Conviction Grades
A single undifferentiated signal feed treats a marginal setup and a high-confidence setup identically. Grade-tiered systems let subscribers allocate more capital to the highest-conviction setups and scale back on weaker ones — but only if the grades are calibrated against live results and explained before you subscribe.
Grade tiers must be defined with specific, quantitative criteria; calibrated against live signal results (not backtests); published before subscription; and consistent across the signal history. Subjective “high / medium / low confidence” labels without calibration data do not satisfy this criterion.
Why Grade Systems Matter to Buyers
- They let you focus capital on the highest-confidence setups if your account size limits simultaneous positions
- They provide a framework for backtesting the grade-level performance of a service before you follow a grade tier
- They reveal whether a high aggregate win rate is driven by a small number of high-grade signals or is consistent across grades
- They expose whether the service’s stated win rate holds at the grade level you intend to follow
What to Look for in a Grade System
| Factor | Strong | Weak |
|---|---|---|
| Grade definition | Quantitative threshold (e.g. minimum avg return per trade) | Subjective language only (“high confidence”) |
| Calibration basis | Live signal results per grade | Backtest-only calibration |
| Publication timing | Grade definitions accessible before subscription | Revealed only post-subscribe |
| Consistency | Grade criteria documented and applied consistently across history | Retroactively applied; or definitions changed without announcement |
| Per-model variation | Grade bars calibrated separately per model (different horizons warrant different thresholds) | Single grade scale applied across all models regardless of holding horizon |
Red Flags
- No grade tiers at all — all signals presented as equivalent conviction
- Grade labels applied to signals retrospectively after outcome is known
- Grade-A signals that never close at a loss in published history (suggests selective reporting)
- Grade calibration that relies solely on backtested data
- Grade definitions changed without disclosure or a changelog
How Vector Ridge’s A–D Grade System Satisfies This Criterion
Vector Ridge operates an A through D conviction tier system applied across all four signal models. Each grade level is defined by quantitative performance criteria calibrated against the live 2026 signal record, not a backtest. Grade A represents the highest conviction level; Grade D is the lowest. Each model has its own Grade-A threshold, reflecting the different performance characteristics of different holding horizons.
These thresholds are publicly accessible. The Day Trade Grade-A bar is 0.70% average per trade. Multi Hour Grade A requires 4.50%. Swing Trade Grade A requires 6.00% average per trade — the highest bar, appropriate to the flagship model’s 7–28 day holding period and 74.4% live win rate.
The 2026 YTD aggregate performance — 690 signals across all models, 70% combined win rate, +1,227% combined P&L — is drawn from the full grade distribution, not a filtered Grade-A-only subset. Subscribers can choose to follow all grades or filter to Grade A and above depending on their own risk tolerance and position-sizing approach.
Every graded signal is cryptographically timestamped the moment it goes out. Because the grade is folded into that sealed hash from the start, no one can quietly bump it up once the result is in without the tampering showing.