Adjustable-rate mortgages earned their bad name in the 2000s and have never fully shaken it. But the 2026 ARM is a different instrument, the rate environment is nearly the opposite of 2014’s, and for a specific kind of borrower the math can work. The 2014 version of this piece compared a 4.33% 30-year fixed with a 3.37% 5/1 ARM. Both numbers are ancient history, and so is the 5/1 ARM itself.

What a 2026 ARM actually is

The old standard was the 5/1 ARM: five fixed years, then a reset every year, priced off LIBOR. LIBOR is gone; its last U.S. dollar settings ceased on June 30, 2023, and new agency ARMs price off a 30-day average of SOFR, the Secured Overnight Financing Rate published by the New York Fed. The products Fannie Mae and Freddie Mac buy today are the 5/6, 7/6 and 10/6 ARMs. The first number is the fixed period in years; the second is the adjustment interval in months. A 5/6 ARM holds its rate for five years and then resets every six months, to the SOFR average plus the margin written in your note. Twice-a-year resets mean a 2026 ARM tracks the market more closely than the old annual version, in both directions. Our mortgage-basics guide compares fixed and adjustable loan types side by side.

The discount, and why it is volatile

An ARM only makes sense if the fixed period is cheaper than a fixed-rate loan, and that spread is not a constant. In the Mortgage Bankers Association’s weekly survey for the week ending August 21, 2026, the average contract rate was 6.78% on a 30-year fixed conforming loan and 5.98% on the five-year-fixed ARMs it tracks: a discount of about 0.8 of a point. ARMs were 7.9% of applications that week, up from 7.7%. Treat those as a snapshot, not a rule: the MBA figures move every week and the discount moves with them, so check the current spread before you assume one exists. Freddie Mac’s survey for the first week of September 2026 put the 30-year fixed at 6.71% and the 15-year fixed at 6.04%; it stopped publishing an ARM rate in 2022.

At the August spread, the money is real. On a $400,000 loan, 6.78% costs $2,602.37 a month in principal and interest and 5.98% costs $2,393.06, a difference of about $209 a month, or roughly $12,600 across the five fixed years. That is the prize; the rest of this article is what you risk to collect it.

Caps: your worst case, in writing

Every ARM has three rate caps, the most important numbers in the loan after the rate itself:

  • Initial adjustment cap: the most the rate can rise at the first reset. Two or five percentage points are common.
  • Periodic cap: the most it can move at each later reset. One or two points are common.
  • Lifetime cap: the most it can rise over the life of the loan. Five points is the most common.

Lenders write these as a trio such as 2/1/5 or 5/2/5; structures vary by product and lender, so read your own loan’s numbers. Then run the worst case. On our $400,000 5/6 ARM at 5.98%, the balance after five years is about $372,100. If the first reset used up a full 2-point initial cap, the rate would be 7.98% and the payment about $2,867. If the rate eventually hit a 5-point lifetime cap, 10.98%, the payment on that balance would be about $3,642, roughly $1,250 a month more than the starting payment. A borrower who can absorb that number has bought an option. A borrower who cannot has bought a bet.

The rate premise has flipped

In 2014 the case against ARMs was that rates sat at historic lows and could only go up. In 2026 the 30-year fixed has spent the year in the mid-6s, down from a peak near 8% in October 2023; the Federal Reserve cut three times in 2025 and held through the first half of 2026, and its median projection puts the policy rate near 3.1% by March 2027; and forecasts have the 30-year around 6.4% through year-end. February 2026 even produced a brief dip to 5.98%. None of that is a guarantee; a reset that lands in a rising market still costs you. But a reset in a flat-to-falling market can leave you at or below your starting rate, a scenario the 2014 article could not imagine. For the full picture of what could move rates, see our mid-2026 rate outlook.

Who an ARM is right for in 2026

  • You will sell inside the fixed period. A known transfer date, a starter home you plan to outgrow, a five-year plan you actually believe. Pick a fixed period longer than your horizon; a 7/6 or 10/6 gives the plan room to slip, so ask what the longer fixed period costs.
  • You would refinance if the reset were ugly, and you could. That means a payment cushion and a credit profile that will qualify again; note that Fannie Mae’s manual-underwriting floor is higher for ARMs (640) than for fixed loans (620).
  • You can carry the capped payment. If the lifetime-cap payment above would break the budget, the discount is not worth it. The ARM is for borrowers who could afford the fixed loan and choose to pocket the spread, not for those who need the lower rate to qualify.

When it does not make sense

Do not take an ARM because it is the only way the payment fits; that is the borrower the reset hurts most. Do not take one if you expect to stay long past the fixed period and would rather not gamble on the refinance window. And do not take one on the theory that rates must fall: that is the mirror image of 2014’s error.

The alternative the original article got right

The 2014 piece noted that the 15-year fixed was priced almost exactly like the 5/1 ARM. That is still true: 6.04% for the 15-year against 5.98% for the ARM in the two surveys above, taken two weeks apart. On $400,000, the 15-year costs $3,384.08 a month, far more than the ARM’s $2,393.06, but the rate is guaranteed for the whole loan and the house is paid off in 15 years. If your real goal is a low rate rather than a low payment, the 15-year gives you the rate without the reset. Our 15-year versus 30-year comparison lays out that trade.