Regime-Aware Sector Rotation with Momentum and Canary Assets

Research summary of a Python implementation and extension to U.S. sector ETFs.

Overview

This project reproduces and extends a regime-aware Hybrid Asset Allocation framework from market-cap ETFs to U.S. sector ETFs. It uses a canary asset momentum signal to define risk-on and risk-off regimes, implemented with month-end signals and a flat 5 bp turnover cost.

Main findings

The strongest full-sample strategy is Sector Top-2 SHY Net, which holds the two top offensive sectors in risk-on regimes and rotates fully to SHY in risk-off regimes. It produces the highest CAGR among tested rules while materially lowering drawdown versus broad equity benchmarks.

Robustness

Walk-forward model selection and paired block bootstrap inference show that the full-sample winner is not always selected in real time, and raw outperformance over SPY and VTI is not statistically decisive. The clearest empirical contribution is the improved risk profile of sector momentum strategies.

Strategy design

The model computes a 13612 momentum score for a canary asset and uses it to toggle between offensive and defensive allocations. Offensive sectors include XLK, XLY, XLI, XLF, XLE; defensive sectors include XLP, XLU, XLV; SHY is used as a clean risk-off allocation.

Limitations & next steps

The summary notes limitations from yfinance data, a U.S. ETF-only universe, simplified transaction costs, proxy factor regressions, and no multiple-testing adjustment. Recommended next steps include formal Fama-French/Carhart regressions, stress-period analysis, and additional robustness metrics.