Cambial Stock Intelligence

Data Coverage & Sentiment

Loading analysis data...
Positive -- Awaiting data
Neutral -- Awaiting data
Negative -- Awaiting data
Mentions -- Awaiting data

Sentiment Mix

Current filtered conversation share

Net --

Momentum

Sentiment and volume over time

Topic Drivers

Conversation themes ranked by volume

Source Split

Where the analysis is coming from

Insurance Insight Catalogue

Primer-derived concepts, guardrails, and claim templates for filing analysis

Primer catalogue loads from the local server.

Concepts

Guardrails

Claim Templates

Data Inventory

Current sourced tables and priority-universe coverage

Company Assessment

Ranked Cambial companies from the latest sizing and signal file

Ticker Company Sector Score Complete Tier Key Need

Build Assessment

Integration readout for the constituent simulation engine package

Awaiting assessment --

Schema --
Pass B --
Lift --

Layer Status

Integration Map

Risks

    Next Build Steps

      Research Validation

      Key literature-backed learnings from the local reading pack

      Literature Exhibits

      Presentation-ready tenets distilled from the literature pack

      Implementation Calibrations

      Concrete method rules and priors extracted from the implementation notes

      HTML Visuals

      Responsive visual summaries of the engine thesis, merge layer, rigor gates, and twin architecture

      Pipeline

      Evidence To Queryable Twins

      Sources
      Spine
      Labels
      Merge
      Simulation
      Twin UI

      Build order: resolve identities, define labels, extract responses, learn residual behavior, then expose grounded twins.

      Merge

      The Residual Is The Signal

      Observed response
      =
      Market
      +
      Proxy advisor
      +
      Bucket herding
      +
      Residual
      Stored twin signal

      `alpha_shrunk` plus residual behavior becomes the reusable constituent profile.

      Rigor

      Decision-Grade Gates

      1Literature
      2Synthetic
      3Real data
      4Backtest
      5Decision-grade

      Every claim needs evidence type, provenance, uncertainty, and leakage-clean validation.

      Twin

      Grounded Agent Architecture

      Observed actions
      Pass A/B/C/D
      Behavioral profile
      LLM-facing twin

      LLM personas sit on top of revealed-action behavioral profiles; they do not replace the merge layer.

      Filing Concept Explorer

      Search filings, reveal associations, and link mentions to metric cells and periods

      Example result loaded from the latest local search.

      Associations

      Applicability Guardrails

      Relationship Map

      Top Filing Mentions

      Metric Links & Periods

      Notable Signals

      Highest-impact findings from the current analysis