Systematic comparison of deep generative models applied to multivariate financial time series

ArXiv ID: 2412.06417 “View on arXiv”

Authors: Unknown

Abstract

Financial time series (FTS) generation models are a core pillar to applications in finance. Risk management and portfolio optimization rely on realistic multivariate price generation models. Accordingly, there is a strong modelling literature dating back to Bachelier’s Theory of Speculation in 1901. Generating FTS using deep generative models (DGMs) is still in its infancy. In this work, we systematically compare DGMs against state-of-the-art parametric alternatives for multivariate FTS generation. We initially compare both DGMs and parametric models over increasingly complex synthetic datasets. The models are evaluated through distance measures for varying distribution moments of both the full and rolling FTS. We then apply the best performing DGM models to empirical data, demonstrating the benefit of DGMs through a implied volatility trading task.

Keywords: Deep Generative Models (DGMs), Financial Time Series, Multivariate Generation, Implied Volatility Trading, Parametric Models, Equities

Complexity vs Empirical Score

• Math Complexity: 7.5/10
• Empirical Rigor: 7.0/10
• Quadrant: Holy Grail
• Why: The paper presents advanced mathematical concepts such as multivariate GARCH decompositions, factor stochastic volatility models, and various deep generative model architectures, indicating high math complexity. It also demonstrates strong empirical rigor through systematic comparisons on synthetic and empirical datasets, evaluating multiple models and applying the best to a specific trading task (implied volatility trading), which shows backtest-ready and implementation-heavy work.

  flowchart TD
    A["Research Goal: Systematically Compare DGMs vs Parametric Models for Multivariate FTS Generation"] --> B["Methodology: 2-Phase Experiment"]
    B --> C["Phase 1: Synthetic Data"]
    C --> D["Computational Process: Evaluate DGMs & Parametric Models<br>via Distance Measures on Distribution Moments"]
    B --> E["Phase 2: Empirical Data"]
    E --> D
    D --> F["Outcomes: DGMs Surpass Parametric Models"]
    F --> G["Applied Outcome: Successful Implied Volatility Trading"]