Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

HireVAE: An Online and Adaptive Factor Model Based on Hierarchical and Regime-Switch VAE

Zikai Wei, Anyi Rao, Bo Dai, Dahua Lin

TL;DR

HireVAE introduces an online and adaptive regime-switching factor model for stock prediction that jointly learns a hierarchical market-stock latent space and regime-specific decoders. By projecting a market latent variable to a 1D regime score and applying a linear stabilization mechanism, the method maintains consistent regime centers while adapting to current market conditions. The approach achieves superior predictive power and portfolio performance across multiple Chinese market benchmarks, and ablation studies show that both the hierarchy and online regime learning are essential. This work offers a practical framework for real-time, regime-aware factor investing with end-to-end training and online inference capabilities.

Abstract

Factor model is a fundamental investment tool in quantitative investment, which can be empowered by deep learning to become more flexible and efficient in practical complicated investing situations. However, it is still an open question to build a factor model that can conduct stock prediction in an online and adaptive setting, where the model can adapt itself to match the current market regime identified based on only point-in-time market information. To tackle this problem, we propose the first deep learning based online and adaptive factor model, HireVAE, at the core of which is a hierarchical latent space that embeds the underlying relationship between the market situation and stock-wise latent factors, so that HireVAE can effectively estimate useful latent factors given only historical market information and subsequently predict accurate stock returns. Across four commonly used real stock market benchmarks, the proposed HireVAE demonstrate superior performance in terms of active returns over previous methods, verifying the potential of such online and adaptive factor model.

HireVAE: An Online and Adaptive Factor Model Based on Hierarchical and Regime-Switch VAE

TL;DR

HireVAE introduces an online and adaptive regime-switching factor model for stock prediction that jointly learns a hierarchical market-stock latent space and regime-specific decoders. By projecting a market latent variable to a 1D regime score and applying a linear stabilization mechanism, the method maintains consistent regime centers while adapting to current market conditions. The approach achieves superior predictive power and portfolio performance across multiple Chinese market benchmarks, and ablation studies show that both the hierarchy and online regime learning are essential. This work offers a practical framework for real-time, regime-aware factor investing with end-to-end training and online inference capabilities.

Abstract

Factor model is a fundamental investment tool in quantitative investment, which can be empowered by deep learning to become more flexible and efficient in practical complicated investing situations. However, it is still an open question to build a factor model that can conduct stock prediction in an online and adaptive setting, where the model can adapt itself to match the current market regime identified based on only point-in-time market information. To tackle this problem, we propose the first deep learning based online and adaptive factor model, HireVAE, at the core of which is a hierarchical latent space that embeds the underlying relationship between the market situation and stock-wise latent factors, so that HireVAE can effectively estimate useful latent factors given only historical market information and subsequently predict accurate stock returns. Across four commonly used real stock market benchmarks, the proposed HireVAE demonstrate superior performance in terms of active returns over previous methods, verifying the potential of such online and adaptive factor model.
Paper Structure (33 sections, 10 equations, 2 figures, 5 tables, 2 algorithms)

This paper contains 33 sections, 10 equations, 2 figures, 5 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Factor Models.
  4. Regime Switching
  5. Variational Autoencoder.
  6. Preliminaries
  7. Methodology
  8. Extracting Features from Sequential Data
  9. Stock Features.
  10. Market Features.
  11. Hierarchical Encoder-Decoder Structure
  12. Posterior Structure in Training Phase
  13. Market Encoder.
  14. Market-Aware Stock Encoder.
  15. Regime-Switch Factor Decoder.
  16. ...and 18 more sections

Figures (2)

  • Figure 1: Identifying market regimes can help make better investment decisions, and there is an example that identifies regime changes based on volatility, where volatility acts as an experience-and-knowledge based indicator that is widely used in practice.
  • Figure 2: The brief architecture of HireVAE.

Theorems & Definitions (2)

  • Example 1: Determining regimes helps better decision making
  • Definition 1: Stock prediction