Acrescope
Home / Methodology / The Risk Model
Acrescope Risk Model · v0.2 · experimental

How the diligence signal is computed — and what it cannot claim

The Acrescope Risk Model produces a diligence signal — a rule-based composite, not a statistical model: seven weighted blocks over public records joined to one parcel. It is a reading aid for the cited evidence below it — not an appraisal, not an insurance rating, and not a substitute for professional diligence. This page is public and will stay public.

The seven blocks

Each block is clamped at its maximum; the raw sum is scaled to 0–100. Two rules govern everything: missing data contributes zero — a parcel we know little about is never scored high for it, which also means a sparse record can read deceptively calm. That is exactly why the signal is labeled experimental, coverage is shown beside it, and the interface presents it as a diligence signal rather than a risk verdict. No record is counted twice (realized burns and modeled fire tier share one clamp; slope belongs to the landslide block alone).

BlockMax pointsWhat moves itRecords
Distress18Recorded filings — foreclosure notices weighted heaviest, then federal tax liens, lis pendens, judgments, mechanics liens. Halved when the newest filing is over three years old; a bonus when filings are accelerating.County clerks & public trustees
Ownership & entity health10Dissolved, delinquent, or otherwise impaired registration status of an entity owner; out-of-state formation adds a small increment.CO Secretary of State
Fire & flood (fused)13One clamp shared by realized burns and modeled hazard so they can never double-count: a recent burn footprint outranks a Very-High modeled tier, which outranks High and Moderate. Special Flood Hazard Area adds a fixed increment.NIFC · CSFS CO-WRA · FEMA NFHL
Geohazards8Mapped landslide susceptibility carries the block; proximity to a mapped fault and EPA radon zone 1 add small increments.Colorado Geological Survey · EPA
Soils instability6Hydrologic group D/C, poor drainage class, severe non-irrigated capability class, and very low available water capacity. Slope is deliberately excluded — steepness is already counted by the landslide block.USDA SSURGO
Water resilience5No attached water right, weighted up when the parcel also carries fire or soil stress — reads the fused fire condition so it cannot re-count already-counted hazard.CO DWR · CDSS
Snow & cold burden5Dormant — contributes zero until a climate-normals table lands. It will activate as its own version bump, never silently.(planned)

What the number means — and does not

How accurate is this?

Not yet calibrated — and we will not pretend otherwise. A defensible accuracy claim needs point-in-time scores checked against what later happened. We began snapshotting every parcel’s score in June 2026 and snapshot monthly; when twelve months of snapshots can be measured against observed outcomes (recorded foreclosure notices, actual burn perimeters), we will publish the hit rates here — whatever they turn out to be. Until then, v0.2 is labeled experimental, and the cited records are the product.

Changelog

v0.22026-06

Added geohazards, soils, and realized-burn blocks; fused the fire signal into one clamp (realized burn outranks modeled tier); added coverage confidence and the High-Friction Rebuild Risk flag; score rescaled to use the full 0–100 range.

v0.12026-05

First composite: distress filings, wildfire/flood tier, entity registration status, and water-rights count.

The records under the model

The model is only as current as its records — each domain updates on its own cadence:

County assessors
2026-08-22
CO Secretary of State
2026-08-22
NIFC
2026-08-22
County clerks & public trustees
2026-08-22
BLM
2026-08-22
CO ECMC
2026-08-20
CO DWR · CDSS
2026-08-20
Conservation easement registries
2026-06-11
CO DWR well permits
2026-06-09
FEMA NFHL
2026-06-03
County assessors & recorders
2026-06-02