Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic Investment Strategies Through Market Classification and Volatility: A Machine Learning Approach

Jinhui Li, Wenjia Xie, Luis Seco

TL;DR

A dynamic investment framework to enhance portfolio management in volatile markets, offering clear advantages over traditional static strategies, outperforming most static methods.

Abstract

This study introduces a dynamic investment framework to enhance portfolio management in volatile markets, offering clear advantages over traditional static strategies. Evaluates four conventional approaches : equal weighted, minimum variance, maximum diversification, and equal risk contribution under dynamic conditions. Using K means clustering, the market is segmented into ten volatility-based states, with transitions forecasted by a Bayesian Markov switching model employing Dirichlet priors and Gibbs sampling. This enables real-time asset allocation adjustments. Tested across two asset sets, the dynamic portfolio consistently achieves significantly higher risk-adjusted returns and substantially higher total returns, outperforming most static methods. By integrating classical optimization with machine learning and Bayesian techniques, this research provides a robust strategy for optimizing investment outcomes in unpredictable market environments.

Dynamic Investment Strategies Through Market Classification and Volatility: A Machine Learning Approach

TL;DR

A dynamic investment framework to enhance portfolio management in volatile markets, offering clear advantages over traditional static strategies, outperforming most static methods.

Abstract

This study introduces a dynamic investment framework to enhance portfolio management in volatile markets, offering clear advantages over traditional static strategies. Evaluates four conventional approaches : equal weighted, minimum variance, maximum diversification, and equal risk contribution under dynamic conditions. Using K means clustering, the market is segmented into ten volatility-based states, with transitions forecasted by a Bayesian Markov switching model employing Dirichlet priors and Gibbs sampling. This enables real-time asset allocation adjustments. Tested across two asset sets, the dynamic portfolio consistently achieves significantly higher risk-adjusted returns and substantially higher total returns, outperforming most static methods. By integrating classical optimization with machine learning and Bayesian techniques, this research provides a robust strategy for optimizing investment outcomes in unpredictable market environments.

Paper Structure

This paper contains 31 sections, 28 equations, 7 figures, 10 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. K-Means Clustering for Market Segmentation
  4. Bayesian Markov Switching Model Using Bayesian Transition Matrix
  5. Mixing Time
  6. Portfolio Methods
  7. Equally-Weighted Investment
  8. Minimum Variance Portfolio
  9. Maximum Diversification
  10. Equal Risk Contribution (ERC) Portfolio
  11. Performance Evaluation Methods
  12. Empirical Implementation and Results
  13. Implementation
  14. Market State Classification
  15. Investment Strategies
  16. ...and 16 more sections

Figures (7)

  • Figure 1: K-means Clustering of 22-Day Volatility into 10 States for SPY Top 11 Portfolio
  • Figure 2: K-means Clustering of 22-Day Volatility into 10 States for Second Asset Portfolio
  • Figure 3: Bayesian Markov Transition Matrix for the First Asset Set
  • Figure 4: Investment Value Over Time for the First Asset Set with an Initial Investment of $1
  • Figure 5: Yearly Sharpe Ratio for the First Asset Set
  • ...and 2 more figures