Problem 01
Certainty arrives too fast
Most market content wraps unresolved signals in confident calls and target numbers — without saying what the evidence is, or what would prove the call wrong.
Scenario Tree
Branches before probabilities
正在形成的訊號 — 尚無定論
Scenario Tree separates possible outcomes, confirmation triggers, refutation paths, and next checks from source-backed Guidances articles.
Why this exists
News reports outcomes. Judgment happens before the outcome.
Most consequential decisions are made while the signal is still unresolved. Scenario Tree is built for that interval — the stretch between the headline and the verdict.
Problem 02
Nothing gets tracked
Headlines scroll past, and yesterday’s call is never checked against today. No record remains that lets you revisit whether a judgment actually held.
Problem 03
Fact and guess blur together
When confirmed facts and forming signals share one sentence, the reader is left to draw the boundary alone. That burden belongs to the product, not the reader.
Operating principles
Four rules the Tree keeps
- 01
Branches, not verdicts
Instead of selling one forecast, we split the range into branches. Every branch states up front what would confirm it and what would refute it.
- 02
Probabilities are estimates, not instructions
A branch’s likelihood is an editorial estimate. It updates when new signals arrive, and the update history stays visible. Context for judgment — never a buy or sell instruction.
- 03
Wrong stays on the record
A refuted branch does not quietly disappear. It closes as ✕ debunked and remains visible. Showing what we got wrong is the price of being trusted.
- 04
No branch without evidence
A branch exists only when it ties to source-backed articles, updates, or verification checkpoints. No evidence, no branch.
DEMO · SAMPLE
Visualization preview
Example tree — how one trigger splits into branches
Illustrative example — not live market data.
情景樹
▸Confirmation path — trigger materializes[52%]
▸Context-only path — partial signal[28%]
▸Refutation path — thesis weakens[14%]
▸Trigger fails — branch debunked[6%]
Worked example: an unresolved policy trigger
How to read it
What the branch glyphs mean
Each branch carries a glyph for its current state. Probabilities are editorial estimates that update as new signals arrive — they are context, not buy/sell instructions.
- StrengtheningNew signal supports this branch
- ConfirmedOutcome path locked by sources
- FormingSignal building, still partial
- WeakeningEvidence draining from the branch
- DebunkedRefuted and closed
- ContextReference context, not proof
01
Paid branch rows, mock probabilities, and placeholder scenarios are not serialized on public SSR.
02
A branch appears only when it can be tied to source-backed articles, updates, or verification checkpoints.
03
This public screen explains the access boundary until the production-safe tree loader is available.