Start with the date.
Then read the range.

A real example: AAPL, September 4, 2026

The response captured on September 8 contained 300 historical examples from 208 stocks and 196 trading days. Of those examples, 299 had a five-day outcome.

The middle result was close to the market baseline, but the historical outcomes spread across both sides. The middle result alone would miss that variation.

What does “relative to the market” mean?

The simple comparison subtracts a date-matched liquid-stock baseline from each stock's return. The result is an excess return, measured in percentage points.

For an illustrative calculation, if a stock rose 3% while its baseline rose 1%, the excess return would be +2 percentage points. If a stock fell 1% while the baseline fell 3%, it would also be +2 points. A positive excess return does not necessarily mean the stock rose.

The advanced workspace may show other return definitions. Use the unit attached to each output; do not mix a raw stock return with an excess return.

What are the lower, middle, and upper results?

The lower end is the 10th percentile, the middle result is the median, and the upper end is the 90th percentile. Together, the lower and upper ends describe the middle 80% of the observed historical outcomes.

Some observations fall outside this range. It is not the worst possible loss, an entire price-path envelope, or an 80% probability for a new stock's next move. The simple tool's historical percentiles are not automatically calibrated forecasts.

Why do the counts change?

A historical example can have a one-day outcome but no usable ten-day outcome. Each time-period control therefore carries its own observed count. An unavailable value is left missing rather than filled with zero.

The closest named examples are a small view into the selected set. The aggregate range uses the full returned set of usable observations for that time period.

When is the comparison too weak?

Read warnings about missing observations, a thin sample, or a failed comparison check before relying on a range. A larger count can still contain repeated stocks or nearby dates affected by the same event.

Similar charts can have different economic causes. The stock's news or current conditions may differ from those behind the historical examples. A comparison may help frame a question without supporting a directional conclusion.

A service error means the tool could not complete the request. An empty result means it did not supply a usable observation. Neither describes an expected return.

How is this different from the coverage record?

The coverage record compares specific published cohort bands with outcomes that later occurred. Its dates, population, method, and warnings matter. It does not certify every historical percentile range returned by another tool.

Daily examples have their own settled-outcome tally. A count of ranges containing later outcomes is not a record of trading profits.