Risk model settings

Settings → Risk model, needing risk_configuration.manage to change and risk_configuration.view to read.

Three things live here.

The tier meter

Two handles on a fixed 1–5 scale. See Tiers for the rules, including the two settings that are rejected because they would leave a band unreachable.

Changing the boundaries affects future assessments only. Existing ones keep the thresholds they were scored against.

The review cycle

Under the meter, how long after a decision each band is looked at again. Defaults are annual for high, two years for medium, three for low.

Higher risk must be reviewed at least as often as lower risk, and the form refuses anything else — a screen taking three free numbers would otherwise let an account review its high-risk merchants least often, which reads as a policy rather than as the mistake it is.

A merchant becomes due when interval has passed since its last completed review — the timestamp written when a review is approved or rejected, not when a check last ran. SchedulePeriodicReviewsJob opens the review, nightly, and opening one is idempotent: a merchant already carrying an open periodic review gets nothing new.

An unassessed merchant is not a low-risk merchant

entities.risk_level has a column default of low, and nothing writes it until an assessment runs. Low is the band reviewed least often, so a scheduler reading the band directly would give every unscored merchant the longest cycle in the configuration — looked at less often than one scored and found safe, with nothing failing to say so.

A merchant that has never been assessed is therefore scheduled on the shortest configured interval, and one with no completed review on record is due immediately. Absence of a score is not a finding of low risk. Held by an invariant in test/invariants/never_claim_clear_test.rb.

A country may shorten a cycle but never lengthen one. country_configurations.periodic_review_interval_days acts as a cap: a jurisdiction that mandates six months tightens a three-year band, while one permitting ten years does not loosen an annual band. The shorter of the two always wins.

Factors and weights

The list of factors, each with its weight, its value rows and their scores. See Factors and weights.

An empty account shows No risk factors are configured, so no merchant can be assessed and offers Install the default factors — eight factors reading data Verity already holds.

Two of them, country risk and legal form, arrive deliberately unscored. Those are compliance judgements, and Verity would rather show an obviously empty table than a plausible invented one.

FATF seeding

The country risk factor can be seeded from the FATF public statements — call-for-action jurisdictions at score 5 with a high override, increased-monitoring jurisdictions at score 4.

It is additive: your own rows are never removed or overwritten. The list carries its publication date and is flagged stale after five months.

What to review periodically

Check	Why
Are country risk and legal form still empty?	Four of 21 default weight points can never be evidenced until they are filled
Is the FATF seed stale?	The banner says so; FATF updates roughly three times a year
Do the weights still reflect policy?	They are relative, so changing one changes every band
Are the boundaries right?	Watch how many merchants land in each band

What this screen cannot do

It does not assess anybody. Thresholds label a score; they do not create one. A merchant's band stays unassessed until an assessment runs against them, and no change here alters that.