Research verdicts counter
How many full backtests run in the desktop app ended in a GO verdict and how many in a NO-GO verdict, counted from users who chose to share them. The figures are read from the database when the page is served and are at most one minute old.
Totals
No verdict has been shared yet. The counter starts at zero and grows only when a desktop user ticks the optional box and completes a full backtest whose verdict is GO or NO-GO.
How to read these figures
Each count is one completed full backtest whose verdict was GO or NO-GO. Preview passes, results replayed from the app's cache and runs with any other verdict are not counted. The column n is the number of verdicts for a market family; a family with fewer than 30 verdicts is marked as a small sample, because its share can move by several points with a single run. The installs column counts the distinct random install identifiers behind a row, so a reader can see when one machine accounts for most of a figure.
Rows are listed in the alphabetical order of their family key and never by a result. The counter holds no return, profit, Sharpe ratio or other performance figure, so no market can be ranked by one here.
The figures are self-reported by the desktop app and are not audited. A NO-GO share describes the strategies that users of the app chose to test; it says nothing about how often a strategy works in a market. The same idea tested many times counts many times.
What is shared, and by whom
Sharing is off by default. A user turns it on by ticking an optional box on the sign-in screen of the app, and turns it off in Settings, General, Data sharing. Each shared verdict carries exactly five fields: a random install identifier created when the box is ticked, the verdict, the market family, the day of the run in UTC and the app version. It carries no account, no email address, no strategy, no parameter and no price. Records are deleted after three years, or earlier from the app. The full description is in the privacy policy.