Case study · M&A advisory · Credit unions

Credit union screening platform on NCUA call report data

Replaced a hand-maintained Excel heat map with a live screening platform over every federally insured credit union, scored the way the firm's own analysts score them.

An investment bank advising credit unionsLive ↗
5,216
credit unions covered
20
quarters of history
296M
call report facts loaded
Days
from kickoff to first live version
From the regulator's quarterly release to a ranked universe
Figure 1
SOURCENCUA quarterly release~20 wide tables, ~3,400 account codesINGESTIONMelt to factscharter · account · quarter · valueVersioned definitionsSchema drift recorded, never silentWAREHOUSEBigQuery20 quarters, 296M facts, clusteredDerived metrics26 FPR ratios + composite scorePRODUCTHeat mapFull universe, ranked and filteredPeer sets and drill-downPer-institution historyGated APISame-origin, keyed accessEvery ratio reconciles to the regulator's own performance report. Every score is computed once.
The build
Stack
PythonFastAPINext.jsBigQueryCloud RunFirebase Hosting

The problem

The firm advises credit unions on mergers and capital. Its edge was a screening heat map: a spreadsheet that pulled ratios from the NCUA 5300 Call Report for a shortlist of institutions and ranked them on the partners’ own composite score. It worked, and it was slow. Every quarter someone re-downloaded the regulator’s data, re-pasted it, and re-checked the formulas. Coverage was limited to the institutions someone had already thought of.

What we built

A screening platform over the full universe of federally insured credit unions, updated from the regulator’s quarterly releases.

  • A long-form fact store. The regulator publishes roughly twenty wide tables per quarter with about 3,400 account codes. We melt them into one fact table, keep every account definition versioned, and record each ingestion so a changed column never silently changes a number.
  • The firm’s own metrics. Twenty-six derived ratios from the regulator’s financial performance report, plus the partners’ composite score, percentile-ranked across the universe. Every ratio was reconciled against the firm’s existing reports before it went live.
  • The heat map, as a product. Filter by state, asset band, and any metric, then see the full universe ranked. Peer sets, per-institution drill-down, and the raw account series behind every number.
  • Full history in a warehouse. All twenty quarters, about 296 million facts, loaded into BigQuery and clustered so a single institution’s drill-down scans a fraction of a percent of the table. Storage runs under a dollar a month.

How it went

The first version was deployed within days of kickoff. The full twenty-quarter history, authentication, and the warehouse followed in the first week. The platform runs on the firm’s own cloud project, behind access codes, with a same-origin API so nothing is exposed that does not need to be.

What the client got

Coverage of every institution instead of a shortlist. Quarterly refreshes that take minutes instead of a day. A score that is finally consistent because it is computed once, not retyped. And a system the firm owns and can extend.

A white-labeled version of the same platform is available as a private preview for other advisors and institutions working with credit union data. Ask us about it.

Want something like this built for your firm?

Two weeks and one working tool is the usual starting point. Tell us the workflow and we will scope it on the first call.