CRE loan comps, explained.

One apartment loan shows what loan comps are for

Sycamore Yard's $33.5 million request, from package to servicing

1. The borrower and broker send a packageOffering memorandum, rent roll, trailing operating statements, purchase contract. For Sycamore Yard: a $33.5 million request on a $58.9 million purchase.
2. The bank orders an appraisal and reviews itAn independent appraiser values the building; the bank's review appraiser checks the work. Here: $59.4 million at a 5.10% cap.
3. The analyst sets the assumptions and sizes the loanRent, vacancy, expenses, reserves and cap rate, tested against credit policy. This is where the bank's own comparable loans are read, or skipped.
4. A credit officer reviews and committee decidesThe memo explains every assumption; exceptions to policy need a written justification and the right approver.
5. Servicing records how the loan turns outOccupancy, expenses and coverage arrive every year. They are the outcome the next analyst needs, and they rarely get back to the memo.

Step 3 is the subject of this guide. The bank, buildings, people and figures in the example are fictional.

Four surprises about CRE underwriting assumptions and appraisals

Loan comps, cap rate, DSCR and policy exception

Thirteen terms are defined in the one-page summary.

Sales comps, lease comps and loan comps: three kinds of evidence a CRE lender can use

In shortSales comps show what similar buildings sold for. Lease comps show what similar space rents for. Loan comps show what the lender itself assumed when it lent on similar buildings, and whether those assumptions held up. Only the third pairs an assumption with an outcome, and only the lender holds it.

"Comps" is short for comparables, and in commercial real estate it usually means market evidence. An appraiser picks recent sales of similar buildings, adjusts them for differences and concludes on a value. A leasing analyst does the same with signed leases to support the rent line. Both are evidence about the market on a given day. The OCC's handbook expects a lender's analysis of a property's income to consider "Comparable rental rates, operating expenses, and sales prices", and says that "Inadequately supported or questionable analysis should be challenged."5

A lender has a third set of comparables that no data vendor sells: its own loans. For each one it knows the rent, vacancy, expenses, reserves and cap rate it accepted, the exceptions it granted and why, and what the building has done since. The credit analyst's official job description already assumes this comparison. O*NET's profile of the occupation lists the task "Compare liquidity, profitability, and credit histories of establishments being evaluated with those of similar establishments in the same industries and geographic locations."6

A bank's loans are also a different population from the market's. Federal Reserve economists found that banks, life insurers and securitization lenders "differ in how they are funded and regulated, and therefore specialize in loans with different characteristics."7 Comparables drawn from the whole market, or from securitized loans, describe somebody else's book.

When each kind of evidence should lead

The rules behind CRE underwriting assumptions and exception reports, as of October 2026

In shortNo rule tells a bank what cap rate or vacancy to use. The rules require a written, board-approved real estate lending policy with clear and measurable standards, supervisory ceilings on loan to value, monitoring of the local market, documented exceptions, and reports that examiners review. Two texts changed in 2026: the model risk guidance in April and the OCC's lending booklet in July.

The supervisory loan-to-value ceilings, and a bank's own limit inside them

The guidelines say "Institutions should establish their own internal loan-to-value limits for real estate loans", that the supervisory figures "should not be interpreted to mean that loans at these levels will automatically be considered sound", and that a riskier subcategory deserves a lower internal limit.1 For a purchase, value means "the lesser of the actual acquisition cost or the estimate of value."1 Loans above the supervisory limits are capped in total at 100% of capital, and 30% for commercial and multifamily property.1

Rent, vacancy, expenses, reserves and cap rate: how five guesses become a value

In shortNet operating income is the rent the building can collect, less vacancy, plus other income, less operating expenses and reserves. Divide it by a cap rate and you have a value. The same building gives three values in the example below, $47.6 million to $59.4 million, because three parties chose different guesses.

The OCC defines each piece. NOI analysis "may assume market vacancy rates that are above or below actual vacancy rates, and expenses that may not represent an actual or immediate cash expense, such as management fees and reserves for capital replacements". "An important objective of the underwriting process is to develop an NOI that represents a stabilized estimate of income and expenses."5

Fannie Mae's multifamily guide, updated October 26, 2026, sets two floors a bank can borrow for its own policy: vacancy, concessions and bad debt together of at least "5% of GPR" (gross potential rent), and a replacement reserve of at least $200 a unit that "must be included whether the escrow is funded or not".20

One building, three sets of guesses

Fictional example. The analyst chose $6,600 of expenses, above the nine-loan median of $6,350, because the three loans closed since March 2025 average $6,617 after insurance renewals.

Where the $296,656 of NOI went

The broker's NOI is $296,656 a year higher than the bank's. Two points of vacancy are worth about $85,000; $400 a unit of expenses about $91,000; the missing reserve $57,000; the rest is rent. Capitalized at 5.75%, the reserve alone is worth about $990,000 of value, all of it missing from a pro forma that leaves the line blank.

Half a point of cap rate is worth millions

Holding the bank's NOI fixed, the value runs from $53.2 million at the broker's cap rate to $44.9 million at the top of the bank's own range. The OCC's warning applies in both directions: "Rising interest rates may lead to higher capitalization rates and lower property values without any change to the actual property's fundamentals."5 A cap rate taken from the cheapest-money years can be supported by many sales and still fail the test of "normal, orderly, and sustainable market conditions".5

Coverage, loan to value and debt yield: which test sets the loan amount

In shortA bank sizes an income-property loan three ways and lends the lowest: the loan whose payments the NOI covers at the minimum DSCR, a share of the lesser of price and value, and the loan on which the NOI earns a minimum debt yield. On the bank's own numbers, coverage is the test that sets Sycamore Yard's loan, at $30.6 million.

The OCC describes each test. DSCR "measures the borrower's ability to service its debt", and "The determination of an appropriate DSCR should consider the loan amortization period and the expected volatility of the cash flow." Debt yield "provides a measurement of risk that is independent of the interest rate, amortization period, and capitalization rate", and is "especially useful during periods of low interest and capitalization rates".5 The Federal Reserve's stress-test documentation says "lenders generally require a minimum DSCR between 1.2 and 1.5 to qualify for a loan".21

At the request, payments are $2,397,286 a year. The broker's NOI covers them 1.27 times; the bank's covers them 1.14 times and gives a debt yield of 8.2%, so the request would need two policy exceptions, coverage and debt yield, before anyone discusses price. At $30.6 million coverage is exactly 1.25x and the debt yield is 8.95%. Loan to value is 70.4% at the request, inside the bank's 75% limit: on these numbers, loan to value leaves room to spare.

Interest-only periods and amortization

Interest-only months raise coverage on paper, because no principal is paid, and leave more debt at maturity. The OCC's booklet expects policy to address "Requirements and limits on interest-only loans for stabilized commercial real estate", and in the April 2026 senior loan officer survey longer interest-only periods were among the most widely reported easings of the past year.5,22 In the example bank's policy, anything over 12 months is an exception.

What makes a past loan comparable, and how 1,284 closed loans become nine

In shortComparability is a set of written criteria: property type and class, market, size, year built and when the loan closed. Each criterion trades sample size for relevance. A good search shows its criteria, says why each loan was left out, and reports how many loans the answer rests on.

Within one property type, class and metro over three years, a community bank may have only a handful of loans; the example bank has nine. That is enough to see a range and an outlier, and too few to estimate a default rate. The answer should always say how many loans it rests on, and when the count is zero it should say so and step aside for market evidence.

Why a lender's cap rate often sits above the appraisal's, and how far appraisals miss

In shortAn appraiser estimates market value on a date, from recent sales. A lender chooses a rate it is willing to lend against for the loan's term. When the sales behind an appraisal date from 2021 and 2022, around the historic low in cap rates that the Federal Reserve dates to 2022,23 a bank that underwrites above them is adjusting for the move. Appraisals also carry their own error, about 11% of value on average.

The Federal Reserve's May 2026 Financial Stability Report says cap rates at purchase "have recovered from historical lows reached in 2022," "rising to a level just below its historical average in the most recent data."23 CBRE's survey for the first half of 2026 found the all-property median roughly flat, with rates up 50 to 100 basis points (0.50 to 1.00 percentage point) in key coastal markets since late 2025.24 An appraisal effective in September 2026 whose comparable sales closed in 2021 and 2022 carries those years' cap rates into the file.

How far appraisals miss the later sale price

The study's authors write: "We also do not find any evidence that the MAPE has significantly narrowed over the past decade compared to previous years when disregarding the large deviations that occurred during the great financial crisis in 2009."2 On a stabilized apartment at a 5.75% cap rate, an 8.6% value error is about the same as moving the cap rate 49 basis points (our arithmetic). A bank that sits about half a point above its appraisals is allowing for roughly one average appraisal miss.

The example bank against its own appraisals

A gap that steady is a practice worth writing down: an analyst who starts from the appraisal's cap rate would be about 45 basis points more aggressive than the bank has been. Juniper Trace is the caution. The bank underwrote it half a point above its 4.90% appraisal and it still missed, because its income rested on 4.0% vacancy. A careful cap rate on an aggressive NOI does not make a careful loan.

Judging whether an assumption held up: NOI, vacancy and expenses against plan

In shortCompare what the memo underwrote with what servicing shows now, line by line, with the date of each figure. NOI against plan is the summary; vacancy and expenses explain it. Research on thousands of securitized loans shows that income optimism is persistent by lender and predicts distress, which is why a bank's own record is worth reading.

Take the underwritten NOI from the approved memo and the trailing twelve months from the borrower's operating statements, on the same definitions. Then look at vacancy and expenses separately, because a loan can hit its NOI with high rents covering an expense overrun that will not repeat. A simple scale reads well on a map or a table: held up (NOI within 2% of plan or better), close (within 6%), missed (worse). The thresholds are a policy choice; write them down.

The expense row tells the sharper story. The five loans closed before mid-2024 run 4 to 9% over their expense plan after insurance renewals; the four underwritten since carry more and are on plan. A median built from all nine would reproduce the older loans' mistake, which is why the analyst in the example underwrote $6,600 a unit, above the $6,350 median. The FDIC's 2026 Risk Review describes the same pressure across the industry: higher operating costs and elevated vacancy squeezing property income.25

Optimism is a habit of lenders, and it shows up later as distress

Griffin and Priest found that income overstatement in these loans "varies widely and persistently across originators, is predictable ex ante" and "is highly predictive of pre- and COVID-period loan distress".4 Their data are securitized loans; a bank's book is a different population. The lesson carries over: habits persist, so a bank whose recent apartment loans ran over on expenses will probably do it again unless someone shows the analyst the record.

Policy exceptions: what examiners count, and why each one needs its outcome beside it

In shortExceptions are legitimate when they are justified in writing and approved at the right level. Examiners read their number and trend, and the OCC expects them aged and stratified by type and officer. The column that helps the next approver most is the outcome: how the last loan with the same exception is doing now.

The OCC's CRE booklet warns: "An excessive or significantly increasing number of exceptions to the CRE lending policy could indicate that the bank is unduly relaxing its underwriting practices, needs to revise its loan policy, or that its policies are inconsistent with the board's risk tolerance."5 The 2026 lending booklet adds the other side: "the absence of exceptions can indicate that the policy is too liberal or vague".15 Some banks, it notes, track exceptions by loan officer, broker or dealer.15

What a useful exception report carries

The OCC's 2026 booklet describes postmortems in the same spirit: comparing "each loan's terms and characteristics with underwriting guidelines", reviewing the borrower's condition "at the time of loan approval and over time", and "Assessing the adequacy and timeliness of collateral valuations".15 An exception report with outcomes beside it is a postmortem that runs before the loss.

Ten ways precedent research goes wrong, and what catches each

In shortThe errors fall into four groups: starting in the wrong place (the broker's pro forma), reading the wrong document (a draft, or the appraisal for the bank's view), comparing the wrong things (terms instead of assumptions and outcomes), and keeping the wrong loans (survivors only). Each has a simple control.

The 2026 CRE lending market in numbers: $3.14 trillion on bank books, $875 billion maturing

In shortUS commercial banks hold $3.14 trillion of CRE loans and the book is growing again. About $875 billion of commercial and multifamily mortgages matures in 2026, each one a re-underwriting. Bank delinquencies are low; securitized multifamily delinquencies jumped in 2026. Standards are easing from a tight level, mostly on loan size, pricing and interest-only periods.

MBA's 2025 survey of loan maturity volumes counts 17%, or $875 billion, of the $5.0 trillion of commercial and multifamily mortgages held by lenders and investors matures in 2026, after $957 billion scheduled for 2025; the figures are balances at December 31, 2025. Depositories face $396 billion, 21% of their balances.27

Each maturity is a new underwriting, done with current assumptions on a building first underwritten with older ones. Extension is part of that work, and a 2026 Federal Reserve study found that after the spring 2023 bank stress, "banks increased income and principal paydown requirements for extensions," which contributed to strong performance of extended loans.28

How CRE loans are performing

At June 30, 2026, 1.10% of insured institutions' multifamily loans, 1.23% of nonfarm nonresidential loans and 0.99% of construction loans were noncurrent, and year-to-date net charge-offs ran at 0.12% to 0.16%.31 Averages that low say little about the next loan; the variation that matters sits inside each bank's book. The OCC's spring 2026 report points to one source of it: "continuing rental rate declines in the Sun Belt, driven by the substantial supply wave of 2022-24", while "demand is steady and multifamily properties remain resilient."32

Standards are loosening from a tight position

Over the year to April 2026, "the most widely reported changes in terms … were higher maximum loan sizes, narrower spreads of loan rates over the bank's cost of funds, and longer interest-only payment periods", modest net shares lowered DSCR requirements for construction and multifamily loans, and the most cited reason was "more aggressive competition from other banks or nonbank lenders".22

In July, "a significant net share of banks reported that lending standards were at the tighter end of their range for CLD loans," with "relatively tight levels of standards for NFNR and multifamily loans".17 The survey panel is mostly large banks (56 domestic banks answered in July), and the largest of them led the easing. A community bank's own exception count is its private version of the survey.

What software can do with a bank's old credit memos, and what the 2026 rules expect

In shortSoftware can find comparable loans from plain-words criteria, read approved memos and appraisals into fields with the page behind each, join servicing with dates, and size the next loan on the result. The April 2026 model guidance leaves generative AI outside its scope, so the controls come from third-party guidance and the bank's own AI policy. The test that matters is whether every figure opens at its page.

The April 2026 interagency guidance, issued as Fed SR 26-2, OCC Bulletin 2026-13 and FDIC FIL-15-2026,14 defines a model as "a complex quantitative method, system, or approach that applies statistical, economic, or financial theories to process input data into quantitative estimates" and excludes "simple arithmetic calculations, such as those found within spreadsheets, as well as deterministic rulebased processes".13

A footnote places generative AI, and AI that acts on its own, outside the guidance, and adds that "a banking organization's risk management and governance practices should guide the determination of appropriate governance and controls for any tools, processes, or systems not covered in this document."13

The OCC and the Fed on AI that reads credit memos

The Treasury's 2024 report warns that generative AI "can be susceptible to 'AI hallucinations,' producing confidently stated but incorrect output".35 A figure that carries its document and page can be checked in seconds; one that does not cannot be checked at all. The third-party guidance keeps the responsibility with the bank, and NIST's framework names the work: GOVERN, MAP, MEASURE and MANAGE.10,36

The Equal Credit Opportunity Act and Regulation B apply to business credit too.37 A CFPB final rule effective July 21, 2026 states that ECOA does not authorize disparate-impact liability; the ban on disparate treatment is unchanged, and a lawsuit challenging the rule is pending.38 Consistent assumptions across similar loans are part of showing consistent treatment.

How you would know a search works

Replay it on closed loans whose outcomes you know, letting it see only what the bank knew at each committee date. Score three things: did it find the loans a senior officer would cite, do its figures match the pages, and did loans whose assumptions sat outside the comps' range perform worse. Then open citations at random, plant a draft memo beside an approved one, and ask about a property type the bank has never financed. A small clean test proves less than it seems.

Choosing each cap rate and approving exceptions stay human

• Choosing each assumption, and writing down why it differs from the comps.
• Judging a building nobody has lent on, where the honest answer is a count of zero.
• Approving exceptions, at the authority the policy names.
• Keeping valuation independent of loan production.8
• The responsibility for the loan, which stays with the bank that signs it.10

How CRE credit teams research precedent today, and where the hours go

In shortAn analyst asks colleagues, searches the loan system for loan numbers, opens memos one by one, pulls current figures from servicing and lines everything up in a spreadsheet. The finding and reading take most of the time, so the research is done fully for large or unusual loans and from memory for the rest. No public figure measures the hours.

Who looks for precedent: credit analyst, credit officers, review appraiser

Four places an analyst digs for a past loan before the memo

Nineteen hours, three-fifths spent finding and reading old memos

When precedent costs two days, it is consulted for large or unusual loans and skipped for the rest, which is where a bank relies on one officer's memory. That arrangement fails three ways: a new analyst underwrites to the outer limit because the inner target is invisible; a second officer across town applies a different one; and when the officer retires the bank loses the memory of which exceptions went badly.

The OCC's booklet puts the inner targets in procedures and guidelines.15 Written down, they become something an analyst can follow and an examiner can test.

Memo folders, bought market comps or a searchable loan record

Sycamore Yard Apartments, priced against nine of the bank's own past apartment loans

In shortThis is the example from the top of the page, inside the product. Priya Nandakumar, the CRE credit analyst, asks one question in plain words. Six agents, each a piece of software with one job, find the bank's comparable loans, read 27 documents and join servicing; the loans drop on a map, colored by whether each assumption held up. Priya sizes Sycamore Yard on the bank's own numbers, signs a cited research note, and Marcus Hale, the senior credit officer, reviews and signs it before Thursday's committee.

1

A question bar over a map of the bank's own lending

The first screen is a map of Calder Point Bank's closed CRE loans across North and South Carolina, Georgia and Tennessee: 1,284 loans since 2014, one dot for about five loans. The loan in hand sits in the rail: Sycamore Yard, 228 units in Huntersville, the broker's 5.15% cap rate and 4% vacancy, committee Thursday, October 15.

Three ready questions sit under the bar, and any question can be typed in plain words.

2

The criteria come back as chips, and every left-out loan has a reason

Priya asks how the bank has underwritten Class A garden apartments around Charlotte since 2023, and how those loans did. The question reader turns it into criteria: multifamily, garden, Class A and B+, Charlotte metro, closed since January 2023, built 2016 or later, 180 to 320 units.

The comparable loan finder searches 1,284 loans, finds 31 multifamily loans in the metro and keeps 9; the other 22 are left out as mid-rise, older or under 180 units. The memo and appraisal reader works through 27 documents and records 486 figures, each with its page.

3

Nine pins, each with the cap rate the bank underwrote and how the loan did

Nine numbered pins land around Sycamore Yard, ranked by closeness. Each carries the cap rate the bank underwrote and its income against plan. The verdict bar counts them: seven held up, one came close, one missed.

The answer beside the map cites every sentence: cap rates of 5.40% to 6.10%, median 5.75%, about 45 basis points above the appraisals; vacancy median 6.0%; the broker's 5.15% below all nine; and on the bank's own numbers the loan sizes to $30.6 million at 1.25x, against $33.5 million requested.

4

Juniper Trace: what the bank assumed, and what happened

Priya clicks pin 5. The card leads with one line: underwritten at a 5.40% cap, actual NOI 17% below plan. In March 2023 the bank underwrote 4.0% vacancy and $5,940 a unit of expenses at 1.30x. On September 30, 2026 the building is 86.4% occupied, expenses run $6,480 a unit, 9% over plan, coverage is 1.08x and the loan is rated 6, on the watch list.

The reason is in plain words: 612 new units opened within a mile, and concessions reached six weeks free. Credit memo and servicing record are one click away.

5

Sycamore Yard, sized on the bank's numbers

On the sizing screen the broker's pro forma, the appraisal and the bank's nine comps sit side by side. Priya clicks "Use the comps medians" and sets expenses to $6,600 a unit, above the median, because the three newest loans average $6,617. NOI comes to $2,737,842 and the value to $47.6 million.

The policy checker shows which test sets the loan: $30.6 million at 1.25x coverage, $32.2 million at the 8.5% debt yield, $35.7 million at 75% loan to value. At the request, coverage is 1.14x against 1.25x and the debt yield 8.2% against 8.5%, so two of the six policy tests fail and four pass. The warning on the screen names the coverage exception; the debt yield test in the list below it fails as well.

6

A cited research note, reviewed before committee

Priya signs and saves research note RN-2026-1147: the question, the nine loans, what the bank assumed, how those loans performed, the sizing and her recommendation to size the loan to $30.6 million or present the request with exceptions. She sends it to Marcus Hale.

Marcus reviews and signs: "Assumptions backed by the bank's own deals, ready for Thursday." Only then can the note be inserted into the credit memo, with every source as it stood that day.

The run in these screens finished in 16 seconds on the sample bank's loans, with a source on every sentence of the answer and the note. Time the same question on your own loans.

A medical office question, assumption ranges, the bank's cap rate against its appraisals, and every exception with its outcome

In shortOne loan is the story. These screens show the rest: the lens that recolors every pin, the ranked table, a second question about medical office, the full cited answer and its sources, the assumption ranges and comps table, a memo page, one loan's history, ranges by property type, the bank's cap rate against its appraisals by quarter, every policy exception with its outcome, the settings and the library.

Most screens open full size when selected. The bank, borrowers, people and figures are fictional.

Agents find and read the precedent; the analyst picks the cap rate, vacancy and expenses, and a credit officer signs the note

In shortEach agent has one job and limits it does not cross. None chooses an assumption, sizes a loan for approval or waives a policy. The analyst picks the cap rate, vacancy and expenses and recommends the amount; a named credit officer reviews and signs the note; committee approves the loan.

Six agents, from question reader to answer writer

The analyst picks the cap rate; a credit officer signs

Every question, answer, note and signature is on the audit trail, with the sources as they stood that day. Agent steps and human decisions are recorded apart, settings changes are versioned, and only approved memo versions are searched.

Priya Nandakumar's six loans in underwriting and her reviewers

From the core system and old credit memos to a ranked comps table, assumption ranges and a research note

In shortThe solution reads the records the bank already keeps and leaves them where they are. Out come a cited answer, a ranked comps table, assumption ranges by property type and market, a sizing of the loan in hand on the bank's own numbers, an exception history with outcomes, and a research note ready for the credit memo.

Core banking, approved memos and servicing data a comps search reads

For each, check whether it has an API, a file feed or only screens, and who owns it.

Your credit memo template, entitlements and policy, kept as they are

Every sentence sourced, thin evidence said aloud, people decide

What unsearched loan history costs you: analyst days, loose assumptions and exceptions granted blind

In shortStart from your own costs. No regulator or study publishes how long precedent research takes or how often a bank's assumptions drift from its own history, so the numbers that matter are yours. A few published figures frame them: what an analyst hour costs, how much CRE debt comes up for re-underwriting, and how sharply risk rises when coverage slips.

Timing precedent research and measuring cap rate drift on your loans

The repeated vacancy exception and the cap rate nobody can support

Analyst days back and losses avoided: the return on searching your own closed loans

In shortPut in your own numbers. The calculator starts from a published analyst wage and asks you for the rest: how many loans need precedent research, the hours they take now and after, what the tool and your testing cost a year, and, if you want, one loss you believe better assumptions would avoid. Every default marked illustrative is an assumption to replace.

Analyst hours, precedent on every loan, one set of inner targets

Your analyst hours back from searching your own closed loans

A bank that underwrites 60 new CRE loans a year and re-underwrites 60 maturities clears the 80-loan line in the chart. A bank that makes 30 CRE loans a year, all of one kind in one town, may do as well with a well-kept spreadsheet.

How searchable your loan history is: volume, approved memos, servicing data and exception records, scored

In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.

Explore, replay 20 to 30 closed loans, or search the whole book

How a loan comps tool decides what is comparable, and sixteen more things to ask its vendor

In shortAsk every vendor the same questions, us included. The answers show what the tool reads, how it decides what is comparable, what evidence you keep, who sees what, and how you would test it on your own closed loans.

We found no published independent test of any loan comps tool's accuracy, as of October 2026. A replay on your own closed loans is the fairest comparison, and the same test measures your current process.

Ten weeks on one property type in one market: replay closed loans, then search beside your analysts

In shortMeasure first, then replay a search on loans whose outcomes you know, then run it beside your analysts on live loans, then decide with four numbers. None of these steps commits you to a vendor.

1. Time three precedent searches on live loans, step by step, and count how many of your last 20 memos cite the bank's own comparable loans.
2. Pick one property type and one market where you have at least 20 closed loans, and gather their approved memos, appraisals, reviews and servicing.
3. Ask a senior officer to list the precedents they would cite for five past loans, before the tool runs.
4. Replay and run in parallel: the five past loans limited to what was known then, and four weeks of live loans with your analysts choosing every assumption.
5. Decide with four numbers: precedents found against the officer's list, figures checked against pages, analyst minutes, and memos that now cite the bank's own loans.

Precedents found, figures right and an honest count of zero

Cap rates above the appraisal, apartment vacancy, debt yield and exception reports: CRE loan comps questions

In shortThe questions CRE credit teams and buyers search for most, each answered on its own.