Hierarchical Risk Parity (HRP) Portfolio Engine

Event: Arbitrage Arena 2026 | Problem Statement: Cross-Asset Survivability
Team: TEAM Z

Project Overview

This project implements a Hierarchical Risk Parity (HRP) model to optimize a cross-asset portfolio (Crypto, Equities, Commodities). Unlike traditional Mean-Variance optimization, HRP uses Machine Learning Clustering to allocate capital based on risk hierarchy rather than unstable return predictions.

Key Achievement: The model successfully navigated the 2022 Inflation Crisis, reducing Max Drawdown by 10% compared to the NASDAQ Benchmark while maintaining a higher Sharpe Ratio (0.77 vs 0.67).

Key Features

• No Look-Ahead Bias: Implemented strict intersection logic to align 24/7 Crypto markets with Mon-Fri Equities data.
• Regime Robustness: Uses scipy.cluster to identify "Safe Haven" clusters (Gold/Silver) without human intervention.
• Stress Testing: Includes specific modules to backtest performance during the COVID-19 Crash and 2022 Rate Hike Cycle.

Tech Stack

• Python 3.10+
• Pandas & NumPy: Vectorized backtesting engine.
• SciPy: Dendrogram clustering and distance matrix calculations.
• Matplotlib/Seaborn: Financial visualization.

Data & Usage

Note: The raw .csv datasets used for this competition are proprietary to Arbitrage Arena 2026 and are not included in this repository.

To run this notebook:

1. Place standard OHLCV .csv files (e.g., BTC_USD.csv, AAPL.csv) in the root directory.
2. Ensure filenames match the file_map dictionary in Cell 2.
3. Run the notebook cells sequentially.

License

This project is for educational and portfolio demonstration purposes.