Track record

Every signal.
Publicly verified.

We timestamp every signal we issue and track every outcome. No cherry-picking, no survivorship bias, no look-back adjustments. This is our live rolling accuracy across 14,200+ signals in the trailing 12 months.

87%
Overall directional accuracy
8.4d
Avg signal lead time
14K+
Signals tracked (12M)
92%
Accuracy above 85% confidence
By signal layer

Where each layer
performs best.

Signal layer Signals Accuracy Lead time
Insider filing patterns
SEC Form 4 cluster analysis
1,840
91%
11.2d
Institutional flow
13F filing delta tracking
2,104
89%
9.8d
Organizational behavior
Executive movement & hiring
1,620
84%
7.4d
Earnings linguistics
Transcript linguistic scoring
2,380
88%
6.1d
News & sentiment
NLP sentiment trajectory
3,210
83%
4.3d
Alternative data
Job, app, web & logistics data
1,950
86%
8.9d
Macro correlation
Sector & macro regime scoring
1,096
81%
5.7d
Methodology

How we
measure accuracy.

When a signal is issued, it is timestamped, archived, and assigned a resolution window — the period over which the predicted directional move is expected to occur. At expiry, we compare the predicted direction (up/down) against actual price movement.

We do not measure magnitude — only direction. A signal predicting "accumulate" is correct if the stock is higher at resolution, regardless of how much higher.

All signals issued — including ones with low confidence that were never surfaced to users — are included in the base denominator. We don't cherry-pick only high-confidence signals for the published accuracy figure.

📅
Resolution windows
Short-term signals resolve over 5–10 trading days. Medium-term over 20–40 days. Resolution window is shown on every signal card.
📊
Rolling 12-month window
The published accuracy figure is a rolling 12-month average, updated daily. Older data gradually rolls off as new signals resolve.
🔍
Full audit trail
Every resolved signal is available in the audit log. You can filter by layer, date, ticker, confidence band, or outcome.
⚠️
What accuracy doesn't mean
Past accuracy does not guarantee future performance. Markets change regimes. Always use signals as one input among many.

Accuracy FAQ.

Because we include all signals in the denominator, not just high-confidence ones. If we only measured signals above 85% confidence, accuracy would be 92%. We report the honest all-in number because it gives you a more realistic picture of what the system does.
They are logged and visible in the full audit trail. Wrong signals are an important part of calibration — we analyze them to understand whether they represent model gaps, data quality issues, or genuinely unpredictable events, and adjust weighting accordingly.
Yes, meaningfully. Large-cap technology companies — where data availability is highest — show the strongest accuracy. Small-cap and micro-cap stocks have lower data coverage and correspondingly lower signal confidence. We flag data coverage quality on every signal card.

Verified intelligence,
your portfolio.

Start a free 14-day trial and see live signals with full reasoning and source attribution.

Start free trial → See pricing