Where the numbers come from

Where the data comes from

All of it is public regulatory filing data. Nothing is purchased, scraped from a vendor, or modelled.

• FFIEC call report filings — the primary source. Ingested from the quarterly CDR bulk files into a dedicated store covering forty quarters of filings for every FDIC-insured institution, at the individual reported-item level rather than as summary ratios.
• FFIEC UBPR— the Uniform Bank Performance Report's supervisory ratios and published peer figures, used exactly as published and never recomputed. These are the external check on our own arithmetic.
• FDIC BankFind — institution identity and the directory: name, charter class, primary federal regulator, branch and merger history.
• FDIC Summary of Deposits — branch locations and branch-level deposits, which is what makes geographic peer groups possible. Annual, as of 30 June.
• Enforcement registers — published formal actions from the OCC and the Federal Reserve, joined to institutions by charter number and exact identity match. Coverage is stated honestly per institution: where your own primary supervisor is not a loaded agency, the surface says so rather than implying a clean record.

Everything refreshes on the regulators' own cadence — quarterly for filings, annually for Summary of Deposits. The quarter a screen is showing is its data vintage, and it is always stated: pull the same call report and you will see the same figures.

The supervisory CRE basis

CRE concentration is measured the way examiners measure it: non-owner-occupied CRE (construction and development, multifamily, and non-owner-occupied nonfarm nonresidential) as a percentage of tier 1 capital plus the allowance for credit losses. The denominator includes the ACL per the 2020 guidance — a CECL-adjusted basis — which is why it can differ from older tier-1-only figures. The ready-made FFIEC fields are preferred for this calculation, and every input shows in the trace.

FIL-104-2006 / SR 07-01 (interagency CRE guidance); denominator per SR 20-8 / FIL-31-2020 (UBPR3792, tier 1 + ACL).

The conjunctive 300% / 50% test

The 300% CRE criterion is not a standalone limit. The supervisory indicator is conjunctive: it is met only when CRE concentration is at or above 300% of capital and total CRE has grown by 50% or more over the prior 36 months. That is why an institution can show a high concentration and still pass — if the growth leg is not also met.

The 36-month growth figure comes from the FFIEC field UBPRNL33, the "3 year NOO CRE growth ratio" (UBPR concentrations of credit, page 7B). It is a cumulative 36-month growth of non-owner-occupied CRE — the same loan basis as the 300% numerator — so the two legs are measured consistently. Its trace shows the value, the 50% threshold, and whether the concentration leg is also met.

UBPRNL33 = (current NOO CRE ÷ NOO CRE 36 months prior × 100) − 100 — FFIEC UBPR user guide, page 7B.

CBLR institutions

If your institution has elected the community bank leverage ratio framework, risk-weighted-asset-based capital rules do not apply. The CBLR election is read from the FFIEC CDR and those rules are suppressed rather than flagged — a metric the framework makes irrelevant for you is not a finding.

The CBLR standard the engine tests is 8%, effective 1 July 2026. Where a dated schedule applies, the applicable figure is the one in force for the quarter being read, and the transition's grace period is stated rather than silently applied.

91 FR 22973 — community bank leverage ratio, 8%, effective 2026-07-01.

Peer figures — two kinds, always labelled

There are two different peer numbers in the product and they are not interchangeable. Every surface that shows one says which it is.

• FFIEC's published peer figure — taken straight from the UBPR for your regulatory peer group. It is a trimmed average: the FFIEC excludes the top and bottom 5% before averaging. It is an average, nevera median, and it is not recomputed here. This is the figure an examiner will be looking at, and it renders as "trimmed avg, theirs".
• Your own group — built in Find & Compare Banks from measures, geography or a shape you draw. A custom group reports an untrimmed median and an ordered position among observed values, never an interpolated percentile against a distribution that was never published. Below five members a reading is suppressed rather than shown thin.

Both are shown side by side rather than one being chosen for you. A group you built yourself can be made to say almost anything; keeping FFIEC's published figure permanently in view is what stops that from being invisible.

No invented warning bands

The only levels the product draws are the ones a regulator published. There are no product-supplied "approaching" bands invented to make a chart look useful — an earlier 80% construction-and- development warning line was removed for exactly that reason, and high-LTV lending is tested against the published five-band supervisory table rather than a single invented line.

12 CFR 365 appendix A — supervisory loan-to-value limits.