regime-switching

Statistical Jump Models in Python, with scikit-learn-style APIs

MarSwitching.jl: Julia package for Markov switching dynamic models 📈

Fit autoregressive models with skewed generalized error distribution (SGED) noise whose parameters vary with level

Semi-analytical dynamic mean-variance allocation and terminal-wealth risk under regime-switching jump-diffusions

Systematic multi-asset allocation strategy using Hidden Markov Models to identify VIX volatility regimes and dynamically rotate between TLT, GLD, and SPY

A quantitative trading framework that leverages daily OHLCV stock data and a Hidden Markov Model (HMM) to dynamically identify market regimes and generate momentum-based trading signals.

This package implements hypothesis testing procedures that can be used to identify the number of regimes in a Markov-Switching model.

Implementations of various trading strategies

End-to-End Python implementation of Ang et al's (2026) Agentic 'Self-Driving Portfolio'. Implements: Black-Litterman equilibrium priors, Grinold-Kroner building blocks, Campbell-Shiller CAPE analysis, Ledoit-Wolf covariance shrinkage, Risk Parity, Hierarchical Risk Parity, and Robust Mean-Variance optimization across 18 asset classes.

Python framework for simulating and estimating Bayesian nonlinear pricing models with regime-switching and adaptive volatility filters.

Automated volatility arbitrage engine exploiting rough volatility mispricing in short-dated equity options. Combines Monte Carlo pricing with Gaussian HMM regime detection to trade only during calm markets. Connects to Interactive Brokers for live/paper trading with full validation suite.

Quantitative regime-switching trading framework using Hidden Markov Models (HMM) to adapt market exposure based on changing volatility and return environments.

Implementation of financial market regime identification models including traditional statistical approaches and deep learning methods (GRSTU), featuring a novel application of Temporal Fusion Transformers to regime classification.

[FUSION 2024] A Gaussian Process-based Streaming Algorithm for Prediction of Time Series With Regimes and Outliers

Regime-Invariant Specialist Pools (RISP): reward-independent specialist retention + episode-invariant decision-focused training for non-stationary markets. Theory, 11-arm experiments, three papers — honest nulls included. Companion to GAUSE.

Unsupervised latent regime discovery for crypto markets. HMM, VAE, and temporal contrastive models identify hidden market states from multi-exchange data. FastAPI + React dashboard. Docker Compose.

Building a balanced Vanguard ETF portfolio with data-driven optimization—exploring advanced methods, robust backtesting, and an interactive Dash app to pick your optimal mix.

This project reimagines the classical Merton portfolio optimization problem using Deep Reinforcement Learning (DRL). Instead of static, closed-form allocation rules, we design an intelligent agent that dynamically adjusts exposures to risky and risk-free assets under changing market regimes.