CRE loan comps search over the bank’s own loans — what it underwrote, and whether it held up
Cap rate, rent and vacancy assumptions backed by the bank's own deals.
See one case, screen by screen ↓It is a search over the bank’s own closed CRE loans, not market sales. Ask in plain words, and it finds the comparable loans, the cap rate, rent, vacancy and expenses the bank underwrote on each, and how each loan has performed since — with the memo or appraisal page behind every figure.
How to choose AI solutions for lending →
Updated
Every CRE loan rests on a handful of assumptions — cap rate, rent, vacancy, operating expenses — and the bank has made each of them before, on a similar building down the road. But the answer to “how did we underwrite this last time, and how did that loan do?” sits in old credit memos, appraisals and review reports, with the outcome in servicing. Nobody can search them.
So the analyst asks around, or starts from the broker’s pro forma and the appraisal, whose sales may date from when money was cheap. Two loans a mile apart get two different cap rates, exceptions are granted without anyone checking how the last one turned out, and the credit officer sees inconsistent underwriting across deals.
Estimated split for one loan’s precedent research, by hand and with the solution.
Six specialist agents drop the bank's comparable loans on a map, read every memo and judge what held up; the analyst chooses and the officer signs.
Sycamore Yard Apartments, a $33.5 million acquisition loan, goes to credit committee on Thursday 15 October. Here is what the analyst does on Monday, screen by screen, in the working solution.
Priya opens a map of Calder Point Bank’s own lending: 1,284 closed CRE loans since 2014 across North and South Carolina, Georgia and Tennessee. The loan in hand sits beside the question bar — Sycamore Yard, 228 units in Huntersville, with the broker’s 5.15% cap and 4% vacancy.
“Every pin opens the memo page it came from. Borrower names follow your entitlements.”
Ask what the bank lent on suburban medical office in the Southeast since 2022, and eight loans drop across Atlanta, Nashville, Charlotte and Greenville. Click Peachtree Corners Medical Pavilion: underwritten at a 6.75% cap, actual NOI 8% above plan, with underwriting at closing and today side by side. Murfreesboro Physicians Park missed by 11% after an anchor with 31% of the space left at expiry.
“Underwritten 6.75% cap, actual NOI 8% above plan.”
Six agents take the question. The criteria come back as chips she can change — multifamily, garden, Class A and B+, Charlotte MSA, closed since January 2023, built 2016 or later, 180–320 units. Of 31 multifamily loans, 9 are comparable and 22 are left out with the reason. The memo and appraisal reader reads 27 documents and records every figure with its page.
“Finished in 16 s · 6 agents.”
Nine pins land around Charlotte, each with the cap rate the bank underwrote. The bank underwrote them at 5.40–6.10%, median 5.75% — about 45 basis points above the appraisals. The broker’s 5.15% cap is below all nine. Every sentence of the answer carries its citation.
“Seven of nine are at or above plan. The one that missed, Juniper Trace, was underwritten at 4.0% vacancy — the broker’s number on Sycamore Yard.”
The pin card leads with one line: underwritten 5.40% cap, actual NOI 17% below plan. Vacancy was underwritten at 4.0%; the loan is 86.4% occupied today, DSCR down from 1.30x at closing to 1.08x, risk rating 6 and on the watch list. 612 new units opened within a mile, and concessions reached six weeks.
The “Did it hold up?” lens recolours every pin for NOI against plan, vacancy or expenses. On expenses, four of nine held up: the five loans closed before mid-2024 ran 4–9% over plan after insurance renewals. Those underwritten since carry more, and are on plan.
Beneath the map, the nine loans — $257.8M in all — are ranked by closeness to Sycamore Yard: Mallard Bend 96, Hartley Mill 95, Linden Row 91, down to Quail Ridge Flats 71. Each row shows the underwritten and appraisal cap, vacancy and DSCR from closing to now, and NOI against plan.
Opening Juniper Trace shows occupancy month by month against the 96.0% underwritten, the operating statement at closing against the trailing twelve months, and what the memo said. The vacancy of 4.0%, below the 5% policy floor, was approved as an exception by Joan Pemberton, chief credit officer, with lease-up to 97% in nine months as the mitigant.
Priya sees the broker’s pro forma, the appraisal and the bank’s nine comps side by side, then clicks “Use the comps medians”: in-place rent $1,561, 6.0% vacancy, $6,600 expenses and $250 reserves a unit, 5.75% cap. NOI is $2,737,842 and value $47.6M. At $33.5M the DSCR is 1.14x against the 1.25x minimum; the loan supported is $30.6M, the lowest of coverage, loan to value and debt yield.
“At your assumptions the request needs a DSCR exception at 1.14x. On these comps the bank has granted one DSCR exception, at 1.24x (Linden Row).”
Priya signs and saves research note RN-2026-1147 — the question, the nine comparable loans, what the bank assumed, how those loans performed, the sizing and her recommendation — and sends it to Marcus. He reviews and signs it; then the note can be inserted into the credit memo, and its PDF carries every source.
“Size the loan to $30.6M, or present $33.5M with a DSCR exception at 1.14x — below any exception the bank has granted on these loans.”
The same loan, seen by the people who carry it to committee — what their week looked like, and what it looks like now.
Six specialist agents drop the bank's comparable loans on a map, read every memo and judge what held up; the analyst chooses and the officer signs.
Turns a plain-words question into the criteria for comparable loans — property type, class, market, vintage, size and period — and shows them as chips the analyst can change.
Searches the bank’s closed loans, scores how comparable each one is, and explains why a loan is in or left out.
Pulls the underwritten cap rate, rent, vacancy, expenses, DSCR and exceptions from each credit memo, appraisal and review report, with the page.
Joins occupancy, coverage, risk rating and payment history from servicing, with the date of each figure.
Tests the assumptions and the sizing against credit policy and the supervisory loan-to-value limits, and reports which test binds.
Writes the cited answer, the comps table and the research note, and says plainly where the bank has little or no history.
Analyst chooses the assumptions; officer reviews and signs. The agents propose; a named person decides.
Ask in plain words, or tell it to size the loan in hand. Every answer cites the memo, review or servicing record behind it.
Why is Juniper Trace on the watch list?
It closed in March 2023 underwritten at 4.0% vacancy, below the 5% floor, approved as an exception on the strength of its lease-up. Then 612 new units opened within a mile; it gave up to six weeks free and occupancy fell to 86.4%. Coverage is 1.08x against 1.30x at closing, so it was downgraded to 6, special mention.
Which of the comps needed a policy exception, and how did they do?
Three of the nine: Juniper Trace (vacancy 4.0%) is at 1.08x and special mention; Linden Row (DSCR 1.24x) is at 1.29x, pass; Hartley Mill (24 months interest-only) is at 1.27x, pass. The vacancy exception is the one that failed.
Size Sycamore Yard at our median cap rate and 6% vacancy
On the bank’s medians — in-place rent $1,561, 6.0% vacancy, expenses $6,600 a unit, $250 reserves and a 5.75% cap rate — NOI is $2,737,842 and value $47.6M. At $33.5M the DSCR is 1.14x against the 1.25x minimum; the loan these assumptions support is $30.6M.
Add the exception history to the research note
Added as section 7 of RN-2026-1147: the three exceptions on the comparable loans, who approved them, and how each loan covers today. It closes with the point Marcus will ask about: the bank has not granted a DSCR exception below 1.24x on these loans.
13 screens from the working solution, on its sample data. Pick one to see it large.
What counts as comparable, and the policy minimums the loan is tested against, are settings — versioned and on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| How far from the subject a loan may be | Same MSA | Same MSA · Within 15 miles · Same submarket |
| Comparable loans closed since | January 2023 | January 2023 · January 2022 · January 2024 |
| Year built, for apartments | Built 2016 or later | Built 2016 or later · Built 2010 or later · Any vintage |
| Size band either side of the subject | ±40% | Units or square feet either side of the subject, stepped up or down |
| Minimum debt service coverage (multifamily) | 1.25x | 1.20x · 1.25x · 1.30x |
| Underwritten vacancy floor | 5% | 4% · 5% · 6% |
| Interest-only period | 12 months | None · 12 months · 24 months |
| Who reviews research notes before committee | Marcus Hale | Marcus Hale · Joan Pemberton |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample loan data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (MBA, 17% of commercial and multifamily mortgage balances mature in 2026; Trepp CMBS delinquency rates, September 2026). People, banks, borrowers and properties named on this page are fictional — characters and sample data in the working solution.
It is a search over the bank’s own closed CRE loans, not market sales. Ask in plain words, and it finds the comparable loans, the cap rate, rent, vacancy and expenses the bank underwrote on each, and how each loan has performed since — with the memo or appraisal page behind every figure.
Market comp data tells you what buildings traded for. Only the bank knows what it assumed when it lent and whether that assumption held up. This solution reads the bank’s credit memos, appraisals, review reports and servicing, and colours each loan by whether NOI, vacancy or expenses held up against plan.
No. The agents find, read and test; the analyst chooses the assumptions and recommends the amount, a credit officer reviews and signs the research note, and committee approves the loan. The policy checker never waives or reinterprets a policy.
It says so. Where only a few loans exist you are told how few, and where none exist that is said instead of a guess — for example, that the bank has no medical office loan outside four metros since 2022.
Yes. Every exception granted on a CRE loan since January 2023 is listed by type — vacancy, DSCR, interest-only, loan to value and others — with who approved it, the mitigant and how the loan is doing now, and can be exported as an exception report.
Borrower names follow the bank’s relationship entitlements: shown to people entitled to the relationship, hidden for everyone else. Relationship managers see their own borrowers, and the comparable loan finder never includes a loan the user may not see.
Closed-loan terms, approved credit memos and annual reviews, appraisals and review reports, monthly servicing and the credit policy. Every night, new memos, appraisals and servicing records are read and the assumption ranges refreshed.
The Agentic Solution Engine builds and deploys it from your requirements — your credit policy, memo and note templates and a sample of closed loans — and it goes live once every quality gate has passed. We will walk you through it on your own loans first.
Eighteen months of maturities ranked by refinance gap, evidence attached.
1-min demoConstruction Draw ManagementDraws funded in a day, not a week, every line tied to its invoice, waivers and inspection.
1-min demoRent Roll & Operating Statement AnalysisNo re-keying, and every rent checked against its lease.
1-min demoWe’ll run CRE Loan Comps Search on a sample of your own closed loans and credit memos.