Newest papers

Regime Detection in Latent Factor Models -- A Bayesian Decision-Theoretic Approach

Karl Kristian Engelund · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.20366 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Model multi-faktor yang digunakan dalam keuangan kuantitatif, tetapi filtrasi sendiri tidak menentukan kapan agen harus bertindak pada perubahan rezim yang diduga. Kami merumuskan deteksi rezim online dengan faktor latent dan rezim tersembunyi sebagai masalah keputusan Bayesian yang menyeimbangkan alarm palsu, keterlambatan deteksi, dan misclassification. Karena probabilitas rezim belakang umumnya tidak-Markovian, masalah ini diwakili sebagai masalah penghentian optimal yang diukur dan disimulasikan menggunakan filtrasi partikel dengan pemrograman dinamis Monte Carlo. Model Heston satu titik perubahan menyediakan benchmark. Dengan varians yang diamati, masalah ini berkurang menjadi masalah batas bebas Markov dua dimensi, sementara di bawah latens varian keputusan didasarkan pada ukuran filtrasi posterior. Eksperimen numerik dalam model Heston dan Vasek yang menggantikan rezim menunjukkan bahwa

Machine translation (Meta M2M100); the English original follows.

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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Particle filter, Bayesian probability, Markov chain Monte Carlo, Monte Carlo method, Variance (accounting), Computer science

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange