Principles
- Evidence before precision.
- Explain exclusions as carefully as inclusions.
- Abstention is a valid result.
- Separate data quality, model fit, peer fit, and forecast uncertainty.
- Use deterministic calculations for valuation and attribution.
Research pipeline
1. Identify the security
Resolve ticker, issuer, exchange, currency, share class, and security type before calculating.
2. Validate and normalize inputs
Check identity, units, freshness, period continuity, currency, shares, and material reconciliations. Missing data is not silently treated as zero.
3. Route appropriate models
Use only methods suited to the company and available evidence. Specialized company types require validated specialized methods.
4. Build scenarios and ranges
Bear, Base, and Bull use displayed assumptions. They are model scenarios, not predicted price boundaries.
What scenarios mean
Bear uses a more conservative assumption set. Base uses central model assumptions. Bull uses a more optimistic set. Actual outcomes may fall outside the range.
Confidence is multidimensional
The interface separately presents data quality, model applicability, forecast uncertainty, peer confidence, and sensitivity. One unexplained score cannot replace these dimensions.
Limitations
Valuation can be materially affected by missing or inaccurate data, reporting differences, assumptions, unsuitable peers, corporate actions, leverage, cyclicality, and model defects. A narrow range can still be wrong.