Non-Stationarity

Non-stationarity is the failure mode in which the statistical properties of a market — volatility, liquidity, correlations — change over time, so a Markov Decision Process estimated with a fixed transition kernel and reward becomes mis-specified, and a policy optimised for past conditions degrades when conditions shift. It appears in this vault because Lalor & Swishchuk 2025 warn that an MDP-based strategy should be tested “in different types of market regimes, in particular regimes that were unseen in the training data, as this could significantly alter the out-of-sample results.”

Connections

• Markov Decision Process Trading Model — suffers_overfitting_risk, source: https://arxiv.org/html/2410.14504v2
• Reinforcement Learning Trading Policy — suffers_overfitting_risk, source: https://arxiv.org/abs/2109.13851
• Sim-to-Real Gap — relates, source: https://arxiv.org/abs/2109.13851