MARKET MEMORY / RESEARCH DESK

What followed
a moment like this?

Turn a market question into historical evidence. See the events, compare them with ordinary windows, and inspect what actually happened next.

BTC · ETH · SOL

Events. Sequences. Similar candles.
The same evidence for you and your AI.

Understand the method
01 / ASK THE HISTORY

Start with a question.

Open research. No account needed.

YOUR QUESTION

When BTC fell 5%, what did another market do?

Find the fall on BTC, then observe ETH or SOL from the same hour. The watched market supplies the outcome and charts.

Hourly candles · stored, closed history · a pinned snapshot · no forecast

Choose a question and run research

YOUR QUESTION IS READY

Let the history answer it.

Run the selected study to see event counts, ordinary windows, uncertainty and the paths behind the result.

02 / READ THE EVIDENCE

A large percentage
is only half the story.

Price crosses nearby lines surprisingly often. A fall can recover. Different markets can react to the same event. A useful study makes those details visible.

  1. 01

    Define the event.

    A move between closes and a fall inside the day answer different questions. The condition is shown before you run it.

  2. 02

    Keep the background in view.

    See what followed ordinary windows under the same context. Compare the difference, not just the event’s percentage.

  3. 03

    Open the evidence.

    Inspect real episodes, uncertainty, missing data and the pinned snapshot. Share original charts without an agent redrawing them.

HOW THE EVIDENCE IS BUILT
01

Say what actually happened.

A fall inside the day is different from a lower daily close. Define the event first, including recoveries and when the observation starts.

The diagrams illustrate the method; dots are not sample counts.
Is this a prediction or a trading signal?

No. These are counts and distributions from stored history. Similarity, a frequent outcome or a positive difference does not establish a tradable strategy. Results remain NO_TRADE.

Why are there two event counts and two background counts?

A matched event can lack a complete outcome because of gaps or the end of history. Resolved events enter the outcome statistics. Ordinary windows can overlap: the pooled count describes the background, while uncertainty uses the independent-window count.

Why does the chart show fewer hours than the statistics?

To keep paths readable, charts display at most 24 forward candles. The selected 12, 24 or 48 hour horizon still determines the statistics. A median chart combines each bar across episodes; it is not one real historical path.

What does “no clear difference” mean?

The returned 95% difference interval includes zero. It does not prove the two groups are equivalent. Trying many questions can also produce accidental findings; confirm a prespecified question on an untouched period.

Can I share exactly the evidence I’m looking at?

Yes. Create an evidence link from a completed result. The report must match the same snapshot, includes original charts, and offers checksum-verifiable text. Cached links last up to 24 hours and may become unavailable earlier; the JSON download remains yours.

ALREADY HAVE YOUR OWN PRICES?

Open the analogue-level lab.

Inspect the reach of a level, target and invalidation you choose.

THE SAME ENGINE, IN YOUR ASSISTANT

Keep asking in your own words.

Use the website’s questions here, or connect an MCP assistant for your own event sequences, contexts and cross-market studies. The engine returns the numbers and original evidence.