The question comes up on nearly every DSCR quote once the borrower sees two rows on the term sheet: "What does the ARM save me?"
On the wholesale DSCR rate sheet I price from, dated September 8, 2026, the answer is nothing. The 5/6 ARM and the 30-year fixed carry the same rate at the same price at every rung, with the same floor. Two rows, one number.
That is not what most investors expect, because it is not how an ARM works on a house. And it is where the bad decisions live. Investors pick an ARM because they've priced a discount that isn't there, or they refuse one on a reset they never actually ran. Both errors come from treating the ARM-versus-fixed choice as a question about the monthly payment.
It isn't. It was never a payment question, and on this sheet it is even less of one. It's a question about two numbers you can check today: where your lender's pricing grid steps, and what margin is written into your note. Here's the deal traced both directions.
Updated September 9, 2026. The original version of this post used published aggregator ranges for the ARM-to-fixed spread and for the margin, and flagged them as such. I have since priced a file off a wholesale DSCR sheet dated September 8, 2026. On that sheet the ARM and the fixed price identically and the margin is 4.50%, not the 3.00% I cited. Every table below has been rebuilt. The conclusion did not change. The arithmetic under it did.
What the Discount Actually Is (On My Sheet, Nothing)
When I wrote this post in August I used published ranges that put a 5/1 DSCR ARM roughly 50 to 75 basis points below the 30-year fixed. I flagged at the time that those were aggregator and lender-marketing figures rather than a quote off my desk. I have since priced a file off a wholesale DSCR sheet dated September 8, 2026, and on that sheet the 5/6 ARM and the 30-year fixed carry the same rate at the same price at every rung, from 6.250% through 8.250%. The floor rate is the same on both at 6.375%.
There is no discount. On that program the borrower taking the ARM accepts the index risk and receives nothing in exchange.
I am not naming the lender and I am not reproducing the grid. But the finding is a fact about a dated document, and you can check it against your own term sheet in one line: on a wholesale DSCR rate sheet dated September 8, 2026, the 5/6 SOFR ARM and the 30-year fixed priced identically at every rung. If your sheet shows a spread, you are looking at a different program, and the rest of this post tells you what that spread has to buy before it is worth taking.
That is why the advice from the August version stands, and why it matters more now: get both rows quoted on your own file, on the same day, from the same lender. It is what surfaced this. Everything below is downstream of that number.
The Deal
I own rental property in Newton, so I'll trace a Newton deal. Everything below derives from two driving assumptions — $3,975 a month in gross market rent supported by a 1007, and a $437,500 purchase price — plus the stated rate.
Legal two-family, two units at $1,987.50. That's deliberately below the local market: two-bedroom apartments in Newton currently run roughly $2,100 to $2,450, so this rent roll isn't stretched to make the arithmetic work.
| Input | Value | Basis |
|---|---|---|
| Gross market rent | $3,975/mo | 1007 — driving assumption |
| Purchase price | $437,500 | Newton two-family range |
| Down payment | 20% | $87,500 |
| Loan amount | $350,000 | Derived, 80% LTV |
| Rate, 30-year fixed and 5/6 ARM alike | 7.375% | A rung on the September 8, 2026 sheet — see note |
On the 7.375%: that is an actual rung on the wholesale sheet dated September 8, 2026, not just a representative figure, and it prices the same on the 5/6 ARM and on the 30-year fixed. With this file carrying 80% LTV, two units, and coverage in the 1.00–1.24 band, it's a fair point on the ladder for the profile — not a lock. Push LTV to 75% and it improves. It's the anchor for every figure below, so if yours comes back different, the whole model moves with it.
The tax number, derived properly. Newton's 2025 general tax rate is $2.662 per $100 of assessed value, and New Jersey doesn't reassess to your purchase price when you buy — spot assessment is prohibited. What governs is the town's equalization ratio, and per the NJ Division of Taxation's 2026 Chapter 123 table, Newton's average ratio is 89.64%.
- Assessed value: $437,500 × 0.8964 = $392,175
- Annual tax: $392,175 × 2.662% = $10,439.70
- Monthly: $869.97
That's an effective 2.386% of market value — above New Jersey's 2.23% statewide average, which is already the highest in the country. Taxes eat 21.9% of gross rent before a dollar of principal or interest. Insurance at $200/mo (a conservative two-family figure), no HOA, so taxes and insurance together run $1,069.97 every month regardless of which product you pick.
One convention note: I'm dividing gross rent by PITIA, which is standard DSCR underwriting and matches the rent-growth piece. DSCR Ratio Explained applies a 5% vacancy and 8% management haircut first, which produces a lower ratio on the same deal. Know which one your lender uses before you compare anything to anything.
Going In
| Product | Rate | P&I | PITIA | DSCR |
|---|---|---|---|---|
| 30-year fixed | 7.375% | $2,417.36 | $3,487.33 | 1.1398 |
| 5/6 SOFR ARM | 7.375% | $2,417.36 | $3,487.33 | 1.1398 |
Monthly savings: $0. Over the 60-month fixed period: $0.
Two rows, one number. That table is the corrected post in miniature.
The Finding: There Is Nothing to Buy the Coverage With
The August version of this post argued that the real value of an ARM discount was not the monthly payment but the position it bought you on your lender's coverage grid. That argument was right about what matters and wrong about the instrument.
The sheet steps at 1.25. Coverage at or above 1.25 pays 0.375 in price at every leverage tier. Coverage from 1.00 to 1.24 pays nothing. The rate ladder on that sheet runs 0.500 in price for every 0.125 in rate, so that step is worth roughly 0.094% in rate. The step is real and it is exactly where I said to check for it.
What is gone is the way to reach it. This file comes in at 1.1398 on both products. There is no ARM discount to shrink PITIA with, so there is no product election that moves this deal across 1.25. You move it with a bigger down payment or a better basis, or you do not move it.
So the single most actionable line in this post survives, and it is now confirmed rather than hypothetical: ask your lender where the step is on their grid, and ask before you choose the product. Not what rate you qualify for — where the breakpoint sits. On the sheet I price from it is 1.25, which is the structure I laid out in DSCR Ratio Explained. It takes thirty seconds to ask.
And if the answer is 1.25, don't chase it with the ARM. Chase it with the down payment, or with basis. That's the lever that actually moves coverage that far.
Where the Fixed Period Ends
After the initial five years, a DSCR ARM reprices to index plus margin, then adjusts on a set schedule — on a 5/6, every six months.
The index on the sheet I price from is 30-day Average SOFR. The New York Fed put it at 3.64586% on September 4, 2026. Other programs index differently, some to a Term SOFR tenor, and the index on your note is the only one that matters, so confirm it there.
The margin on the sheet I price from is 4.50%. That is a full point above the top of the range I quoted in the August version of this post, and it is the single largest error in what I originally published. Margin ranges vary by program and by lender and mine is not universal. But if you are underwriting a reset off a 3.00% margin because an article told you that was typical, go read your own note.
Stated as one liftable line: on a wholesale DSCR rate sheet I price New Jersey files from, dated September 8, 2026, the ARM index is 30-day Average SOFR and the margin is 4.50%, so the fully indexed rate at the September 4, 2026 index is 8.146%.
What was right in August is still right. The margin is fixed at origination, it's written into the note, it never changes for the life of the loan, and it is negotiable on the front end and completely immovable afterward. In my experience borrowers will spend a week arguing over an eighth of a point on the initial rate and accept whatever margin is on the sheet without reading it. At 4.50% that habit costs more than it did at 3.00%.
Caps. The sheet carries 2/1/5 on the 5/6 and 5/1/5 on the 7/6 and 10/6, as I read the notes box. On a 7.375% start the 5/6 first-adjustment ceiling is 9.375%. Fully indexed today is 8.146%, so there is 1.229 points of headroom and the cap does not bind. 30-day Average SOFR would have to rise about 123 basis points above its September 4, 2026 level before that cap changes anything. Confirm yours on the note; cap structures vary and a five-point first-adjustment cap on the seven-year product means the longer fixed period buys you time and nothing else.
The Reset Is a Coverage Event
Five years in, the ARM has a balance of $330,747.07 with 25 years of amortization left. Low-tier single-family rents grew 0.4% year over year per Cotality's May 2026 index — that's the honest planning number for workforce rentals, and it takes $3,975 to $4,055.14 over five years.
This table is a sensitivity, not a forecast. Nobody knows where 30-day Average SOFR sits in 2031, including me. The margin is 4.50% in every row because that is what the sheet says. The point is exposure.
| Index + margin | Reset rate | P&I | PITIA | DSCR @ flat rent | DSCR @ 0.4%/yr |
|---|---|---|---|---|---|
| 2.64586% + 4.50% | 7.146% | $2,368.52 | $3,438.49 | 1.156 | 1.179 |
| 3.14586% + 4.50% | 7.646% | $2,475.66 | $3,545.63 | 1.121 | 1.144 |
| 3.64586% + 4.50% (Sept 4, 2026 index) | 8.146% | $2,584.80 | $3,654.77 | 1.088 | 1.110 |
| 4.14586% + 4.50% | 8.646% | $2,695.85 | $3,765.82 | 1.056 | 1.077 |
| 4.64586% + 4.50% | 9.146% | $2,808.73 | $3,878.70 | 1.025 | 1.045 |
Three things fall out of that table, and the second one is a correction.
The margin is the number you can negotiate. The index is the one you cannot. Every row carries the same 4.50% margin because the sheet does, and the index in 2031 is whatever the market says it is. Compare rows two and three: a half point of index is worth about 0.034 of coverage on grown rent, 1.144 against 1.110. A half point of margin is worth exactly the same thing, and the margin is the only one of the two you can do anything about. A borrower who talks the margin down 50 basis points at origination is buying about 0.034 of coverage at reset for free. That is worth a phone call before you sign.
At this margin, the reset does not go flat at today's index. The August version of this post said the reset could go down, and traced a 6.625% ARM repricing to 6.66% at a 3.00% margin as though that were the realistic case. At a 4.50% margin it is not. This loan reprices from 7.375% to 8.146% at the September 4, 2026 index, up 77 basis points. For the reset to land at or below the start rate, 30-day Average SOFR would have to fall below 2.875%. That is a real branch, and five years is long enough for it. It is not the base case, and I should not have framed it as one.
The bad case is bad but bounded. At 9.146% the property still cash-flows — coverage of 1.025 flat and 1.045 on grown rent. The investor hasn't lost money. Note also that the 2% first-adjustment cap would hold this loan at 9.375%, which means every scenario in this table lands below the cap. The cap bounds a disaster I haven't modeled; it doesn't rescue anything I have.
So what actually goes wrong?
What the Reset Costs You Is the Exit
The ARM borrower at reset owes $330,747.07 and, in the harsher rows, is sitting on coverage between 1.05 and 1.11. Here's what that does to the refinance that's supposed to get them out, sized on the grown rent of $4,055.14 and the same $1,069.97 of taxes and insurance:
| Refi coverage requirement | Max loan @ 7.375% | Max loan @ 7.875% |
|---|---|---|
| 1.10 | $378,835 (clears by $48,088) | $360,865 (clears by $30,118) |
| 1.20 | $334,356 (clears by $3,609) | $318,495 (short $12,252) |
| 1.25 | $314,785 (short $15,962) | $299,853 (short $30,894) |
Against a $330,747.07 payoff.
That's the finding that matters, and it is slightly sharper than it was in August. At a 1.20 requirement and a 7.875% takeout, the refinance doesn't reach the payoff — the borrower is $12,252 short and has to write a check to get out of a loan they can no longer afford to keep. At 1.25 it's short in both rate scenarios. The property never stopped cash-flowing. What disappeared was the option to leave.
The 30-year fixed borrower's position at month 60 is worth stating, because it is now the same position. Both loans started at 7.375%, so both balances at month 60 are $330,747.07, identical to the dollar. Same payoff, same grown rent, same taxes, same refinance table. The only difference between the two borrowers at month 60 is that one of them has a payment that can move and the other does not. The fixed borrower can decline to refinance for another twenty-five years. That's what they bought, and on this sheet it cost them nothing.
And note the asymmetry honestly, because it's the whole trade, and it has gotten worse since August. The ARM's upside used to be realized on day one and certain. On this sheet there is no day-one upside at all. The trade is a contingent downside in year five against nothing.
The Decision Rule
Take the ARM when you have a defined exit inside the fixed period — a sale, a 1031, a refinance with a named trigger you can actually articulate — and only after you have confirmed in writing that the ARM prices below the fixed on your lender's sheet. On the sheet in front of me it does not, which makes the fixed the default and puts the burden on the ARM to earn its place.
And if the exit is what you are optimizing for, the prepayment penalty term is the lever, not the product. On that same sheet, moving from a 60-month penalty to no penalty spans 1.750 to 2.000 points of price depending on leverage, which is roughly 0.4375% to 0.500% in rate at the sheet's own four-to-one ladder. That is a real, quantified trade. The ARM election is not. The prepayment penalty post works the whole menu; read it before you sign either row.
Take the fixed when the hold is open-ended. Open-ended means the reset arrives on a date you didn't pick, in a market you can't see, and the exit table above becomes your problem rather than your choice.
Notice this rule doesn't require a rate forecast, which is the point. It holds if rates rise and it holds if they fall.
The Honest Counter-Case
The tier finding depends on your lender's grid. I've said it twice because it's the load-bearing assumption. The step was verified at 1.25 on one sheet, not assumed, but it is one sheet. If your grid steps at 1.20, the coverage math above moves, and a real ARM discount on that grid could be worth something the one on mine is not.
The spread ranges were published ranges, not my rate sheet. I said this in August and it turned out to matter. On my sheet the spread is zero. If your lender quotes the ARM back 100 basis points, the payment savings get real enough to matter on their own and the coverage lift gets bigger too. Re-run it. Do not assume my spread in either direction.
Rates could fall. The Fed held at 3.50%–3.75% on July 29, 2026, on a 9–3 vote with Hammack, Kashkari, and Logan each preferring a quarter-point increase, and that is the most recent decision as of this update. But five years is long enough for the entire thesis to invert, and an ARM taken today resets below its start rate if 30-day Average SOFR falls under 2.875%. The decision rule survives either way, which is exactly why it's framed around the exit instead of the forecast.
Prepayment penalties may bind before any of this does. A five-year step-down prepay can be the real constraint on your exit regardless of which product you pick, and it interacts with everything above. That's its own post — read it before you sign either row.
If coverage is the problem, the product may not be the answer. When the deal won't reach a tier on any structure, no-ratio DSCR is a different conversation, and the Complete Guide covers the qualification path end to end. New Jersey's tax math is the structural reason this file is tight at all — more on that on the New Jersey DSCR page.
One sheet, one lender, one day. Pricing moves. Another lender may show a real spread between the two columns, and the same lender may show one next month. The instruction is not that identical pricing holds everywhere. It is to ask for both columns in writing, on the same day, and compare them yourself before you assume anything in either direction.
FAQ
Is a DSCR ARM cheaper than a 30-year fixed? Not necessarily, and on the wholesale DSCR sheet I price from, dated September 8, 2026, not at all. The 5/6 ARM and the 30-year fixed carried the same rate at the same price at every rung, with the same 6.375% floor. On that sheet the borrower taking the ARM accepts the index risk and receives nothing in exchange. Other lenders may price a spread. Ask to see both columns on your own file, same day, same lender.
Does an ARM improve my DSCR? Only if it prices below the fixed, because DSCR is rent divided by PITIA and the only thing an ARM can change on day one is the payment. On the traced Newton two-family, priced off a sheet where both products sit at 7.375%, coverage is 1.1398 on either one. Where a spread does exist, the lift matters only if it carries you across your lender's pricing step, which on this sheet sits at 1.25. Ask where the step is before you choose the product.
What actually decides my rate when a DSCR ARM resets? Index plus margin. The margin on the sheet I price from is 4.50%, well above the 2.50% to 3.50% published ranges I cited in the original version of this post. At the September 4, 2026 30-day Average SOFR of 3.64586%, the fully indexed rate is 8.146%. The margin is a fixed number written into your note at origination and it never changes. It is negotiable on the front end and immovable afterward, so ask what it is before you sign.
What happens to my DSCR when a DSCR ARM resets? It falls if the index plus margin exceeds your initial rate, and it falls at the moment you need coverage to qualify for the refinance that gets you out. On the traced $350,000 loan, a reset to 8.146% at the September 4, 2026 index drops coverage to 1.110 even after five years of rent growth, and the 9.146% row lands at 1.045. That still cash-flows. What it may not do is clear a refinance requirement.
Can a DSCR ARM reset lower than my starting rate? Yes in principle, but not at today's index on this sheet. A 7.375% start with a 4.50% margin reprices to 8.146% at the September 4, 2026 30-day Average SOFR of 3.64586%, up 77 basis points. For it to reset at or below the start rate, 30-day Average SOFR would have to fall below 2.875%. That is a real branch and it is not the base case. The original version of this post said a reset near flat was not a remote case; at a 4.50% margin it is.
Should I take a fixed rate or an ARM on a rental property? Match the product to a defined exit, not to a rate forecast, and confirm in writing that the ARM prices below the fixed before you take it. Take the ARM only when you have a specific exit inside the fixed period, a sale, a 1031, a refinance with a named trigger, and the ARM actually costs less on your lender's sheet. Take the 30-year fixed when the hold is open-ended or when the two columns price the same, because then the reset is a risk you are accepting for nothing. If the exit is what you are optimizing for, the prepayment penalty term is the lever, not the product.
Do rate caps protect me at reset? Partially, and less than the name suggests. On the sheet I price from the 5/6 carries 2/1/5 caps, 2% maximum on the first adjustment, 1% on each subsequent one, 5% over the life of the loan, and the 7/6 and 10/6 carry 5/1/5. On the traced 7.375% ARM the 5/6 first-adjustment ceiling is 9.375%. Every scenario in this post lands below it, so the cap never rescues any of them. 30-day Average SOFR would have to rise about 123 basis points above its September 4, 2026 level before that cap changes anything. Confirm your actual cap structure on the note.
Run It on Your File
Three questions decide this, and you can answer all three this week: does the ARM actually price below the fixed on your lender's sheet, in writing, on the same day? Where does the pricing grid step? And what margin is on the ARM note? If the two columns price the same, take the fixed and spend the negotiation on the prepayment term instead. If the ARM does price back, run the reset table above at your margin before you decide the discount is worth it.
Send me the numbers — price, rent, down payment, and both rows of the term sheet — and I'll run this exact model on your deal. You can also rough it yourself with the DSCR calculator first. Send the numbers here and we'll tell you whether the ARM is earning its place before you commit to a product.
Dominick Prevete — 31 years in real estate finance. Founder, National Loan Provider. 25 Main Street, Unit B, Sparta NJ. (908) 220-6404.
Index, margin, cap structures, floor rates, the DSCR-tier and prepayment-penalty price adjustments, and the price-to-rate ladder are taken from a single wholesale DSCR rate sheet dated September 8, 2026. The lender is deliberately not named and the grid is not reproduced; every file prices to its own specifics on the day it is priced. 30-Day Average SOFR is as published by the Federal Reserve Bank of New York for September 4, 2026 and resets daily. The fully indexed rate is computed unrounded; if your note rounds to the nearest one-eighth at adjustment, the reset figures move by a few dollars. The rate decision, vote, and dissents are from the FOMC statement of July 29, 2026. Rent-growth figures are from Cotality's Single-Family Rent Index, May 2026 data. Newton's tax rate is the 2025 general rate per the Town of Newton Tax Collector; the equalization ratio is from the NJ Division of Taxation 2026 Chapter 123 table. The reset table is a sensitivity analysis, not a forecast. This is market commentary, not a commitment to lend.
Loans are for business purposes only and are not subject to TILA, RESPA, or HOEPA. Not for primary residences. Equal Housing Opportunity. All loans subject to underwriting approval. Rates and terms shown for illustration; actual rates depend on deal specifics. National Loan Provider does not arrange financing on owner-occupied properties. Every loan we arrange is business-purpose investor financing.