Ask a market question.
Explore the history.

One memory, two ways in.

You can look up a stock in the browser. An agent can ask for the same completed-session history through MCP or REST. Both return dates, how many similar days we found, the range of what followed, and the limits. The steps below are the browser view; the agent guide shows the calls.

1. Choose a stock and a trading day.

Enter a stock symbol, such as AAPL. Leave the date blank to use its latest available completed trading day, or choose a past date.

Check the date on the result. It describes that completed day, rather than a real-time price. A dated link opens a form; it does not run a new request until you press Find historical examples.

2. Look at the historical examples.

The tool first looks for days that were in a similar situation — trend, position versus the range, volume, what peers were doing — then ranks those by the shape of the chart. It shows how many examples it found.

A similar-looking chart can still have different news, earnings, or market conditions. More matches are not automatically better evidence, especially when the same stock or nearby dates repeat.

3. See what followed.

Choose 1, 5, or 10 trading days to see the historical outcomes at that point. Read the lower end, the middle result, and the upper end together. Switching between these periods uses the same returned comparison.

The simple tool reports returns relative to a date-matched market baseline. A result of +2 percentage points means the stock outperformed that baseline by 2 points; it does not necessarily mean its price rose 2%.

The range describes the middle 80% of the historical observations. Some outcomes fell outside it. It is not an 80% probability for the next move in your chosen stock.

Check the limits before drawing a conclusion.

Every time period has its own observed count. Missing outcomes stay missing. If a comparison check fails, read the reason; a failed or unavailable result can support no conclusion.

The separate coverage record evaluates a specific cohort-band method. Its percentage does not automatically apply to the raw historical ranges in the simple comparison.

Keep the evidence you used.

Download the original data to preserve the exact response, including warnings and missing fields. A dated link lets you run the stock and date again, but underlying data or methods can change.