Outcome-independent evidence packets

Cases organized by what they teach—not what happened to the stock later.

Each case asks an analyst question, compares filing language, checks earlier disclosure, and states what the evidence cannot prove.

Curated teaching gallery

“Genuinely new” means new relative to the filings and headlines available in this study—not new to the world.

New risk disclosure

No matching earlier disclosure was found in the bounded study sources.

Equifax · EFX

Litigation or regulatory escalation

SectionItem 1ANoveltyGenuinely newConfidence0.941Why usefulShows the meaning—and limit—of a bounded novelty result.

Dollar Tree · DLTR

Impairment or write-down

SectionItem 1ANoveltyGenuinely newConfidence0.924Why usefulConnects changed risk language to a concrete accounting concern.

Previously anticipated change

The 10-K added or sharpened detail, but earlier study sources already raised the topic.

Kraft Heinz · KHC

Debt or refinancing pressure

SectionItem 1ANoveltyPreviously disclosedConfidence0.947Why usefulSeparates sharper covenant wording from first disclosure.

Freeport-McMoRan · FCX

Newly added risk factor

SectionItem 1ANoveltyPreviously disclosedConfidence0.990Why usefulShows how an intervening 8-K changes the novelty judgment.

Liquidity or credit change

Cases where financing capacity or resource allocation is the analyst’s review question.

Truist Financial · TFC

Liquidity deterioration

SectionItem 7NoveltyGenuinely newConfidence0.738Why usefulTests interpretation with lower confidence and direct liquidity language.

Devon Energy · DVN

Material capital-spending change

SectionItem 7NoveltyGenuinely newConfidence0.858Why usefulShows how strategy and capital allocation appear in changed language.

Operational or cybersecurity change

A change that tests whether the topic is new or a formalization of earlier disclosure.

RH · RH

Newly added risk factor

SectionItem 1ANoveltyPreviously disclosedConfidence0.990Why usefulConnects cyber/privacy language to earlier evidence.

Low-materiality or null example

Not every detected change deserves the same review priority.

Chemed · CHE

Removed or softened language

SectionItem 7NoveltyPreviously disclosedConfidence0.810Why usefulDemonstrates a lower-attention comparison without outcome storytelling.

Experimental extension

Three Outcomes the Engine Can Produce

These frozen cases were selected without later performance and generated without future outcomes. Formal analyst validation remains pending.

Publishable hypothesis

RH

A grounded, falsifiable hypothesis with 1.00 factual evidence coverage and no actionable trade view.

Explore RH hypothesis
Hypothesis with uncertainty

DVN

A publishable research agenda that keeps unresolved uncertainty visible and remains hypothesis-only.

Explore DVN hypothesis
Rejected after grounding

EFX

Full-prior evidence corrected novelty and numeric roles, removing an unsupported narrative.

View EFX rejection

Experimental extension

Three outcomes the Research Idea Engine can produce

These cases were not selected using later performance. Formal analyst validation remains pending.

Publishable hypothesis

RH

A grounded disclosure change becomes a falsifiable research question, with no actionable trade view.

Review RH
Hypothesis with uncertainty

DVN

A second grounded hypothesis keeps uncertainty, required data, and falsification conditions visible.

Review DVN
Rejected after review

EFX

Full-prior evidence corrected novelty and numeric roles, removing unsupported mechanisms.

Review EFX