Evaluating the Building Blocks of a Dynamically Adaptive Systematic Trading Strategy

TL;DR

This paper proposes a dynamically adaptive trading system that uses State Switching Markov Autoregressive models to identify market regimes and tailor trading strategies accordingly, aiming to improve performance over traditional methods.

Contribution

It introduces a novel approach combining regime detection with strategy adaptation using Markov models, enhancing robustness in volatile markets.

Findings

Effective identification of market regimes using Markov models

Tailored trading strategies improve performance in different regimes

Enhanced robustness over traditional static strategies

Abstract

Financial markets change their behaviours abruptly. The mean, variance and correlation patterns of stocks can vary dramatically, triggered by fundamental changes in macroeconomic variables, policies or regulations. A trader needs to adapt her trading style to make the best out of the different phases in the stock markets. Similarly, an investor might want to invest in different asset classes in different market regimes for a stable risk adjusted return profile. Here, we explore the use of State Switching Markov Autoregressive models for identifying and predicting different market regimes loosely modeled on the Wyckoff Price Regimes of accumulation, distribution, advance and decline. We explore the behaviour of various asset classes and market sectors in the identified regimes. We look at the trading strategies like trend following, range trading, retracement trading and breakout trading in the given market regimes and tailor them for the specific regimes. We tie together the best trading strategy and asset allocation for the identified market regimes to come up with a robust dynamically adaptive trading system to outperform simple traditional alphas.