Reproduce “Improving on the Markov-Switching Regression Model by the Use of an Adaptive Moving Average” on OpenResearch

About this paper

Piotr Pomorski, Denise Gorse

The KAMA+MSR model integrates Kaufman’s adaptive moving average with a two-state Markov-switching regression model to classify financial markets into four distinct regimes based on both volatility and trend. This enhanced model demonstrates superior out-of-sample trading performance, particularly in equities and fixed income, compared to traditional MSR benchmarks, offering more stable and nuanced market segmentation.

Key insights

• Combining a volatility-focused MSR with a trend-adaptive KAMA allows for a more comprehensive and stable four-state market segmentation that accounts for both volatility and directional trend, overcoming the limitations of single-factor models.
• The KAMA+MSR model provides a more robust and less volatile identification of market regimes compared to multi-state MSR extensions, which typically exhibit instability and frequent switches.
• Focusing trading strategies on specific, opportune market regimes (low variance bullish and high variance bearish) identified by the combined model can lead to improved risk-adjusted returns and overall profitability.

Summary by alphaXiv.