Three answers, every time

A price feed hands you homework. This hands you an answer.

What happened, who moved it, and how far the price should have gone. Three fields carry that answer: driver, attribution and reaction. Everything else in the event is context for them.

What lands in your system

Three answers, and your agent can size a position.

These are the questions a system has to settle before it risks money. Every published verdict answers all three.

What happened

information · liquidity · unknown

Tell a fact from money moving.

A large holder rebalancing moves the price the same way breaking news does. Your agent stops paying to find out which one it was.

Attribution weighs whether the broader order book moved at the same time. When it did, the driver is liquidity, even if a convincing headline landed in the same minute. That distinction is the difference between reading news and reading flow.

Who moved it

named source, relevance and novelty 0 to 100

Get the source, and what it is worth.

A candidate conceding and a commentator speculating move a price by the same amount for very different reasons.

You get the source, its class, how strongly it referred to the contract, whether it spoke first, and the lag in seconds between cause and move. Lag is computed from source clocks only, so a replay reproduces the same attribution. A source we did not observe is never named.

What it is worth

under · proportionate · over · unknown

See whether the move went far enough.

This is the field that becomes a trade. Everything before it is context for this one.

Every wick is scored against the price 5, 15 and 60 minutes later, and the reaction field is judged against those outcomes rather than a later narrative. The comparison sharpens with every scored wick, and the record page publishes how often it was wrong.

You always know where you stand

Unknown is a published value, not a silence.

A move we cannot explain publishes as a move with driver unknown. That is exactly why you can size on the ones that do carry a cause.

You never trade an invented reason

Below a judge confidence of 0.55 no cause is named. Below an attribution of 0.30 the driver is unknown. Below the publish floor the verdict is withheld and you get the move alone.

Silence still carries information

An unexplained move tells you the price moved and no credible source said why. Your agent can widen, wait, or stand aside on that alone.

Abstention cannot hide a weak model

The count of moves judged in private is published beside the miss rate, so you can see how often a verdict was withheld rather than published.

Every named source was observed

A cause is named only when that source was captured and scored. A source that spoke after the move is never presented as the reason for it.

Why there is no sentiment score

Tone is not direction.

A statement can sound confident and still push the contract toward no. A tone score sold as direction is a wrong answer delivered confidently.

A sentiment score says

This sounded optimistic. Cheap to compute, and silent on whether the contract you hold pays out.

A stance value says

This pushes the outcome toward yes, judged against the wording this contract settles on. The judge is given that wording when stance is computed.

Stance is pro yes, pro no, or unclear. No sentiment field exists in the envelope at any version, and a test refuses one at the publish boundary, because its absence is a guarantee rather than an oversight.

Your agent sees the price move. Send it the reason.

Subscribe with an npayload key. The free tier is realtime, capped at five events an hour, and eight lines of code print your first event.

Subscribe on npayload

$ npx npayload subscribe wickfeed.lit

No card. No login. No dashboard.