Most teams work the credit union market from a static list and an educated guess. CUSignals consolidates all ~4,300 U.S. credit unions into a single panel, keeps it current with a daily news feed, and answers three questions about them: who is running short of funding, who is likely to merge out, and who runs which platform. The first two are scored against what actually happened next. The third is a register of observed facts, not a forecast. We publish which is which — including the signal we retired this week for failing its own test.
Two scored signals, one register
A credit union analyzing itself already has a core system. CUSignals serves the other side of the market — the funding desks, acquirers, advisors and vendors whose revenue depends on reading an institution accurately from the outside. The two signals below are scored against outcomes observed after the fact; the register is observed fact to begin with. Each card says which it is.
Deposit brokers · FHLB · whole-loan desks
“Who is running out of funding runway, and who is cash-rich and hunting for yield?”
Acquiring credit unions · advisors · investment banks
“Which credit unions are likely to merge out — and what is each one worth to an acquirer?”
Fintech and CUSO vendors · Sales and RevOps leadership
“Which credit unions run my competitor, since when — and when is the contract up?”
How it works
Collection is only the starting point. The product is the transformation, the daily news layer that keeps it current between filing cycles, and the fact that every row carries the evidence that produced it.
Collect
Drawn from multiple data sources and processed through a proprietary pipeline: each input is resolved to the correct institution, deduplicated and normalized against a verified account-code dictionary, producing one clean panel covering every credit union, rebuilt on each new filing cycle.
Derive
Capital, funding, growth and credit-quality ratios, quarter-over-quarter and year-over-year momentum, and rank within peer group and asset tier.
Score
Documented financial dynamics with published weights — not a black box. Each score decomposes into weight × feature, so you can see what drove it.
Activate
Updated daily from news and industry channels: leadership changes, vendor announcements, M&A activity and regulatory actions, each dated and tied to a charter. Scores re-base on every new filing cycle, and the movers view names who crossed a band, whose funding turned stressed, and whose participation role flipped.
Validated, not asserted
Two of the three signals are now scored against outcomes observed after the fact rather than against their own training data — Merger Radar against mergers the regulator attributes to a credit union's own condition, Liquidity Radar against borrowings that actually rose. The third was scored and failed. Saying which is which is the point of this section.
Measured on 89 events in the held-out period, which puts the standard error on that AUC near 0.04 — none of it should be read to a third decimal. The test window held no distress merger above $500M in assets, so above that threshold the model is untested rather than weak, and it puts no names on the list: of the top 50, 49 are under $100M. Merger Radar is, on this evidence, a small-credit-union product. Vendor Signal has been scored against announced adoptions, and it failed — AUC 0.509 and 0.543 over two cohorts, which is the coin-flip band, where ranking by total assets alone scored 0.677 and 0.665, and where not one of the 25 highest-scored credit unions announced an adoption in the year that followed. The label observes announcements rather than purchases, so a zero may be a quiet buyer and the measured precision is a lower bound — that qualifies the result, it does not rescue it. Liquidity Radar has now been scored against observed funding behaviour, and it held — its stress score ranked the credit unions whose borrowings actually rose in the following quarters at AUC 0.76 to 0.83 across three as-of quarters, and never below 0.73 across every horizon and materiality threshold tested, beating borrowing momentum, loan-to-share and asset size in every combination. Its funding-need flag caught about half of them at two to three times the base rate. That label reads borrowings only, and the filing carries no separate brokered-deposit line, so a credit union that funded itself in the brokered market reads as a miss: the recall is a floor rather than a measurement. Backtests describe a past period and are not a forecast. The disclosures set out every limitation in full.
Questions
The answers an evaluator needs before a first conversation, stated plainly.
Read the full FAQ — buying and billing, day-to-day use, security and compliance.
Not seats, not queries — the size of the account universe you can see and export, updated daily.
