Detecting Regime Change in Finance: A Framework with Hidden Markov Model

Abstract

Regime detection plays a vital role in understanding the dynamic nature of financial markets. This becomes especially crucial during times of political or economic turbulence, where shifts in market sentiment can occur rapidly. In financial markets, the conventional method of summarizing prices involves creating time series data, where transaction prices are sampled at fixed time intervals. This study proposes a novel approach called Directional Change, which sample market prices at peaks and troughs. Unlike conventional methods, DC allows data- driven sampling, capturing price changes as they occur. The study presents novel indicators for extracting valuable information from DC-recorded data, providing a new perspective on analyzing market dynamics and detecting regime changes with the help of Hidden Markov Model. After identifying regimes,the study extends to other assets, where the regimes are posi- tioned in a 2D indicator space to visualize their relative positions and assess whether regimes from different markets share similar statistical properties. This study presents a new method- ology for detecting regime shifts in the markets, offering valuable insights for market analysis and tracking.