Long-term signal engine
Daily Dashboard
Today's Suggested Stance
Strict buying, stricter selling.
Conviction Hold
Buy only high-ranked names. Once owned, keep holding unless rank and trend damage become severe.
Entry rank 1-20
Severe review >150
Exit candidate >260
Recommendation Ladder
Current Model Picks
Data Pipeline
Offline
Connecting
Waiting for backend.
Waiting for backend.
Research gate
Waiting for backend.
Waiting for backend.
Operations gate
Waiting for backend.
Waiting for backend.
Provider gate
Waiting for backend.
Waiting for backend.
Automation
No schedule loaded.
No schedule loaded.
Portfolio Health
Manual holdings, tax mode, concentration and signal overlap.
| Ticker | Recommendation | Rank | Trend | Weight |
|---|
Manual Holdings
Enter holdings to estimate concentration, tax risk, and strategy overlap.
Portfolio Diagnostics
Backtest Evidence
2010-2026What Survived
- Conviction hold beat QQQ pre-tax with lower turnover.
- After-tax return nearly matched QQQ while reducing drawdown.
- MACD entry filters did not improve the price-only model.
- Quarterly throttling reduced responsiveness too much.
Known Risks
- Wikipedia universe bootstrap is not a licensed point-in-time dataset.
- Momentum can crash after panic-state rebounds.
- Tax model is approximate and not personal tax advice.
- Fundamental and valuation gates need production data coverage.
Model Rules
Roadmap Gates
Experiment 4
Point-in-time fundamentals, FCF, ROIC, gross margin stability.
Experiment 5
Valuation and earnings-revision overlay to avoid expensive momentum traps.
Product gate
Account-aware recommendations: taxable, IRA, research-only.