Which thresholds produced these numbers?

Rendered from the engine's own /methodology response, so what is on screen is what was applied. No threshold is written into this dashboard, and the list below is whatever the engine served rather than a set of keys this page knows about.

These thresholds are chosen, not calibrated

The thresholds were chosen based on the magnitude of the Blend incident of February 2026 and on conservative judgement, not calibrated against a set of incidents. Every flag is reported separately so that consumers can apply their own thresholds.

What is this document?

Enough to check that the figures elsewhere on this site came from the version named here.

Version in force: 1.0.8-draft
The version the engine reports, not the one this page was built against
Thresholds served: 15
Read by key name; new keys can appear without a major version bump

The full methodology document

Every threshold, as served

Keys are shown verbatim, because the key name is what a consumer looks up in the contract. Every key ending in Pct is a percentage rather than a fraction — that is the one unit convention the contract states, and it is the only one applied here.

Every threshold served by the methodology endpoint, with the asset a figure is denominated in where the engine reports one.
Key	Value
`genuineTradeStaleDays`	30
`genuineTradeWarnDays`	7
`holderTop10HighPct`	80 %
`holderTop1ExtremePct`	50 %
`liquidationDelta`	0.1
`liquidationHaircut`	0.5
`manipulationCheapAbsolute` denominated in USDC:GA5ZSE…K4KZVN	10,000 USDC
`manipulationCriticalDelta`	0.5
`manipulationMargin`	0.25
`manipulationRatioLowPct`	0.1 %
`oracleWindowSeconds`	900
`priceDivergencePct`	10 %
`spreadExtremePct`	20 %
`thinDepth5PctAbsolute` denominated in USDC:GA5ZSE…K4KZVN	50,000 USDC
`washTradeSuspectedPct`	50 %

A threshold denominated in an asset is shown with its full code and issuer. A bare ticker would name nothing: ninety-seven distinct assets share the AQUA ticker alone.