Every number in the library is one measurement, counted the same way across every qualifying session. No modelling, no smoothing. Here is exactly how a hit rate is built, when we trust it, and what the markers mean.
A hit rate is the share of qualifying sessions where a defined event happened. Nothing is weighted or predicted. The report page states the exact rule in one line under the verdict, so a yes is never a matter of opinion.
Does the opening gap fill?
Counted as: price trades back through the prior session close during the session, XAUUSD, full trading day.
The n on every row is the number of sessions behind the number. Conditional questions have smaller samples than daily ones, that is expected. Anything under 200 observations is flagged amber so you read it as suggestive, never settled.
The badge is shorthand for how far the rate sits from a coin flip. It is a label on the distance from 50%, not a recommendation. We phrase every question so the tendency reads above 50%, and rates are shown rounded to whole percents.
The full-history rate is 96%, but over the last 90 days it has run 91%. When the recent window diverges from the full sample by more than a few points, the row carries the flag.
A base rate is a long-run average, and regimes change. The markers show when the last-90-day rate has moved materially from the full sample. Free to see that it moved. Pro shows you exactly when it turned.
These are histories, not forecasts. The market has no obligation to repeat.
Back to the hit rates