Home / Ideas / Cross-Sectional Short-Term Reversal (Enhanced, Sector-Neutral)
Cross-Sectional Short-Term Reversal (Enhanced, Sector-Neutral)
buy last week's losers, short last week's winners among large caps
The classic 1-week reversal has strong long-run evidence (Quantpedia reports Sharpe ~1.09, CAGR ~16% on top-100 large caps, 1990–2009) but has decayed to near-zero in raw form; enhanced variants that neutralize industry/factor returns show "more than double" the risk-adjusted performance of the classic version. Idiosyncratic reversal is exactly what the current record-low-correlation, high-dispersion regime rewards.
Why it's relevant now
1-month implied correlations near theoretical lows and sector dispersion at extremes (semis -9% vs banks +1.5%) means single-name overreactions are large and largely idiosyncratic — the cleanest setup for sector-neutral reversal.
Universe
~100 most liquid US large caps (or S&P 500 members) with sector labels via ETF proxies (XLK/XLF/XLE/... for neutralization). Free daily OHLCV.
How it works
Each week rank stocks by prior-week return after subtracting their sector ETF return; long the decile of biggest relative losers, short the decile of biggest relative winners; equal-weight, hold 1 week, weekly rebalance.
Expected performance
Research-derived Sharpe estimate: 0.5–1.0. classic 1.09 discounted ~25% for costs/decay; enhanced neutralization recovers the upper end).
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 Weekly cross-sectional short-term reversal on US large caps: long prior-week relative losers, short relative winners, sector-neutralized via sector ETFs, hold one week.
Get SignalChain — $49 →
Research & sources
- quantpedia.com — quantpedia.com
- newyorkfed.org — newyorkfed.org
- quantpedia.com — quantpedia.com
More mean-reversion and statistical arbitrage 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.