Recurrent neural network based parameter estimation of Hawkes model on high-frequency financial data

Paper: arXiv 2304.11883

Authors: Kyungsub Lee

Abstract

This study examines the use of a recurrent neural network for estimating the parameters of a Hawkes model based on high-frequency financial data, and subsequently, for computing volatility. Neural networks have shown promising results in various fields, and interest in finance is also growing. Our approach demonstrates significantly faster computational performance compared to traditional maximum likelihood estimation methods while yielding comparable accuracy in both simulation and empirical studies. Furthermore, we demonstrate the application of this method for real-time volatility measurement, enabling the continuous estimation of financial volatility as new price data keeps coming from the market.

Complexity vs Empirical Score

• Math Complexity: 6.5/10
• Empirical Rigor: 6.5/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces an innovative application of recurrent neural networks to bypass computationally expensive Maximum Likelihood Estimation (MLE) for Hawkes processes in high-frequency regimes. It balances point-process mathematical foundations with practical empirical backtesting on financial data, showing significant speedups for real-time volatility estimation.

Research Flowchart

  flowchart TD
    A["Research Goal: Real-Time Hawkes Parameter & Volatility Estimation"] --> B["Data Inputs: High-Frequency Financial Data & Simulated Point Processes"]
    B --> C["Methodology: Train Recurrent Neural Network (RNN) Model"]
    C --> D["Process: RNN vs. Traditional Maximum Likelihood Estimation (MLE)"]
    D --> E["Outcome 1: Significantly Faster Computational Speed vs. MLE"]
    D --> F["Outcome 2: Comparable Accuracy in Parameter Estimation"]
    E & F --> G["Final Application: Real-Time Continuous Volatility Measurement"]