ARM vs. fixed-rate mortgage: which one actually wins?
The adjustable rate is lower today, so the question is what you are giving up for it and how long you would hold the loan. Two numbers frame it: the fee-adjusted gap between the two rates, and the median seller now stays eleven years — an all-time high. Here is the math in both directions, including the worst case under the caps, so you can weigh it on your own horizon.
The ARM is cheaper right now — check the points column too.
In the week ending 17 July 2026, the mortgage industry’s weekly application survey put the average 30-year fixed conforming rate at 6.69% and the average adjustable at 5.97% — a spread of 72 basis points. Both at 80% loan-to-value.
That is a real gap and it is worth having a conversation about. But look at the next column, because this is where ARM marketing quietly stops reading. The fixed rate carried 0.62 points. The adjustable carried 1.11 points — half a point more in fees, paid at closing.
Half a point buys roughly 12 to 15 basis points on a 30-year fixed. So the honest, fee-adjusted spread is not 72 basis points. It is closer to 55 or 60. Still meaningful — but if you have been comparing a headline ARM rate against a headline fixed rate, you have been comparing two loans with different amounts of money already spent on them.
And a survey average is not your quote. Spreads on adjustable products are tight, and the loan-level adjustments stack differently than they do on a fixed rate, so the ARM discount is largest on the strongest files and narrows as credit score falls or loan-to-value rises. A 740 at 80% often finds the ARM barely better than the fixed once the adjustments are applied. Price both on your actual numbers before you spend any time on the strategy question — if the gap is a quarter point, there is not much left to weigh.
One more thing about that survey: it still calls its adjustable series “5/1 ARM.” The conforming product has been a 5/6 — six-month adjustments — since the industry moved off LIBOR. When even the industry’s own benchmark is labeled with the old product name, you can see how much of the public information about ARMs is a few years behind. How the current product actually works is worth reading before you compare anything.
The median seller now stays eleven years.
The National Association of Realtors tracks how long sellers own their homes before selling. The 2025 figure is a median of eleven years, the highest ever recorded — and the reason is the same force that is shaping every housing decision right now. Millions of people are sitting on mortgages at rates they cannot replace, so they are not moving. The lock-in effect did not just freeze inventory. It extended everyone’s tenure.
| Plan | Fixed period | Years the median owner spends adjusting |
|---|---|---|
| 5/6 | 5 years | About 6 |
| 7/6 | 7 years | About 4 |
| 10/6 | 10 years | About 1 |
Set that against the caps and it gets sharper. A 5/6 spends six years climbing a staircase. A 7/6 spends four years exposed — and its first adjustment cap is 5%, not 2%. Only the 10/6 roughly matches the median tenure, and it carries the same 5% first-adjustment cliff.
This does not kill the ARM argument. It relocates it. The case for an adjustable rate is no longer a population statistic that applies to everybody — it is a claim about your circumstances, and it needs a specific reason behind it. A residency that ends. A relocation clause in a contract. A house you already know you will outgrow when the second child arrives. Those are real, and for those borrowers the ARM is frequently the correct answer.
“I feel like we probably won’t stay that long” is not one of those reasons. That is the sentence that was true when people moved in seven years, and it is the sentence that is now costing people money.
What you actually bank in the fixed period.
Two things make the fixed period worth more than a simple rate comparison suggests. The first is that your interest cost is front-loaded — on a 30-year loan at these rates, about 86% of the first payment is interest, and roughly 47% of all the interest you will ever pay falls in the first ten years, because that is when the balance is highest. A rate saving early is worth more than the same saving later. The second is that a lower rate pays down principal faster.
So here is the comparison done properly: net of the extra points, and counting the principal difference.
$400,000, thirty-year term, 5/6 ARM at 5.97% against a fixed at 6.69%
The monthly difference
Fixed: $2,578 a month. ARM: $2,390. You keep $188 a month.
Over the five-year fixed period
$11,278 of payment savings, less the $1,960 of extra points at closing, gives $9,318.
Plus the principal you did not notice
At the lower rate more of each payment goes to principal. After sixty months the ARM balance is $3,173 lower than the fixed balance would be.
Total five-year advantage: about $12,491. That is real money, and it is understated by the headline rate gap rather than overstated. The question is what happens in month 61.
The worst case is not a cliff. It is a staircase, and it is steep.
Here is the same loan, assuming the index rises enough that every adjustment goes to its maximum. This is the ceiling the contract permits, not a forecast.
| Month | Rate | Payment | vs. the fixed at $2,578 |
|---|---|---|---|
| 1–60 | 5.97% | $2,390 | $188 cheaper |
| 61 | 7.97% | $2,864 | $286 more |
| 67 | 8.97% | $3,112 | $533 more |
| 73 | 9.97% | $3,364 | $786 more |
| 79 onward | 10.97% | $3,621 | $1,042 more |
Two things jump out of that table, and they point in opposite directions.
The first adjustment is survivable. Month 61 takes you to $2,864 — only $286 a month above what the fixed rate would have cost you all along. You banked $12,491 getting there, which covers that increase for roughly three and a half years. If the plan was to be gone by year six or seven, the worst case at the first adjustment is not frightening.
The ceiling is a different animal. Eighteen months after the first adjustment, the same loan is at $3,621 — $1,042 a month above the fixed payment, and it stays there. That is $12,500 a year, every year, for the remaining term. The five-year saving is gone inside the first twelve months at the ceiling.
So the real test is not “can you handle the first adjustment.” Almost everyone can handle the first adjustment. The test is whether you can handle month 79 — and whether you have an exit that does not depend on the market cooperating. If your answer to the ceiling payment is “I would just refinance,” read the next section carefully, because that is precisely the plan that stopped working for a lot of people.
A goal worth having, and not a guarantee you can rely on.
This is the most common thing I hear, and it is worth taking seriously rather than dismissing, because sometimes it is right.
But notice where the eleven-year tenure figure came from. It is an all-time high because rates moved against people. Millions of homeowners who fully intended to refinance, or to move, discovered that the market had closed the door. They are not stuck because they made a foolish choice. They are stuck because a plan that depended on rates behaving met rates that did not.
An adjustable-rate mortgage whose exit strategy is “refinance later” is that same bet, with a contractual deadline attached. The fixed rate is the version of the deal where you do not need to be right about anything.
What is genuinely in your favor
An ARM that is a qualified mortgage cannot carry a prepayment penalty — penalties are restricted to fixed-rate qualified mortgages. So there is no charge for leaving, ever. And your servicer must give you notice 210 to 240 days before the first adjusted payment is due, which is seven to eight months to act.
You are not trapped, and you will not be surprised. You are simply exposed to whatever the market is doing on a date you do not get to choose.
Three questions to answer for yourself.
1. Do you have a documented reason your horizon is short?
Not a feeling. A reason with a date attached: a training programme that ends, a relocation clause, a visa term, a house that structurally will not work once circumstances you already know about arrive. If the reason would survive being written down and read back to you in five years, it counts.
2. Does the file work at the qualifying rate, not just the note rate?
A 5/6 is underwritten at the greater of the note rate plus 2% or the fully indexed rate. If your debt-to-income only clears at the note rate, the ARM is not a cheaper version of the same loan — it is a loan you cannot actually afford, priced attractively. Interestingly, a 7/6 or 10/6 is generally underwritten at the note rate itself, so if the shorter ARM is what makes your file work, the answer is a longer fixed period, not a shorter one.
3. Can you carry the ceiling?
Note rate plus five, on the re-amortized balance. In the example above that is $3,621 against a starting payment of $2,390. If that number would end your household, the answer is the fixed rate, regardless of how attractive the first five years look.
Where I see the ARM win most cleanly: a borrower with a genuinely short and specific horizon, comfortable capacity above the qualifying rate, and enough liquidity that the ceiling is an annoyance rather than an emergency. Where I see it go wrong: a borrower who is stretching, who chose the ARM because it was the only way the payment fit, and whose exit plan is a rate forecast. That is the same file, five years apart.
And one structural note if you are building rather than buying: a single-close construction loan permits converting from an adjustable amortization to a fixed one at completion — the only amortization change allowed, and it runs one direction. How construction-to-permanent financing works covers where that sits.
ARM share is running below its long-run average.
Adjustable-rate mortgages were 7.7% of all mortgage applications in the week ending 17 July 2026. The long-run average is about 11.3%. At the peak of the last cycle, between 2003 and 2006, ARMs ran as high as 36% of applications.
So we are running below the historical norm and a very long way below the period everyone remembers. And the composition has changed as much as the volume: when ARM share last spiked, more than two thirds of ARM applications carried five-, seven- or ten-year fixed periods. The modern ARM borrower is buying a long fixed period, not a teaser rate. That is a healthier market than the headline product name suggests.
I mention this because the low share cuts two ways, and I would rather give you both. It suggests the market does not currently think the spread justifies the exposure — which is a data point worth respecting. It also means an ARM is not a crowded trade, and the borrowers it genuinely fits are getting a product few people are competing to sell them well.
I make the same commission either way.
That is worth stating plainly, because the ARM conversation is one where you should know whether the person advising you has a stake. I do not. Whichever way this goes, my file looks the same.
What I actually do is run both, all the way out — the five-year advantage net of points and principal, and the worst case at the ceiling — and put them next to each other. Then I ask the three questions above. Most of the time the conversation ends with the fixed rate, because most people do not have a documented reason their horizon is short, and eleven years is a long time to be exposed.
But not always. For the borrower who has a real reason, the ARM is leaving twelve thousand dollars on the table if they take the fixed rate out of vague unease. That borrower deserves to have the math done properly rather than to be talked out of it by somebody who finds the product embarrassing.
Let’s run both and see what the trade is actually worth.
Send me the scenario and I will price the ARM and the fixed side by side — net of points, with the principal difference counted, and with the worst-case payment at the ceiling spelled out. Then you decide.
Talk First
Text or email the situation — price, down payment, and honestly how long you expect to be there and why. One business day, and no attempt to talk you into either one.
Or Get Real Numbers
Send the full scenario and I will come back with both loans priced at wholesale, the margin and caps stated, and the ceiling payment calculated on your actual balance.
Send my scenario →Rate and points figures are from the Mortgage Bankers Association Weekly Applications Survey for the week ending 17 July 2026, at 80% loan-to-value, and change weekly. Median seller tenure is from the National Association of Realtors 2025 Profile of Home Buyers and Sellers. ARM application share figures are from the same weekly survey; the long-run average and 2003–2006 peak are from MBA historical analysis. Payment illustrations assume a $400,000 loan, a 30-year term and the maximum permitted adjustment at every reset; the worst case is a contractual ceiling, not a forecast. Your actual terms depend on credit, property, occupancy, program and investor. Forest Hills Mortgage · Matt Mergo · NMLS #563819. Equal Housing Opportunity.