Each public post that states a position is graded over a fixed horizon against its own target, stop and against Bitcoin. Deleted posts stay, because every post is archived.

Sampled KOLs (marked provisional) are drawn in strata of quarter, coin group and direction, re-weighted to the population, and reported with the wider of the seed interval and the bootstrap interval. They carry no rank.

Unpriced calls: up to 10% no effect; 10–20% a zone is shown only when the worst and best case land in the same zone, and the range is shown; above 20% no zone and no rank. No price is ever estimated.

One exit closes only the most recent open call of the same asset and direction within 30 days. Older calls close only when the exit explicitly refers to them.

We never accept payment for a score. A Trust Score cannot be bought, sponsored or changed for a fee. Referral links live on their own page and never appear next to a score.

Rules v1.14w

Investigator findings. A documented finding by an investigator (for example ZachXBT) or a court counts even if it predates the period we grade. Only primary sources with a link count. The deduction depends on the strength of the evidence and its age.

Expired calls. A call that reaches neither target nor stop by its horizon counts with half weight in the hit rate. Its actual return at the horizon end counts in full in the return part of the score.

Resolution rate. The share of resolved calls that reached their target or stop. Shown next to the score, not part of it.

Unpriced calls. A call with no market price is never estimated. The share is shown for every KOL.

Official tier list: the boundaries

S ≥ 75 · A ≥ 65 · B ≥ 55 · C ≥ 45 · D ≥ 30 · F < 30, applied to the unrounded Trust Score.

Why these numbers: 50 is the midpoint of the Neutral zone (41–60), so C (45–55) is centred on it and means “no clear edge either way”. Each tier above is 10 points wide, roughly the half-width of a typical 90% range here, so one tier step is about the smallest difference the data can tell apart. S starts at 75, inside the Trusted zone (61–80). D (30–45) covers the upper Caution zone and the bottom of Neutral; F (< 30) is reserved for records that clearly lost against target, stop and Bitcoin.

The tier is a summary, not a verdict. When the 90% range crosses a boundary, the tile lists every tier the range covers, for example “C (range D–A)”. KOLs without a rank or a zone keep their score-based tier but carry a “provisional” badge. Each tile links to the KOL page with the full call log.

What the badges mean

Statistically close to #a–#b. The 90% range overlaps other scores, so the place is a band, not a single rank.

Provisional – small sample. Fewer than 30 effective calls, one call moves the score by more than 5 points, or the sample is too concentrated. No rank.

Provisional – too many unpriced calls. More than 20% of calls have no market price. No zone and no rank. No price is guessed.

Provisional – unpriced calls change the zone. Between 10% and 20% are unpriced, and the worst case and the best case fall in different zones.

Provisional – second sample disagrees. A second draw of the posts landed outside the 90% range.

Zone colours: Trusted green, Neutral grey, Caution yellow, Fade red. A name with no zone is shown as “No zone”.

Nothing here is financial advice. Figures are reproduced from the archived record and can be regenerated from the source tables.