market-regimes

An advanced ML trading dashboard for Nifty 50. JUDAH uses an automated XGBoost pipeline that grid-searches 360 combinations daily across 6 horizons. The live Streamlit engine fuses a 4-pillar probability ensemble to generate mathematically precise, high-conviction options strategies (Spreads/Strangles) and Strike recommendations.

Multi-asset market regime detection and trading strategy using KMeans, HMM, walk-forward retraining, and macro signals

Personal research into why markets move — regimes, flows, and a permanent daily record. Not financial advice.

A regime-aware reinforcement learning workbench for synthetic trading research. It combines a hidden-regime market simulator, multiple agent baselines, a live terminal dashboard, and experiment tooling for ablations, OOD sweeps, and artifact-driven analysis.

Randomized, regime-aware benchmark for evaluating trading strategy outputs across changing market conditions.

Research framework testing whether market regimes make systematic strategies more robust out of sample: causal features, purged walk-forward validation, cost-aware backtesting and significance testing.

Educational investment research lab with market regime dashboards, portfolio diagnostics and quantitative research modules.

Notebook-based regime modelling pipeline for Indian markets using multi-asset historical analogue analysis.

Clean-room quantitative research: volatility & dispersion regime detection tested with HAC inference, permutation testing, and out-of-sample holdouts.

Structural analysis of the “10 best days” market timing claim using SPX regime clustering and volatility geometry.

Institutional-grade market regime detection, systemic risk telemetry, and walk-forward backtesting engine in Python.

Quantitative test of the "bonds always protect" narrative across inflation regimes. UC2 — Usual Claims Series

Public evidence portfolio for macro-financial research engineering: Python pipelines, real-data ingestion, volatility-regime analysis, ML validation, backtesting logic, and research reporting.

Quantitative test of the "VIX = fear" narrative. UC3 — Usual Claims Series

The Auditing LLM Trading Dataset repository provides an end-to-end benchmarking framework designed to evaluate, audit, and compare Large Language Models (LLMs) against traditional machine learning and deep learning models in financial trading environments.

This project simulates wealth accumulation during the pre‑retirement phase using multiple financial return models. The goal is to compare deterministic and stochastic approaches to long‑term portfolio growth.

Python dashboard analysing how BTC, ETH and SOL behave across macro and cross-asset market regimes.