Rating
1966
Battle Count: 50
Relevance
9/10
Highly relevant for quantitative traders and risk managers who need to decide *when* to act on regime changes rather than just estimating them. It provides a rigorous framework for balancing the costs of acting too early (false alarms) vs. too late (detection delay) in stochastic volatility and interest rate models.
Implementation Complexity
8/10
Implementing PFMCDP requires expertise in stochastic filtering, Monte Carlo methods, and regression techniques. It involves simulating paths, running particle filters, and performing backward induction with regression, which is more complex than standard Kalman filtering or simple threshold rules.
Reproducibility
3/5
The paper provides detailed mathematical formulations, algorithm steps (PFMCDP), and specific parameter settings for numerical experiments (Section 5). However, no public code repository URL is provided in the text. The data is simulated, and the simulation parameters are fully specified, allowing for reproduction if the algorithm is implemented from scratch based on the description.
Abstract
Regime-switching multi-factor models are widely used in quantitative finance, but filtering does not by itself determine when an agent should act on inferred regime changes. We formulate online regime detection with latent factors and hidden regimes as a Bayesian decision problem balancing false alarms, detection delay, and misclassification. Since posterior regime probabilities are generally non-Markovian, the problem is represented as a measure-valued optimal stopping problem and approximated using particle filtering with Monte Carlo dynamic programming. A single-change-point Heston model provides a benchmark. With observed variance, the problem reduces to a two-dimensional Markov free-boundary problem, while under latent variance decisions are based on the filtered posterior measure. Numerical experiments in Heston and regime-switching Vasicek models show that the latent-information procedure can outperform feasible plug-in rules.
The interactive Everscope explorer (charts, battles, favorites) loads below.