§ Documentation
Reference
How Sentinellis works
We're transparent about our pipeline, data sources, and the math behind every score. No black boxes — read what powers every report.
How It Works
End-to-end pipeline: parallel data fetch, news relevance scoring, financial enrichment, AI synthesis with inline citations.
- How a report is built
- News relevance pipeline
- Financials enrichment
- AI synthesis & inline citations
- Cache, retries, fail-soft
Data Sources
Every provider we query, what they return, freshness windows, and what to expect when coverage is thin.
- Yahoo Finance (financials, exec)
- NewsData.io (primary news)
- Google News RSS (fallback)
- Anthropic Claude (synthesis)
- Coverage gaps & fallbacks
Methodology
The math behind every score: Sentinellis Health, Piotroski F-Score, Altman Z-Score, ROIC vs WACC, confidence.
- Sentinellis Health Score (0–10)
- Piotroski F-Score (9 tests)
- Altman Z-Score (bankruptcy risk)
- ROIC, WACC, NOPLAT
- Confidence score breakdown
Limitations
Honest about what we can't do: private companies, non-US coverage, news lag, AI variance, no investment advice.
- Private company coverage
- Non-US / emerging markets
- News recency window
- AI variance across runs
- Not financial advice
FAQ
Most common questions answered: report time, accuracy, refunds, account, security.
- How long does a report take?
- Is this financial advice?
- What if a report is incomplete?
- Can I cancel anytime?
- How is my data handled?
Glossary
100+ financial metrics defined in plain English with formulas and interpretation guidance.
- Margins (gross, op, net)
- Returns (ROIC, ROE, ROA)
- Leverage (D/E, coverage)
- Valuation (P/E, FCF yield)
- Risk scores (Z, F, Health)
§ Feedback
Something missing?
We're actively expanding these docs. If a topic isn't covered or an answer feels incomplete, tell us and we'll add it.
Suggest a topic