This article is educational analysis, not mortgage advice; all rate figures reflect the week ending September 10, 2026 unless a different date is stated inline, and your actual terms depend on your lender’s rate sheet and Note Rider.
TL;DR — Quick Verdict
- The 5/1 ARM contract rate averaged 5.82% versus 6.85% for the 30-year fixed in the Mortgage Bankers Association survey for the week ending September 4, 2026 — a spread of 1.03 percentage points.
- On a $343,280 loan (80% of the $429,100 median existing-home price reported by the National Association of REALTORS® for August 2026), that spread saves roughly $231 per month, or $13,848 across the 60-month fixed period.
- Break-even does not arrive before month 81 in any scenario modeled — meaning a borrower who sells or refinances inside five years keeps the savings outright, and only a sharp reset ever claws them back.
- The 5/1 ARM carried 0.84 points including origination fee, roughly $2,884 on that loan — a cost that pushes the true break-even later than the payment comparison suggests.
- ARM share climbed to 8.5% of applications in the week ending September 4, 2026 — the highest level since June — up from 8.0% the week before and 7.1% in mid-July, as fixed rates pushed to their highest level since June 2025.
- Take the ARM only if your documented exit horizon is under five years and you can absorb the worst-case reset payment on current income.
Just over a full percentage point. That is the entire prize an adjustable-rate borrower is playing for in September 2026, according to the Mortgage Bankers Association’s Weekly Applications Survey. It sounds thin, and until recently it was — but on a mid-sized loan it now moves about $230 a month, which is why Rocket Mortgage, United Wholesale Mortgage, and most credit unions keep 5/1 and 7/1 products on the rate sheet.
The trouble is that almost every ARM comparison stops at the monthly payment. That comparison is wrong, because it ignores three things that decide the outcome: the points you paid to get the teaser rate, the amortization difference that leaves you with a smaller balance on the ARM at reset, and the actual worst-case payment your Note Rider permits. This analysis builds the full break-even model on a $343,280 loan using verified September 2026 rates, shows where the crossover month actually falls under three reset scenarios, and identifies the specific borrower profiles for whom the math works. Freddie Mac’s own Primary Mortgage Market Survey notes that ARMs remain concentrated in larger nonconforming loans — a clue about who is genuinely benefiting.
What ARM and Fixed Rates Actually Cost in September 2026
Start with the numbers on the table, not the ones from memory. Freddie Mac’s Primary Mortgage Market Survey put the 30-year fixed-rate mortgage at 6.76% as of September 10, 2026, up from 6.71% the previous week — the third straight weekly increase and the highest reading of 2026. The 15-year fixed-rate mortgage averaged 6.09%. Freddie Mac’s survey no longer reports a 5/1 hybrid figure, so the ARM comparison has to come from the Mortgage Bankers Association, which surveys contract rates and points directly from lender applications.
Points matter enormously here and are almost always omitted from consumer rate tables. The MBA figures below include the origination fee, which is why the 5/1 ARM’s 0.84 points is not a rounding detail — it is roughly $2,884 in upfront cost on the modeled loan.
Sources: Mortgage Bankers Association Weekly Applications Survey, week ending September 4, 2026 (verify at mba.org); Freddie Mac Primary Mortgage Market Survey, September 10, 2026. Freddie Mac states it can no longer report average fees and points under current Loan Product Advisor requirements.
Note the two different 30-year figures. Freddie Mac’s 6.76% reflects borrowers with excellent credit and 20% down; MBA’s 6.85% is a broader application-weighted contract rate. This analysis uses the MBA pair throughout so that the ARM and the fixed come from the same survey methodology on the same week — mixing sources would manufacture a spread that no single lender actually offers. Anyone benchmarking against a live quote should also review current 30-year fixed rate trends before locking.
How the Break-Even Calculation Actually Works
Picture a buyer closing on a $429,100 home — the median existing-home price NAR reported for August 2026 — with 20% down. The loan is $343,280. She is choosing between the 6.85% fixed and the 5.82% 5/1 ARM.
Monthly principal and interest on the fixed loan runs approximately $2,249. On the ARM it runs approximately $2,019. The gap is $231 a month, or $13,848 over the 60-month fixed period. But she paid 0.84 points on the ARM versus 0.67 on the fixed — a difference of 0.17 points, or $584. Net cash advantage entering the reset: about $13,264.
Amortization adds a second, quieter benefit. At the lower rate, more of each payment attacks principal. After 60 payments the ARM balance sits roughly $3,893 below the fixed balance. Combine cash savings and the balance advantage and the ARM borrower is ahead by roughly $17,157 the day the reset clock starts. That is the cushion the reset has to burn through.
Break-even is the month at which cumulative post-reset overpayment consumes that cushion. If the reset pushes her payment $450 above the fixed payment, the cushion lasts about 38 months — break-even lands near month 98, or roughly year eight. If the reset is milder, she may never cross. Running the same exercise on mortgage points and rate buydown math shows why upfront cost has to enter the model rather than sitting in a footnote.
5/1 ARM vs 30-Year Fixed: Which Is Better for a Five-Year Horizon?
Everything depends on what the rate becomes in month 61, and that is governed by three contract terms rather than by the market alone: the index, the margin, and the caps. Most ARMs originated since the LIBOR transition reference an SOFR-based index. The 30-Day Average SOFR published by the Federal Reserve Bank of New York read 3.65% on September 14, 2026 — the most recent published value at the time of this update — and has held in the 3.6%–3.7% range through most of 2026 based on the FRED series readings; the exact daily value on any reset date is what governs.
Margins and caps are lender-specific and appear only in your Adjustable Rate Note Rider — there is no national average that binds any individual loan. Rather than assert a figure, model your own: fully indexed rate equals index plus margin, then apply the initial adjustment cap. Three scenarios below use the 3.65% index reading with margins and caps spanning the common contractual range.
Modeled by Real Cost Report on a $343,280 loan using MBA contract rates for the week ending September 4, 2026, and 30-Day Average SOFR readings from the Federal Reserve Bank of New York via FRED series SOFR30DAYAVG. Margins and caps are illustrative; substitute the values in your own Note Rider.
Verdict
For a documented five-year horizon, the 5/1 ARM wins in every scenario modeled — break-even never arrives before month 81, well past the exit. For an open-ended horizon, the fixed wins, because the adverse scenario costs $173 a month indefinitely and the worst case costs $833. The deciding variable is not your rate forecast; it is whether your exit date is a plan or a hope.
What Most Borrowers Get Wrong About ARM Break-Even
Four errors show up repeatedly, and each one shifts the crossover month by years.
Mistake 1: Comparing rates instead of total cost
A borrower sees 5.82% against 6.85% and stops. The consequence is ignoring 0.84 points on the ARM against 0.67 on the fixed — a $584 swing that delays break-even by roughly two and a half months. Correct action: build the comparison on annual percentage rate and total cost of borrowing, the approach detailed in comparing lenders by APR and total borrowing cost.
Mistake 2: Assuming you will refinance before the reset
NAR’s 2025 Profile of Home Buyers and Sellers found the typical seller had owned for 11 years — a record high — and buyers expect a median tenure of 15 years. Plans to exit at year five collide with lock-in effects and life circumstances. Correct action: qualify for the worst-case payment, not the teaser payment, and treat refinancing as an option rather than a plan. Prevailing conditions in a mortgage rate forecast five years out are unknowable.
Mistake 3: Confusing the Fed with your reset
Borrowers watch Federal Open Market Committee meetings expecting their ARM to follow. Short-term policy rates do influence SOFR, but the relationship between policy and mortgage pricing is indirect — a distinction covered in why mortgage rates track the 10-year Treasury. Correct action: track your specific index series, not headlines.
Mistake 4: Treating the initial cap as the real risk
Most attention goes to the first adjustment. The lifetime cap is the figure that governs the tail. Correct action: compute the payment at the lifetime cap and confirm it clears your debt-to-income ratio at current income.
Who Should Take the ARM in 2026 — and Who Should Not
ARM share of applications climbed to 8.5% in the week ending September 4, 2026 — the highest level since June — up from 8.0% the week before and 7.1% in mid-July, per the Mortgage Bankers Association. Joel Kan, CMB, MBA’s Vice President and Deputy Chief Economist, noted that the 30-year fixed rate pushed to its highest level since June 2025 that week, even as more borrowers shifted into ARM loans to soften the payment shock. Roughly one borrower in twelve is choosing an ARM, and the market is telling you why: 1.03 percentage points of spread is real money to give up against open-ended rate risk.
Take the ARM if three conditions hold together. Your exit is contractual rather than aspirational — a military relocation, a fixed-term employment posting, a property already slated for sale. Your loan exceeds the $832,750 baseline conforming loan limit FHFA set for 2026, where ARM discounts run wider, as Freddie Mac notes ARMs are most popular for higher loan size nonconforming loans; the mechanics appear in jumbo versus conforming loan rate differences. And your income comfortably absorbs the lifetime-cap payment — $3,082 monthly in the worst-case row above, against $2,249 fixed.
Skip the ARM if any single condition fails. Retirees and pre-retirees on fixed income face a particular asymmetry: the reset can arrive after earned income has stopped, and an $833 monthly increase against a defined-benefit or drawdown budget is not absorbable. Borrowers with credit below the top tier should also reconsider, since ARM pricing tends to widen faster by tier than fixed pricing — see credit score impact on mortgage rates by tier. Before choosing on the ARM axis at all, run fixed versus adjustable rate mortgage cost comparison and the 15-year versus 30-year total interest comparison — at 6.09%, the 15-year fixed delivers much of the ARM’s rate discount with none of the reset risk.
Frequently Asked Questions
Is a 1.03-point spread wide enough to justify an ARM?
More so than in recent months. A 1.03-point spread between the 5.82% 5/1 ARM and 6.85% fixed contract rate in the MBA survey for the week ending September 4, 2026, produces about $231 monthly on a $343,280 loan. That is meaningful for a three-to-five-year horizon and still notable over thirty years. The market appears to be responding — ARM share climbed to 8.5% of applications, its highest level since June.
What index will my ARM reset against?
Most ARMs originated after the LIBOR transition reference a SOFR-based index, commonly the 30-Day Average SOFR administered by the Federal Reserve Bank of New York, which read 3.65% on September 14, 2026 and has held in the 3.6%–3.7% range during most of 2026. Your Note Rider names the exact index, the lookback convention, and the margin added to it. Never assume — the rider governs.
Does a 7/1 ARM change the break-even math?
Yes, in two directions. A 7/1 typically prices above a 5/1, shrinking the monthly savings, but it extends the protected period by 24 payments. Applying the base scenario, roughly $231 monthly over 84 payments rather than 60 accumulates a larger cushion before any reset. The 7/1 generally suits borrowers whose exit horizon falls between years five and seven.
Can I count on refinancing out of the ARM before it resets?
Treat it as an option, not a plan. Refinancing requires qualifying income, sufficient equity, and acceptable rates simultaneously five years from now. NAR’s 2025 Profile found median seller tenure at a record 11 years, suggesting exits routinely run longer than intended. Qualify for the lifetime-cap payment — $3,082 monthly in the worst-case model above — before signing.
How We Researched This Article
Rate inputs came from two primary surveys, deliberately not blended. Contract rates and points for both the 5/1 ARM and the 30-year fixed came from the Mortgage Bankers Association Weekly Applications Survey for the week ending September 4, 2026, released September 9, 2026, because it is the only recurring survey that reports ARM contract rates alongside fixed rates from the same lender application pool on the same week. Freddie Mac’s Primary Mortgage Market Survey for September 10, 2026, supplied the 30-year and 15-year fixed benchmarks; Freddie Mac discontinued its 5/1 hybrid series and states it can no longer report average fees and points under current Loan Product Advisor requirements, which is why MBA rather than Freddie Mac anchors the ARM comparison.
Property and loan sizing used the National Association of REALTORS® Existing-Home Sales report for August 2026, released September 11, 2026, which put the median existing-home price at $429,100 — down 1.2% from July’s $434,100 but still up 1.6% year-over-year, marking the 38th consecutive month of annual price gains. Existing-home sales fell to a 3.98 million seasonally adjusted annual rate, dipping below 4 million for the first time since June 2025, while total inventory rose to 1.62 million units, the highest level in more than a decade. The modeled loan of $343,280 assumes 20% down, matching the borrower profile Freddie Mac’s survey covers. Conforming thresholds came from the Federal Housing Finance Agency’s 2026 conforming loan limit announcement, unchanged at $832,750. Behavioral assumptions about holding periods came from the NAR 2025 Profile of Home Buyers and Sellers.
Index values referenced the 30-Day Average SOFR administered by the Federal Reserve Bank of New York, retrieved through FRED series SOFR30DAYAVG at the Federal Reserve Bank of St. Louis, which published a value of 3.65% for September 14, 2026 — essentially unchanged from the 3.62% reading used in the prior edition of this analysis, and still within the 3.6%–3.7% band that has held for most of 2026.
Two limitations deserve emphasis. All payment figures, break-even months, and reset scenarios are modeled, not measured — margins and caps are contractual terms that vary by lender and by loan, and no national average binds any individual borrower, so the scenario table spans a plausible contractual range rather than asserting a single figure. Payments cover principal and interest only, excluding property taxes, insurance, and any mortgage insurance, which do not differ between the two products and therefore do not shift the crossover month. Research was last conducted September 2026. All figures were verified against named primary sources before publication.