Every statistic ORIN publishes links here. This page explains how each one is produced, what counts as a resolved outcome, and the conditions under which we will not publish a number at all — which, at time of writing, is most of them.
What a grade means
A setup is detected only when it passes every hard criterion listed on its Playbook page. Detection is binary; grading is not. The grade combines eight weighted factors — setup quality, trend alignment, location and reward:risk, entry location against the mean, volume confirmation, momentum, structural confirmation and volatility regime — read across four timeframes. Factors that disagree subtract rather than being excluded, which is why a strong setup in a poor location grades in the eighties rather than the nineties.
Score bands map to letters as follows: 90+ is A+, 80–89 A, 70–79 B, 60–69 C, 50–59 D, below 50 F. That mapping is fixed, and a card showing a letter inconsistent with its score is blocked at build time.
How resolution is defined
A graded setup resolves favourably if price reaches the target zone before touching the invalidation level, and unfavourably if it touches invalidation first. Setups that do neither within the resolution window for their timeframe band are recorded as expired and excluded from resolution rates — but counted in the sample, so an expired-heavy setup cannot look better by having its failures quietly dropped.
Resolution windows by timeframe band: 1–5 minute, 24 hours · 15 minute–1 hour, 5 days · 4 hour, 15 days · daily, 60 days.
The sample floor
No cell publishes a figure below 500 resolved observations. Under that it displays collecting with its current count, so you can see it filling rather than being told to wait. This threshold is not a stylistic choice: below a few hundred observations, a resolution rate on a setup with a genuine edge is statistically indistinguishable from one without.
The rule we do not bend
No fabricated, extrapolated or estimated statistic appears anywhere on this site. Not as a placeholder, not as an illustration, not with an asterisk. In the codebase this is structural rather than editorial: a published statistic cannot be constructed from a literal number, only from a reading that carries its own sample size and source, and the build fails if that constraint is violated.
What these numbers cannot tell you
- They are not your results. A resolution rate describes what happened to graded setups, not what happened to anyone trading them. Entry timing, slippage, sizing and the decision to take the trade at all sit between the two.
- They exclude what the chart cannot see. Scheduled events, halts and news are not in price data. Resolution rates measured across periods containing them are describing an average that includes outcomes no technical read could have anticipated.
- Past resolution is not a forecast. A setup that resolved favourably 60% of the time historically has no obligation to continue, particularly if the behaviour becomes widely traded.
Auditability
Every published cell traces to a query against the event log, and every scoreboard is downloadable as CSV with its source and retrieval date in the header. If a figure on this site cannot be traced to a query, it should not be there — reporting one that cannot is a bug, and the corrections policy covers what happens next.