Expanded PII and payment-card risk language in Item 1A.
Research integrity before idea volume
How Research Ideas Are Grounded
The selected excerpt is display evidence. Novelty is tested against the full available point-in-time disclosure record.
Bounded validation
2 published ยท 6 safe failures
Zero actionable trade views. Formal blinded human ratings remain pending.
Experimental feature: These are AI-generated research hypotheses grounded in cited filing evidence. They are not facts, personalized investment advice, or validated trading signals. Analyst review is required.
Eight controls
From disclosure to analyst decision
Each stage can stop the packet. A plausible narrative is not enough.
- Full prior-filing searchSearch prior 10-K sections, full filings, earlier annual reports, 8-Ks, prior news, and current context.
- Atomic claim decompositionSeparate each factual proposition so unsupported content cannot hide inside a broader sentence.
- Financial-quantity normalizationAssign amounts an economic role, unit, period, conditionality, and paid/remaining basis.
- Novelty verificationFind the earliest eligible prior occurrence and state what changed and what did not.
- Counterevidence retrievalActively search for evidence that weakens novelty, mechanism, or instrument fit.
- Hypothesis generationRetain only evidence-linked, cautious, falsifiable research questions.
- Skeptical reviewTest the opposite interpretation, genericity, unsupported causality, and quantity-role conflicts.
- Analyst approvalKeep the final decision and notes local to the analyst workflow.
Trust checklist
What must be visible
Published packets expose evidence and limitations, not hidden reasoning traces.
- Complete prior filing searched
- Earlier 8-Ks checked
- Prior news checked
- Exact evidence shown
- Facts separated from inference
- Financial quantities assigned roles
- Counterevidence shown
- Falsification condition required
- Unsupported mechanisms rejected
- No position sizing generated
- Analyst review required
What this can do
Structure a research agenda
- Show verified evidence and novelty
- Propose a possible mechanism
- Define required data and falsification conditions
- Surface the strongest contrary interpretation
What this does not do
Recommend a trade
- No live generation
- No buy or sell instruction
- No position size, leverage, stop loss, or price target
- No claims of analyst approval or completed human validation
Public-safe guided example
One frozen RH hypothesis, five inspectable decisions
This compact view uses only the validated V2 public-safe contract. No V1 speculation or future outcome is loaded.
Complete prior disclosure context reviewed before the final novelty decision.
Investigate issuer-specific control or compliance changes; no incident is established.
Peer and control evidence should distinguish issuer specificity from boilerplate.
PASS_HYPOTHESIS_ONLY with no actionable trade view.
Current gate
The 60-case phase remains blocked.
Broader generation requires at least six of eight publishable packets, zero critical grounding defects in published packets, complete retrieval coverage, transparent safe failures, and a frozen methodology. The bounded V2 produced only two publishable hypotheses.
Review the bounded results