Home / Ideas / Crash-protected cross-sectional momentum
Crash-protected cross-sectional momentum
vol-scaled / dynamic momentum to survive the reversal spikes
Daniel-Moskowitz momentum crashes and Barroso-Santa-Clara volatility-managed momentum show that scaling exposure inversely to recent momentum-portfolio volatility roughly doubles the strategy's Sharpe and cuts the fat left tail — the single most impactful known improvement to cross-sectional momentum.
Why it's relevant now
2025 saw "violent reversals" that gutted trend books; with elevated 2026 volatility and dispersion, the reversal/crash risk in momentum is live, making dynamic vol-scaling especially valuable rather than a nice-to-have.
Universe
Same cross-sectional momentum book as idea 1 or 3 (sector ETFs or S&P 500 names), with a portfolio-level vol overlay.
How it works
Build a standard 12-1 cross-sectional momentum long/short (or long-top) portfolio, then scale gross exposure by target_vol / trailing-realized-vol of the momentum portfolio, rebalanced monthly.
Expected performance
Research-derived Sharpe estimate: 0.6–1.1. Barroso-Santa-Clara report roughly doubled Sharpe vs unscaled).
Backtest this idea with SignalChain
This is a research lead — not a finished backtest. SignalChain takes an idea like this and runs the whole pipeline inside Claude Code: it researches the concept against academic and practitioner sources, sets benchmarks, writes and lints a VectorBT backtest, runs it, and grades the result PASS/FAIL. One command:
/signalchain Volatility-managed momentum: 12-1 cross-sectional momentum portfolio scaled inversely to its own trailing realized volatility, monthly rebalance
Get SignalChain — $49 →
Research & sources
- arXiv — Is Factor Momentum More than Stock Momentum? — arxiv.org
- Man Group — Factors — man.com
- VT Markets — A Trader's Guide to the Great Rotation in 2026 — vtmarkets.com
More cross-sectional momentum and trend-following ideas
Not financial advice. This page describes a research idea, not a recommendation. Any performance figures are hypothetical, research-derived estimates and are not indicative of future results. SignalChain is a research and educational tool; you are solely responsible for any decisions you make.