Rejected thesis
Failed-breakdown / bear-trap
Thesis (modestly working only with drawdown): A close below the trailing 20-day low on volume, followed by a reclaim above that level within 1-3 days, marks a fake-out — shorts get squeezed and the price recovers.
Code: research/failed_breakdown.py
Report: research/failed_breakdown_report.json (n=24,204)
Where it lands: same family as the broader contrarian set — unconditional edge is essentially 0 ; deep-drawdown sub-cohort produces +3.4 % at 40 d OOS but the magnitude is dominated by other triggers (RSI div, consensus, gap-cap).
| Bucket | n | 20d edge | 40d edge |
|---|---|---|---|
| all | 24,204 | +0.22 % | -0.07 % |
| dd ≤ -30% | 4,608 | +1.74 % | +2.83 % |
| OOS 2020+ dd ≤ -30% | 2,981 | +1.99 % | +3.42 % |
Verdict: Modest. Not productized as a standalone cohort but is one
of the triggers feeding consensus_capitulation.
Working signals — discoveries from session 2
These were the most impactful findings of the second pass.
| Sample | 24,204 |
|---|---|
| Validator | research/failed_breakdown.py |
| Verdict | Rejected |
Every result here is reproducible from the script named above. Reports are in the repository.
- Volatility squeeze
- Pre-FOMC drift
- Uptrend pullback
- Momentum continuation
- Industry-relative momentum (at 4-week lookback)
- Earnings-reaction reversal (anti-PEAD)
- Beat-persistence pre-earnings drift
- Sector ETF mean-reversion
- 52-week-high breakout
- Volume accumulation (above-200DMA +...
- Cross-sector momentum
- Analyst-upgrade cluster