in distributed systems. Sensors, 2024, 24(24): 7949.
[28] Tan Y, Wu B, Cao J, et al. LLaMA-UTP: Knowledge-Guided Expert Mixture for Analyzing Uncertain Tax Positions. IEEE Access, 2025.
[29] Zheng W, Liu W. Symmetry-Aware Transformers for Asymmetric Causal Discovery in Financial Time Series. Symmetry, 2025, 17(10): 1591.
[30] Ivascu C F. Option pricing using machine learning. Expert Systems with Applications, 2021, 163: 113799.
[31] Hu X, Zhao X, Wang J, et al. Information-theoretic multi-scale geometric pre-training for enhanced molecular property prediction. PLoS One, 2025, 20(10): e0332640.
[32] Zhang H, Ge Y, Zhao X, et al. Hierarchical deep reinforcement learning for multi-objective integrated circuit physical layout optimization with congestion-aware reward shaping. IEEE Access, 2025.
[33] Wang J, Zhang H, Wu B, et al. Symmetry-Guided Electric Vehicles Energy Consumption Optimization Based on Driver Behavior and Environmental Factors: A Reinforcement Learning Approach. Symmetry, 17(6), 930.
[34] Hu, X., Zhao, X., & Liu, W. (2025). Hierarchical Sensing Framework for Polymer Degradation Monitoring: A Physics-Constrained Reinforcement Learning Framework for Programmable Material Discovery. Sensors, 2025, 25(14): 4479.
[35] Wang Y, Ding G, Zeng Z, et al. Causal-Aware Multimodal Transformer for Supply Chain Demand Forecasting: Integrating Text, Time Series, and Satellite Imagery. IEEE Access, 2025.
[36] Han X, Yang Y, Chen J, et al. Symmetry-Aware Credit Risk Modeling: A Deep Learning Framework Exploiting Financial Data Balance and Invariance. Symmetry, 2025, 17(3): 20738994.

Download as PDF