How We Analyze

Every analysis on this site goes through a structured multi-model pipeline. Multiple AI models examine the same evidence from different angles, debate each other's findings, and converge on evidence-graded assessments. The methodology is the product.

Multi-Lens Analysis

Each company is examined through 13 specialized lenses — from accounting integrity to tail risk. Each lens asks a different question and produces standardized, comparable signals.

Structured Discourse

For every lens, multiple models independently analyze the evidence, then critique each other's work. A synthesizer reconciles disagreements. Positions stabilize through debate, not averaging.

Evidence Grading

Every finding carries an evidence grade (E0-E3). Unverified claims are labeled, not hidden. You see what was checked, what wasn't, and where models disagree — so you know where to focus your own diligence.

The Pipeline

1

Source Collection

SEC filings, earnings transcripts, FRED data, industry reports, and alternative data sources are gathered and validated.

2

Structured Extraction

Raw documents are distilled into dense, fact-only dossiers. No interpretation yet — just the data, organized for analysis.

3

Independent Lens Analysis

Each lens runs independently. Multiple models analyze the same dossier through the lens framework, producing signal assessments with evidence grades.

4

Adversarial Discourse

A dedicated critic finds flaws, gaps, and unsupported claims. The original analyst defends or revises. A synthesizer reconciles. Positions are tested, not rubber-stamped.

5

Convergence

A moderator checks whether positions are stabilizing. If not, discourse continues. We seek convergence through evidence, not forced consensus.

6

Cross-Lens Synthesis

Findings across all lenses are aggregated. Corroborating signals strengthen confidence. Contradictions are preserved and surfaced.

What We Cover

The same discourse pipeline powers three analytical verticals.

30
Analysis Lenses
17
Tracked Signals
213
Companies Analyzed
1774
Forecast Markets

Go Deeper

What This Produces

  • Evidence-graded signal assessments per lens
  • Model agreement/disagreement maps
  • Cross-lens synthesis with corroboration scoring
  • Probability estimates on forward-looking questions
  • Transparent discourse transcripts

What it does not produce:

  • Buy/sell recommendations
  • Price targets or fair value estimates
  • Personalized financial advice