Core multifamily underwriting: a 3.8% levered IRR, below Treasuries

An acquisitions analyst at a core-plus fund asks Compound what to pay for a real 258-unit apartment community from the CMBS loan filings, starting at its 2019 appraised value, with rent growth checked against building permits and new debt priced on the Treasury yield of November 13, 2026.

Compound Research · Updated Nov 15, 2026 · 14 min read

Short answer

Core multifamily underwriting prices a stabilized apartment property from its in-place income, one rent growth rate and a loan. Compound's one-sheet model of a 258-unit, 2018 property near Tampa returns a 3.8% levered IRR at the $57.5 million appraised value, and the cap rate matches the loan rate at about $52.1 million.

Try it now
On this page (11)
  1. 1. The subject property: what the CMBS filing reports
  2. 2. The eight underwriting inputs and where each came from
  3. 3. Average rent from revenue and occupancy, checked against ZORI
  4. 4. Rent growth: ZORI, building permits, vacancy and concessions
  5. 5. The five-year projection and the exit
  6. 6. Sizing the loan at 55% loan-to-value against the loan in place
  7. 7. Going-in cap rate, levered IRR and the three flags
  8. 8. What you can pay: levered IRR at five prices
  9. 9. What the public data cannot tell you
  10. 10. Our view: the price this model supports
  11. 11. FAQ

At its appraised value of $57.5 million, a 258-unit apartment community built in 2018 returns a 3.8% levered IRR over a five-year hold. The 10-year Treasury yielded 5.15% on November 13, 2026. The same model puts the price at which the going-in cap rate equals the loan rate at about $52.1 million, and the levered IRR reaches 7.8% only at $46.0 million.

You are on the acquisitions team of a core-plus fund, a stabilized deal has come in, and the first question is what you can pay. We asked Compound to answer it on a real property before any seller had sent a file. It ran on November 15, 2026 and returned one Excel workbook and a short summary.

Compound picked Charleston On 66th in Largo, Florida, in the Tampa-St. Petersburg-Clearwater metro. The property secures loan 15 of GS Mortgage Securities Trust 2020-GC45, and the trust’s Form ABS-EE for the period ended October 9, 2026 reports 258 units built in 2018 and 95.74% occupied. Compound reported that it was the only screen result in the listed metros with every field filled. The request set the starting price at the latest appraised value, and the latest appraisal in the filing is $57.5 million, dated October 17, 2019. That price is an assumption of the exercise. Nothing here is a current value for the property, a forecast of what it will earn, or advice to buy or sell it.

What the workbook shows:

  • Borrowing lowers the return at the appraised value. The unlevered IRR is 5.3% and the levered IRR is 3.8%, with a 1.19x equity multiple. The new loan costs 6.65% and the property earns 6.02% on the price.
  • Two of the three flags trip. Negative leverage is one. The other is rent growth: the model uses 1.0% a year while the three-year ZORI average is -0.31% and permits rose from 7,545 units in 2024 to 8,771 in 2025.
  • The sale brings in less than the purchase. Year 6 net operating income of $3.64 million at a 6.52% exit cap rate is a $55.8 million sale.
  • The loan in place costs 3.43%. It is interest-only until November 1, 2029, with $1.30 million of annual debt service against $2.44 million on the new loan.

Each input on the Assumptions sheet names its source, or says “analyst assumption” and gives the reason. Property income, occupancy, the appraisal and the loan in place come from CMBS loan-level filings on SEC EDGAR. Metro asking rents come from Zillow’s ZORI index. Vacancy, concessions and median rents come from Apartment List. Permits come from the Census Bureau’s Building Permits Survey. The new loan is priced off the Daily Treasury Par Yield Curve.

Prompt to try

“I am pricing a stabilized apartment community for a core-plus fund and want a core multifamily underwriting model that answers one question, what can I pay, built on a real property. From the CMBS loan-level filings, pick one U.S. apartment property of roughly 200 to 400 units, built in 2015 or later, at least 93% occupied, in a Sun Belt metro such as Phoenix, Dallas, Austin, Nashville, Charlotte, Atlanta or Tampa, whose latest report gives units, occupancy, revenue, operating expenses, net operating income, appraised value and the loan terms. Use it as the subject and its latest appraised value as the starting price. Set the blended average rent from reported revenue and occupancy and check it against Zillow ZORI and Apartment List rent for the metro. Set the one rent growth rate from the last three years of ZORI for that metro, adjusted for new supply: show Census building permits for buildings of five or more units in the metro for each of the last five years next to Apartment List vacancy and concessions, and say in a note how they moved your growth rate. Set the expense ratio from the property's reported expenses over effective gross income. Size one amortizing loan at 55% loan-to-value, priced off today's 10-year Treasury yield plus a spread, and show it next to the loan the property has now. Build the screen itself as a one-sheet model: an inputs block of eight inputs (price, units, average rent, rent growth, expense ratio, loan-to-value, rate, and exit cap rate as the going-in rate plus a reversion spread); a five-year annual projection; a returns panel with going-in cap rate, average unlevered yield over the hold, levered IRR, equity multiple and the exit-minus-going-in cap rate spread in basis points; a row of five prices around the appraised value with the levered IRR at each; and a 2-by-2 sensitivity of levered IRR to rent growth and exit cap rate. Flag negative leverage, an exit cap rate tighter than the going-in rate, and a rent growth rate above the metro's three-year ZORI average while permits are rising. Use the stored public data wherever it has the figure, and web search only for what it lacks. Put every input on an Assumptions sheet with one column that says where it came from: a cited source, or "analyst assumption" for a figure I would set myself, with the reason for the value you chose. Every output must be a live formula that traces back to the Assumptions sheet. Add a Sources sheet listing each data source, what it gave and its date. Deliver one Excel workbook and a short summary of the result, the flags, and what I would need from the seller to firm it up. Cite every sourced figure.”

Try this prompt free →

That is the request as we sent it, without its closing line, which asks that no name or email be put in the workbook.

The subject property: what the CMBS filing reports

The filing gives one servicer-reported figure each for units, occupancy, revenue, expenses, net operating income, the appraisal and the loan. Compound copied each to the Assumptions sheet with the filing’s accession number beside it, and everything else in the model is built on these rows.

Item on the Assumptions sheetValueUnit
Units258units
Year built2018year
Physical occupancy, latest report95.74%%
Revenue (effective gross income), T12 7/1/2025-6/30/20266,648,047$
Operating expenses, T123,218,227$
Net operating income, T123,429,821$
Latest appraised value (MAI)57,500,000$
Appraisal date10-17-19date
Current balance37,375,000$
Interest rate (fixed)3.43%%
Interest-only term120months
Origination date11-01-19date
Maturity date11-01-29date
Debt service, T121,300,904$
DSCR on NOI, as reported2.64xx

Source: Compound’s Assumptions sheet, rows 7 to 24. Each row cites SEC Form ABS-EE, GS Mortgage Securities Trust 2020-GC45, loan 15, accession 0001888524-26-018792.

The financials cover the twelve months ended June 30, 2026. Revenue was $6.65 million, operating expenses $3.22 million and net operating income $3.43 million. Row 13 of the sheet subtracts expenses and net operating income from revenue and shows 0, so the three reported figures tie.

Expenses take 48.41% of revenue, which is $12,474 a unit a year (our arithmetic). The sheet’s source note for the appraisal reads “the latest appraisal on file is the 2019 securitization appraisal”, so the starting price is a value set when the loan was made.

The property itself is doing better than its metro. Reported occupancy of 95.74% leaves 4.3% of units vacant (our arithmetic), and Apartment List’s vacancy index for the Tampa metro was 8.5% in October 2026.

The eight underwriting inputs and where each came from

Five of the eight inputs come from the filing, from the request or from arithmetic on the two. Rent growth is an analyst assumption, and the interest rate and exit cap rate are each a sourced or computed figure plus a spread that Compound assumed and labeled.

Charleston_on_66th_Core_Underwriting.xlsx
C58fx=C10/(C7*C9*12)
BCDE
5ItemValueUnitSource / basis
55MODEL INPUTS (THE EIGHT THE MODEL READS)
56Purchase price$57,500,000$Request: latest appraised value as the starting price (row 14)
57Units258unitsReported units (row 7)
58Average monthly rent per occupied unit$2,242.85$/monthComputed: T12 revenue / (units x occupancy x 12); includes other income in reported revenue
59Rent growth, annual1.00%%Analyst assumption: 1.0%. Starts from the ZORI multifamily 3-yr average (row 31, -0.3%); +130 bps because 5+ unit permits are 34% below the 2022 peak and down 10% in Jan-Sep 2026, so deliveries thin over the hold; held to 1.0% because vacancy (8.5%) and concessions (52%) sit near their highs and 2025 permits rose 16%
60Expense ratio (operating expenses / effective gross income)48.41%%Computed: T12 operating expenses / T12 revenue (rows 11 and 10)
61Loan-to-value55.00%%Request
62Interest rate6.65%%Computed: 10-yr Treasury (row 66) + loan spread (row 67)
63Exit cap rate6.52%%Computed: going-in cap rate (row 71) + reversion spread (row 72), per the Request
65SUPPORTING ASSUMPTIONS
6610-yr Treasury par yield, 11/13/2026 close5.15%, as publishedU.S. Department of the Treasury, daily par yield curve, 10-yr, as of 11/13/2026 (Compound stored public data); latest close before this model
67Loan spread over 10-yr Treasury1.50%%Analyst assumption: 150 bps, a typical spread for a fixed-rate agency or conduit loan on a stabilized apartment property at 55% LTV
68Amortization30yearsAnalyst assumption: 30-year schedule, standard for agency multifamily debt
70Selling costs1.50%% of sale priceAnalyst assumption: 1.5% for broker, transfer and closing costs at exit
71Going-in cap rate (Year 1 NOI / price)6.02%%Computed: in-place NOI grown one year at the rent growth rate, over the purchase price
72Reversion spread (exit over going-in)0.50%%Analyst assumption: 50 bps; the asset ages from 8 to 13 years and the 10-yr Treasury (5.15%) leaves a thin cap rate spread today
ModelAssumptionsSources
From Compound's Assumptions sheet: the eight inputs the Model reads and six of the ten supporting assumptions, as Compound wrote them. The source column is wrapped here so each reason can be read in full. Blue is a sourced figure, blue on yellow an analyst assumption and black a formula.

Price and loan-to-value say “Request”. Units, average rent and the expense ratio are the filing’s figures or a formula on them. The Model sheet reads all eight with formulas such as =Assumptions!C56, so a change to one yellow cell reprices every output.

The three judgment calls are the cells to argue with:

  • Rent growth, 1.0%. The reasoning takes a section of its own below.
  • Loan spread, 150 basis points. The sheet calls it “a typical spread for a fixed-rate agency or conduit loan on a stabilized apartment property at 55% LTV”. Added to the 5.15% Treasury yield it gives 6.65%. A lender’s quote replaces this cell.
  • Reversion spread, 50 basis points. The exit cap rate is the going-in rate plus this spread. Compound’s reason is that “the asset ages from 8 to 13 years and the 10-yr Treasury (5.15%) leaves a thin cap rate spread today”. The going-in cap rate of 6.02% sits 87 basis points above that Treasury yield (our arithmetic).

Amortization of 30 years and selling costs of 1.5% are also labeled analyst assumptions.

Average rent from revenue and occupancy, checked against ZORI

The filing reports revenue and occupancy and no rent, so the model backs the rent out: $2,243 a month for each occupied unit. That figure is 26.4% above Zillow’s multifamily rent index for the metro and 52.8% above Apartment List’s metro median, and some of the gap is other income.

Charleston_on_66th_Core_Underwriting.xlsx
E77fx=E76/D76-1
BCDEFGHI
74MARKET AND SUPPLY: TAMPA MSA
75202120222023202420252026 YTD
765+ unit permits (units)5,43413,2929,9717,5458,7716,357
77Change vs prior year144.6%(25.0%)(24.3%)16.2%(9.6%)
78Apartment List vacancy (Dec; Oct for 2026)3.4%5.8%6.7%6.4%8.0%8.5%
79Apartment List concession rate (Dec; Oct for 2026)49.0%51.9%
81Rent check (property vs metro)Rent ($/month)Premium
82Property, average rent per occupied unit$2,243
83ZORI multifamily, Sep 2026$1,77426.4%
84Apartment List median, Tampa metro, Oct 2026$1,46852.8%
85Apartment List median, Largo, Oct 2026$1,39460.9%
ModelAssumptionsSources
The market and supply block of Compound's Model sheet, rows 74 to 85, unedited: five years of permits, vacancy and concessions, then the rent check. Negative values are in parentheses. The 2026 permit change compares January to September with the same months of 2025.

The formula is revenue divided by units, occupancy and twelve months. Compound wrote a caveat on both sheets: reported revenue includes other income, so $2,243 is what an occupied unit brings in each month, and base rent is somewhat lower.

Monthly rent: the property against three market gauges

Dollars per month. The property figure is trailing revenue per occupied unit and includes other income

Charleston On 66th, revenue per occupied unit$2,243Zillow ZORI, multifamily, Tampa metro, Sep 2026$1,774Apartment List median, Tampa metro, Oct 2026$1,468Apartment List median, Largo, Oct 2026$1,394

Compound's Model sheet, rows 82 to 85: SEC Form ABS-EE asset data for the twelve months ended June 30, 2026; Zillow ZORI, September 2026; Apartment List Rent Estimates, October 2026.

ZORI tracks asking rents across apartment buildings of every age in the metro, and Apartment List’s medians of $1,468 for the metro and $1,394 for Largo cover every listed property. This building opened in 2018. What the public data cannot say is how much of the 26.4% is newer units and how much is fees and other charges. The seller’s rent roll answers that in one page, and it is the first item on Compound’s list of what to ask for.

Rent growth: ZORI, building permits, vacancy and concessions

Compound set rent growth at 1.0% a year, 130 basis points above the three-year ZORI average of -0.31%, and wrote its reasons on the sheet. The model then flags that choice, because the request said to flag any growth rate above the average while permits are rising.

Start with the trailing record. Zillow’s multifamily index for the Tampa metro was $1,791 in September 2023, $1,768 in 2024, $1,784 in 2025 and $1,774 in 2026. The three yearly changes are -1.2%, 0.9% and -0.5% (our arithmetic), and the sheet averages them to -0.31%. Two other gauges on the Assumptions sheet point the same way. Apartment List’s metro rent fell 1.73% in the year to October 2026, and the ZORI index that adds single-family rentals rose 0.79% a year over the three years. The 1.0% in the model is above all three.

The request asked for the supply check that would justify moving off the average.

Apartment units permitted in the Tampa metro

Units authorized in buildings with five or more units, by calendar year, with the first nine months of 2025 and 2026

20215,434202213,29220239,97120247,54520258,771Jan-Sep 20257,029Jan-Sep 20266,357

Compound's Assumptions sheet, rows 38 to 44: U.S. Census Bureau Building Permits Survey, Tampa-St. Petersburg-Clearwater metro.

Permits peaked at 13,292 units in 2022, fell to 7,545 in 2024 and rose 16.2% to 8,771 in 2025. In the first nine months of 2026 the count was 6,357 against 7,029 a year earlier, down 9.6%.

Apartment vacancy in the Tampa metro

Apartment List vacancy index at each December and at October 2026, with the subject property for comparison

Dec 20213.4%Dec 20225.8%Dec 20236.7%Dec 20246.4%Dec 20258.1%Oct 20268.5%Charleston On 66th, 1 lessreported occupancy (our arithmetic)4.3%

Compound's Model sheet, row 78: Apartment List vacancy index, Tampa-St. Petersburg-Clearwater metro. Property occupancy: SEC Form ABS-EE asset data.

Vacancy rose from 3.4% in December 2021 to 8.5% in October 2026. Apartment List’s concession rate, the share of properties offering at least a month free, went from 28.3% in January 2025, when the series starts, to 51.9%. Compound’s note under the block reads:

Note on rent growth: the ZORI multifamily 3-yr average for Tampa is -0.3% a year. 5+ unit permits fell from 13,292 units in 2022 to 7,545 in 2024, rose 16% to 8,771 in 2025, and are down 10% in Jan-Sep 2026; the thinner pipeline argues for some recovery over the hold, so growth is set 130 bps above ZORI at 1.0%. Vacancy (8.5%) and concessions (52% of properties offering a month free) sit near their highs and 2025 permits rose, which holds the rate at 1.0% rather than a typical 3% and trips the supply flag.

We read the same rows more cautiously. The lower permit counts of 2024 and 2026 help the later years of a five-year hold, which supports rents that stop falling. It is thinner support for rents that grow from Year 1, and the sensitivity further down shows what a flat-rent case does to the return.

The five-year projection and the exit

One growth rate drives the projection: rent grows 1.0% a year, occupancy stays at the reported 95.74% and expenses stay at 48.41% of income. Net operating income rises from $3.46 million in Year 1 to $3.60 million in Year 5, and the sale is priced on Year 6.

Charleston_on_66th_Core_Underwriting.xlsx
H27fx=I21/$C$13
BCDEFGHI
15FIVE-YEAR PROJECTION
16Year 0Year 1Year 2Year 3Year 4Year 5Year 6
17Average monthly rent$2,243$2,265$2,288$2,311$2,334$2,357$2,381
18Occupancy95.7%95.7%95.7%95.7%95.7%95.7%95.7%
19Effective gross income ($)6,648,0476,714,5286,781,6736,849,4906,917,9856,987,1657,057,036
20Operating expenses(3,218,227)(3,250,409)(3,282,913)(3,315,742)(3,348,899)(3,382,388)(3,416,212)
21Net operating income ($)3,429,8213,464,1193,498,7603,533,7483,569,0853,604,7763,640,824
22Debt service(2,436,257)(2,436,257)(2,436,257)(2,436,257)(2,436,257)
23Cash flow after debt service ($)1,027,8621,062,5041,097,4911,132,8291,168,520
24DSCR (x)1.42x1.44x1.45x1.46x1.48x
25Loan balance, end of year31,625,00031,281,46030,914,36630,522,10230,102,94229,655,043
27Gross sale price (Year 6 NOI / exit cap)55,801,875
28Selling costs(837,028)
29Loan repayment(29,655,043)
30Unlevered cash flow ($)(57,500,000)3,464,1193,498,7603,533,7483,569,08558,569,623
31Levered cash flow ($)(25,875,000)1,027,8621,062,5041,097,4911,132,82926,478,323
ModelAssumptionsSources
The five-year projection on Compound's Model sheet, rows 15 to 31, unedited. Year 0 is the reported twelve months ended June 30, 2026, and Year 6 is the forward income the exit cap rate is applied to. Green cells read the Assumptions sheet.

Income is units times rent times twelve times occupancy, and expenses are a fixed share of it, so net operating income grows at the rent growth rate and nothing else. Expenses therefore grow 1.0% a year as well, and do not grow at all in the flat-rent case of the sensitivity. If expenses rose faster than rent, net operating income would be lower in every year than this projection shows.

The exit is the row to read closely. Year 6 income of $3.64 million divided by the 6.52% exit cap rate gives a gross sale price of $55.8 million. That is $1.7 million below the $57.5 million purchase price (our arithmetic). Selling costs take another $837,028, and the loan balance at sale is $29.7 million.

The levered row shows what that does to the equity. The buyer puts in $25.9 million. Cash flow after debt service adds up to $5.5 million over the five years, and the sale returns $25.3 million after costs and the loan (both our arithmetic). The sale hands back slightly less than the equity that went in, so all of the profit in the 1.19x multiple is the cash collected along the way.

Sizing the loan at 55% loan-to-value against the loan in place

The new loan is $31.6 million at 6.65%, amortizing over 30 years, and Year 1 income covers its payments 1.42 times. The loan the property has now is $37.4 million at 3.43%, interest-only, and costs $1.30 million a year against $2.44 million for the new one.

Charleston_on_66th_Core_Underwriting.xlsx
C38fx=PMT(C36/12,C37*12,-C35)*12
BCD
33FINANCING: PROPOSED LOAN vs LOAN IN PLACE
34ProposedIn place
35Loan amount ($)31,625,00037,375,000
36Interest rate6.65%3.43%
37Amortization (years)30IO to 11/1/2029
38Annual debt service ($)2,436,2571,300,904
39DSCR on Year 1 NOI (x)1.42x2.66x
40Debt yield on Year 1 NOI10.95%9.27%
41LTV at purchase price55.00%65.00%
42Debt constant (debt service / loan)7.70%3.48%
ModelAssumptionsSources
The financing block of Compound's Model sheet, rows 33 to 42, unedited. The in-place column is the loan reported in the CMBS filing, measured on the same Year 1 income as the proposed loan.

The new loan is $5.75 million smaller than the balance in place and costs $1.14 million more a year (both our arithmetic). Its debt constant, payments over loan amount, is 7.70% against 3.48%. Coverage falls from 2.66 times to 1.42 times on the same income. The filing reports 2.64 times on trailing income.

The debt constant explains the first-year cash return. Payments on the loan run 7.70% of its balance a year and the property earns 6.02% on the price, a gap of 168 basis points (our arithmetic). Year 1 cash-on-cash is 4.0%. Amortization does build equity: the loan balance falls by $1.97 million over the hold (our arithmetic).

The loan in place matures on November 1, 2029, and the Model’s note says its prepayment premium runs through April 30, 2029. Whether a buyer could assume a 3.43% loan with three years left is a question for the seller, and Compound’s summary lists the payoff and assumption terms among the items to request.

Going-in cap rate, levered IRR and the three flags

At $57.5 million the property yields 6.02% going in and an average of 6.15% over the hold, and the unlevered IRR is 5.3%. Adding the loan cuts the IRR to 3.8%, and two of the three tests the request asked for read FLAG.

Charleston_on_66th_Core_Underwriting.xlsx
F58fx=IF(AND(C58>D58,C59>D59),"FLAG","OK")
BCDEF
44RETURNS
45Going-in cap rate (Year 1 NOI / price)6.02%
46Average unlevered yield, Years 1-56.15%
47Unlevered IRR5.3%
48Levered IRR3.8%
49Equity multiple (x)1.19x
50Exit cap minus going-in cap (bps)50
51Year 1 cash-on-cash4.0%
52Price at which the cap rate equals the loan rate ($)$52,092,016
54FLAGS
55ValueTest againstDifferenceStatus
56Negative leverage: interest rate vs going-in cap6.65%6.02%0.63%FLAG
57Exit cap tighter than going-in cap6.52%6.02%0.50%OK
58Rent growth vs ZORI 3-yr average (flag if above while permits rise)1.00%(0.31%)1.31%FLAG
595+ unit permits, 2025 vs 20248,7717,5451,226Rising
ModelAssumptionsSources
The returns panel and the flags on Compound's Model sheet, rows 44 to 59, unedited. The two outlined cells are the flags that trip. Row 59 feeds the permit half of the third flag.

The unlevered IRR of 5.3% is below the 6.02% going-in yield because the sale price is below the purchase price. The levered IRR of 3.8% is 132 basis points under the 10-year Treasury yield the loan is priced from (our arithmetic).

The three flags work as follows.

  • Negative leverage: FLAG. The test is the 6.65% interest rate against the 6.02% going-in cap rate, a difference of 63 basis points.
  • Exit cap rate tighter than going-in: OK. The exit rate of 6.52% is 50 basis points wider, by construction.
  • Rent growth above the ZORI average while permits rise: FLAG. The formula is =IF(AND(C58>D58,C59>D59),"FLAG","OK"). Growth of 1.0% is above -0.31%, and 2025 permits exceed 2024 permits by 1,226 units. The test uses the last two full years.

Compound added one row the request did not ask for. Row 52, the price at which the cap rate equals the loan rate, divides Year 1 income by the loan rate: $52,092,016. That is 90.6% of the appraised value, or $201,907 a unit against $222,868 at the appraisal (our arithmetic). At that price the going-in cap rate equals the 6.65% interest rate, and above it the cap rate is the lower of the two. This is a first-year test on the interest rate alone. The loan’s payments include principal, so its debt constant is 7.70%, and the price at which the levered IRR catches up with the unlevered IRR is lower. The price row in the next section shows where.

What you can pay: levered IRR at five prices

The model answers the question with a row of five prices from 80% to 120% of the appraised value. The levered IRR is 7.8% at $46.0 million, 3.8% at $57.5 million and 0.9% at $69.0 million.

Charleston_on_66th_Core_Underwriting.xlsx
D65fx=D64+Assumptions!$C$72
BCDEFGH
61LEVERED IRR AT FIVE PRICES
62% of appraised value80%90%100%110%120%
63Purchase price ($)$46,000,000$51,750,000$57,500,000$63,250,000$69,000,000
64Going-in cap rate7.53%6.69%6.02%5.48%5.02%
65Exit cap rate8.03%7.19%6.52%5.98%5.52%
66Levered IRR7.8%5.6%3.8%2.3%0.9%
68SENSITIVITY: LEVERED IRR
69Rent growth (rows) by exit cap rate (columns)
706.02%7.02%
710.00%4.2%(2.2%)
722.00%9.5%3.7%
ModelAssumptionsSources
The five-price row and the two-by-two sensitivity on Compound's Model sheet, rows 61 to 72, unedited. Each IRR is computed from a row of cash flows further down the sheet, in rows 93 to 101. Negative values are in parentheses.

Levered IRR at five purchase prices

Five-year hold, 55% loan-to-value at 6.65%. The exit cap rate at each price is that price's going-in cap rate plus 50 basis points

$46,000,000 (80% of appraised value)7.8%$51,750,000 (90% of appraised value)5.7%$57,500,000 (100% of appraised value)3.8%$63,250,000 (110% of appraised value)2.3%$69,000,000 (120% of appraised value)0.9%10-year Treasury par yield, Nov 13, 20265.1%

Compound's Model sheet, rows 62 to 66. Treasury: Daily Treasury Par Yield Curve, 10-year, close of November 13, 2026.

Two prices on the row clear the Treasury yield. At $51.75 million the going-in cap rate is 6.69%, just above the 6.65% loan rate, and the levered IRR is 5.6%. At $46.0 million it is 7.8%. The unlevered IRR at those two prices is 6.2% and 7.2% (our arithmetic: the same projection and exit rule with no loan). So at $51.75 million the loan still lowers the return, and at $46.0 million it raises it. In IRR terms the loan turns from a cost to a help somewhere between those two prices, below the $52.1 million on row 52. The step from 90% to 100% of the appraised value costs 1.8 points of IRR (our arithmetic).

One design choice shapes the row. The request defined the exit cap rate as the going-in rate plus the reversion spread, so the model moves the exit rate with each price. At $46.0 million the going-in rate is 7.53% and the exit rate 8.03%, which sells the property for about $45.3 million (our arithmetic). If you think the exit cap rate is set by the market and has nothing to do with what you paid, hold it at 6.52% in every column. The low prices then look better than this row shows and the high prices worse.

The two-by-two sensitivity tests the two inputs that carry the most judgment, at the appraised value:

Rent growthExit cap rate 6.02%Exit cap rate 7.02%
0.00%4.2%(2.2%)
2.00%9.5%3.7%

Source: Compound’s Model sheet, rows 70 to 72. Rent growth is 1.0% either side of the base case and the exit cap rate 50 basis points either side.

With flat rents and an exit cap rate of 7.02% the levered IRR is negative, at -2.2%. With 2% growth and an exit at the going-in rate it is 9.5%. Moving the exit cap rate across its two columns changes the result by about six points in either row, and moving rent growth across its two rows changes it by five to six (our arithmetic on the four cells).

Compound’s summary puts the answer in three sentences: at the appraised value of $57.5M ($223k a unit) the levered IRR is 3.8%, leverage stops hurting at about $52.1M, and a levered return near 8% needs a price near $46M, or $178k a unit. The second sentence is the cap rate test on row 52. Measured by IRR, the turning point is lower, as the price row shows.

What the public data cannot tell you

A CMBS filing gives one line each for revenue, expenses and net operating income, and the market indexes describe a metro. Firming up the price takes the seller’s files, and Compound’s summary lists them.

  • Rent roll and trailing operating statement. The filing has total revenue only. The rent roll splits base rent from other income, concessions and loss-to-lease, and tells you whether $2,243 is a rent or a rent plus fees.
  • Monthly occupancy history. The filing gives one occupancy figure, 95.74%, and the model holds it for five years. The history shows whether that figure is steady in a metro where vacancy has risen.
  • A current value. The appraisal is dated October 17, 2019. The request asked for no sale comparables or broker cap rates, and the price is the request’s assumption. A current appraisal or broker opinion replaces it.
  • Taxes and insurance. The 48.41% expense ratio is one reported total. Compound’s summary asks for the 2026 tax bill, an estimate of the bill once a sale triggers reassessment, and a current insurance quote, and notes that the ratio does not include a reassessment.
  • Condition and capital spending. The filing reports no capital spending, and a screen with eight inputs carries no reserve. A property condition report and the capital history set one.
  • Unit mix and square footage. The filing gives a unit count. Rent per square foot by floor plan lets you compare this building with its direct competitors.
  • Debt. The filing gives the balance, rate and maturity of the loan in place. The payoff cost and any right to assume it come from the loan documents, and a lender’s term sheet replaces the assumed spread on the new loan.
  • The submarket. ZORI, the Apartment List series and the permit counts all describe the whole metro. Largo’s own pipeline is a question for a market study.

Our view: the price this model supports

On these inputs the model does not support the 2019 appraised value as a price for a buyer using 55% leverage at today’s rates. The range worth testing with the seller’s numbers starts near the $52.1 million where the cap rate equals the loan rate and runs down toward the 90% and 80% prices on the row. Only at the lower of those does the loan add to the IRR.

Three things in the workbook lead us there. The loan costs more than the property earns. The sale price is below the purchase price even with rent growing. And the growth rate that produces those figures is already above every trailing measure of Tampa rents on the Assumptions sheet. A flat-rent case is the one we would treat as the base, and the sensitivity puts it at 4.2% with the exit at the going-in rate and at -2.2% with the exit 100 basis points wider.

The other side deserves a fair hearing. The property is 95.74% occupied in a metro where the vacancy index is 8.5%. Permits in the first nine months of 2026 are below a year earlier, which helps the later years of the hold. A buyer who uses little or no debt sees the 5.3% unlevered return and none of the negative leverage. And the 3.43% loan in place, if it can be assumed, would change the levered figures for its remaining three years.

Three things would change our view. A rent roll showing base rent close to $2,243 with little other income would make the 26.4% premium to the metro index a fact about the building. A lender quote inside the 150 basis point spread would move that $52.1 million figure up. A tax bill after reassessment would move net operating income, most likely down.

All of this is a reading of a model built on a 2019 appraisal and public data. It makes no claim about what the property is worth today.

A price that survives this screen goes to the full multifamily acquisition pro forma, with a unit mix, line-item expenses and the tax reassessment. A property that needs renovation to earn its price belongs in the value-add multifamily screen. The metro question on its own is a real estate market analysis, and once the seller’s rent roll arrives, the rent roll analysis replaces the average rent in this model.

Core multifamily underwriting model (Excel)Compound’s workbook from this run: the one-sheet Model with the inputs, five-year projection, loan comparison, returns, flags, five-price row and sensitivity; an Assumptions sheet with a source or reason beside every input; and a Sources sheet.Download · Free

FAQ

What is the difference between core and core-plus multifamily underwriting?

Both price a stabilized property on the income it already earns. Core underwriting assumes no renovation and low leverage, so the return is mostly the going-in yield. Core-plus allows somewhat more debt. The model on this page is the core case with a core-plus loan: one rent growth rate, occupancy held at 95.74% and 55% loan-to-value.

How do you set rent growth when underwriting a stabilized apartment property?

Start from the trailing change in a metro rent index, then adjust for supply and write the adjustment down. In this workbook the three-year ZORI average is -0.31%, and Compound moved it to 1.0% after reading five years of permits, vacancy and concessions. The reasoning sits in the source column beside the input.

What is negative leverage in a multifamily acquisition?

Negative leverage means the loan’s interest rate is higher than the going-in cap rate, so borrowing lowers the return on equity. Here the loan costs 6.65% and the property yields 6.02% at the appraised value. The workbook computes the price at which the two are equal, $52,092,016, as Year 1 net operating income divided by the loan rate. That is a first-year test. Over the hold, compare the levered and unlevered IRR as well: at $51.75 million in this model the cap rate is above the loan rate and the levered IRR is still the lower of the two.

What exit cap rate should a five-year multifamily model use?

A common convention is the going-in cap rate plus a reversion spread for the building’s age and for rate risk. This model adds 50 basis points, giving 6.52%, and tests 50 basis points either side. The levered IRR at flat rents moves from 4.2% to -2.2% across those two exit rates, so the spread deserves as much attention as rent growth.

How do you work out the most you can pay for a target IRR?

Lay out a row of prices and compute the levered IRR at each, then read across to your target. In this workbook the row runs from 80% to 120% of the appraised value: 7.8% at $46.0 million, 5.6% at $51.75 million and 3.8% at $57.5 million.

Can you underwrite an apartment property from CMBS filings before the seller sends anything?

Yes, for a first price. A CMBS asset data file reports units, occupancy, revenue, expenses, net operating income, the appraised value and the loan terms, and Compound cites each figure to the filing’s accession number. The seller’s rent roll, operating statement and tax bill then replace the single reported totals.

CoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix ResearchCoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix ResearchCoatueCentri ConsultingAvra783 Capital PartnersOctahedron CapitalArix Research

See this workflow on your own deal

Try it now