The Trust Index

The scale

Every scored event gets a Trust Score from 0 to 100, read on five bands. The score is conservative by design: a market has to earn its way up.

Four questions, one number

The score blends four measured pillars. Three of them — the safety questions — form the Integrity axis (80% of the score); the fourth is the Trade quality axis (20%). Safety outweighs convenience on purpose: a market you can trade easily but can't trust is the worst kind.

Blend, then cap

The pillars are blended by weight — and then hard caps apply. A great average can't paper over a disqualifying weakness:

• Market integrity, Info fairness, or Resolution quality below 30 caps the total near 40–50.
• A severe activity flag caps the total at 30. Trade quality lowers the blend but never caps it.

Screens, in plain language

Behind the Integrity pillars run individual screens — could a small group know this outcome early, can a participant trade a result they control, did prices move before the news, did one trade dominate the tape. Each screen reports in desk language, never pass/fail:

Screens that don't apply to a market's type say so, and screens whose data source isn't connected yet are reported as awaiting data — never silently counted as clean.

The honesty rules

• Missing data is never guessed. When an input can't be measured, the remaining pillars are reweighted and the stated confidence drops — the score never fills a gap with an invented number.
• Priors are labeled as priors. A score that comes from the market's structure rather than a measurement says "structural prior" right on the card.
• One computation, everywhere. The score is computed once, server-side. The report page, the PDF and Markdown exports, and this site all render the same payload — no surface recomputes or rounds its own version.