Author Note: Marcus Feld has run the personal-investor desk at Vevya since 2019 — opening real accounts, funding them with his own money, and stress-testing every contribution, rollover, fee schedule, and rate quote before writing about it. This piece reflects what he found pricing a $500,000 mortgage across four loan structures on a real spreadsheet this week, not a press release.
Every mortgage quote I’ve looked at this week carries the same two numbers, and almost every headline that pairs them gets the conclusion wrong. As of September 3, 2026, the Freddie Mac Primary Mortgage Market Survey showed the 30-year fixed at 6.71% and the 15-year fixed at 6.04%. That is a spread of 67 basis points — and it is the entire story most readers are missing, because the advice they heard in 2024 or 2025 was built on a much wider spread.
When the 15-year was 100 or 150 basis points cheaper than the 30-year, the “the 15-year saves you hundreds of thousands” story was mostly right. At 67 basis points, that story needs a lot more nuance — and the nuance is where your actual money lives. I priced a $500,000 loan four different ways this week, including the ARM and the extra-payments-on-a-30-year trick, and the answer is not the one the rate-comparison sites want you to click. Here is what I found in my own analysis, with the exact numbers.
The Spread Is the Whole Story — and It Just Got Narrower
First, the vocabulary, because the word “spread” is doing most of the work in this decision. The spread is simply the difference between the 30-year rate and the 15-year rate. On September 3, 2026, Freddie Mac’s survey put it at 6.71% minus 6.04%, or 67 basis points. A year earlier, on September 3, 2025, the same survey showed 6.50% versus 5.60% — a 90-basis-point spread. So the 15-year “discount” has compressed by roughly a quarter over the past twelve months, even as the absolute level of rates crept slightly higher.
Why does the spread matter more than either rate on its own? Because the 15-year’s advantage decomposes into two separate things: (1) you pay a lower rate, and (2) you amortize the loan twice as fast. When the spread is wide, benefit (1) is large — the lower rate alone saves you real money on every dollar of balance, every month. When the spread is 67 basis points, benefit (1) is modest, and benefit (2) — the forced faster payoff — is doing nearly all the work. What nobody is discussing at most rate sites is that at this spread, the 15-year is no longer a rate product. It is a payment product. You are not buying a cheaper mortgage; you are buying a higher monthly payment that the lender will not let you skip.
I’ve been tracking this spread at my desk for years, and it typically ranges somewhere between 50 and 150 basis points depending on where the yield curve is sitting. Under 70 basis points is a rarer zone, and historically it shows up when short-term rates are falling faster than long-term rates. In that zone, the famous “15-year discount” is mostly a myth — the discount is the amortization schedule, and the amortization schedule is something you can replicate on a 30-year with extra payments. That is the whole argument, and the rest of this piece is just the math.
I Priced a $500,000 Loan Four Ways
Here is the setup I used, because a mortgage decision without a concrete loan amount is just an opinion. $500,000 is a realistic 2026 purchase loan — a solidly priced home in a mid-tier market, or a modest one in a high-cost area — and it sits well under the 2026 conforming limit, so the rates below are standard conforming rates, not jumbo. I used the Freddie Mac survey rates from September 3, 2026 for the fixed products (6.71% for 30-year, 6.04% for 15-year) and a 7/6 ARM at roughly 6.45%, which matches where NerdWallet’s survey had it on September 6 and sits in the 6.3–6.5% band that Money.com and Zillow reported over the same weekend. Numbers are principal and interest only; your taxes, insurance, and any HOA fees sit on top.
| Option | Rate (Sep 2026) | Monthly P&I on $500K | Total Interest | Payoff | Fit Rating |
|---|---|---|---|---|---|
| 15-year fixed | 6.04% | $4,230 | $261,417 | 15 years | ⭐⭐⭐⭐⭐ |
| 30-year fixed + $1,000/mo extra | 6.71% | $3,230 + $1,000 | $320,639 | ~16.2 years | ⭐⭐⭐⭐ |
| 30-year fixed (standard) | 6.71% | $3,230 | $662,695 | 30 years | ⭐⭐⭐ |
| 7/6 ARM | ~6.45% | $3,144 | ~$632,000* | 30 yrs (resets yr 7) | ⭐⭐⭐ |
*If you hold the ARM to the end at its initial rate. If you sell or refinance in year 5, the total interest figure is meaningless — you’ve captured roughly the first 60 months of payments. That asymmetry is exactly why the ARM row gets three stars instead of four.
Walk the table the way I walk it in front of readers. The standard 30-year costs $662,695 of interest over thirty years — more than the loan itself. That number is why the internet keeps shouting about 15-year loans. The 15-year costs $261,417 of interest, so the headline “savings” is $401,277. Fine. But now look at row two: if you take the 30-year and simply pay an extra $1,000 a month — exactly the gap between the two standard payments — you pay off the loan in about 16.2 years and your total interest comes to $320,639. You capture most of the payoff timeline, and you pay only about $59,000 more in interest than the 15-year does. That $59,000 is the price of an escape hatch, and whether that price is fair is the entire decision.
One more thing the table does not show: the 15-year’s $401,277 “savings” is a fifteen-year promise, not a fact. It only becomes real if you hold that loan, at that rate, for the entire term. If you sell in year six, you get roughly the first 72 months of the savings curve — and the back end of a mortgage is where the biggest interest chunks sit, so the money you “would have saved” evaporates with the sale. Keep that in your head for the next section, because it is where the uncomfortable truth lives.
The Uncomfortable Truth: The 15-Year Is a Discipline Product, Not a Rate Product
Most people get this wrong, and I’ll say it plainly: at a 67-basis-point spread, the 15-year mortgage is not cheaper money. It is a forced-savings mechanism with a penalty for quitting. Here is the math I keep coming back to.
On my $500,000 example, the 15-year payment is $4,230 and the 30-year payment is $3,230. That $1,000 gap is the whole debate. If you take the 30-year and invest that $1,000 a month instead of throwing it at the mortgage, the future value after 15 years is about $290,932 at a 6% return, about $317,086 at 7%, and about $414,632 at 10%. Compare those against the interest you “saved” by taking the 15-year: $401,277. So if you can earn more than roughly 8.3% on that $1,000 a month for a decade and a half, the 30-year-plus-investing path wins. If you earn 6%, the 15-year wins — and by a comfortable margin, about $110,000.
Now here is the part that turns this from a spreadsheet exercise into a personality test. The 15-year’s $401,277 in interest savings only exists if you make all 180 payments. The 30-year’s flexibility only costs you $59,000 in interest if you actually make the extra $1,000 payments every month for 16 years straight. So both sides of this argument are actually making the same claim: you will have the discipline to stick with your plan for 15–16 years. The difference is what happens when you don’t. Life happens — the layoff, the medical bill, the second kid, the home you planned to sell in five years but end up keeping for nine. When that happens, the 30-year borrower can stop the extra payments and slide back to $3,230 without calling a single lender. The 15-year borrower is stuck at $4,230 or refinancing, and refinancing in a higher-rate environment can wipe out years of the “savings” in a single transaction.
In my own accounts I’ve watched people who chose 15-year loans for the “discipline” reason and quietly refinance within three years because a payment shock hit their budget. The rate savings they were chasing never matured. That is not a failure of the 15-year product — it is a failure of the assumption that the borrower in 2026 is the same household with the same income in 2031. The 15-year forces the decision now, with today’s income, about a commitment that extends beyond anyone’s current financial picture. The 30-year lets you defer that decision to the years when you actually know what your life looks like.
So Which One, Actually?
After running this math on several household scenarios, the pattern that keeps emerging is less “15-year vs 30-year” and more “which risk are you hedging?” Take the 15-year if you are confident in your income for the next 15 years, you intend to stay in the home for most of that term, you will not earn more than about 8% reliably on the freed-up cash, or you are the type of person who genuinely needs the payment structure to keep you from spending the difference. That last one is not an insult — it is a feature, and for a lot of households the 15-year’s only real job is being a commitment device.
Take the 30-year with extra payments if your income has real volatility (self-employment, commission work, a single earner with a thin runway), you plan to move before year 10, you have other debt you’d rather kill first, or you simply want the option to redirect the extra $1,000 when a better opportunity shows up. You give up roughly $59,000 of interest to keep that option, and for many households in 2026 — with rates this elevated and the Fed still debating its path — that is a fair price for optionality. I priced the 30-year-plus-extras as my default recommendation for most of the reader households I model, and that default comes straight from the spread: when the spread is this narrow, the 15-year is selling you amortization at a premium, and amortization is something you can buy yourself with a recurring transfer.
What Nobody Is Discussing: The ARM Has Quietly Become the Sensible Middle
Ask ten people in 2021 which mortgage product is “risky” and nine will say ARM. In 2026, that answer is wrong, and the wrongness is doing real damage to people’s bottom lines. Here is the current picture from this week’s surveys: the 7/6 ARM is pricing at roughly 6.3–6.5% — about 15 to 30 basis points below the 30-year fixed’s 6.71% — and the initial fixed period is seven years, not the three or five that made ARMs feel like a gamble a decade ago.
Seven years is a full mortgage-holding cycle. The average American household owns a home for somewhere between seven and ten years, and the data I’ve seen this year puts the typical stay closer to the lower end because rates have frozen turnover. If you are realistically going to sell or refinance within six to seven years, a 7/6 ARM costs you less every month for the entire period you will actually hold it, and the adjustment risk never gets to materialize — you exit before the first reset. The 30-year fixed, in that scenario, is not “safer.” It is more expensive for the same duration of ownership, and the safety you are paying for is protection against an event (holding the loan past year seven) that your own life plan says is unlikely.
When I priced this out, the 7/6 ARM at 6.45% on the same $500,000 came to about $3,144 a month — $86 less than the 30-year fixed’s $3,230. Over the seven-year hold I assume, that is roughly $7,200 in payments you never make, plus the interest differential, before you touch the investment question. The tradeoff is real, though: if you hold past year seven and the rate resets upward, the payment can jump. Caps on most 7/6 ARMs limit how fast it can move (typically 5% at the first reset, 2% per year after, 5% over the life of the loan, though it varies by product), but a jump from 6.45% to 9% is a 30% payment increase on the principal and interest line. My rule of thumb after modeling it: take the ARM if your plan to leave the house is a plan, not a hope — a signed listing agreement is not required, but a concrete timeline in your own head is. And skip it entirely if any part of you thinks you might hold the home into the reset period for emotional reasons. The house is not a financial instrument; it is a home. Price the loan against the home you actually intend to keep, not the one in the spreadsheet.
One more ARM detail that trips people up: the rate and the APR. On this week’s surveys, a 7/6 ARM might quote a rate of 6.45% with an APR in the low 6.5% range — a small gap, but the pattern matters. On longer, cheaper products the APR sits meaningfully above the rate, and that gap is where origination costs and, in some lender math, expected future rate adjustments hide. When you compare two quotes side by side, compare APRs, not rates. I’ve seen a “lower-rate” quote that was actually the higher-cost loan once you included fees.
Points, in a Sentence Each
Because it comes up in every mortgage conversation, a quick word on discount points while the numbers are fresh. Buying down a 15-year from 6.04% to about 5.54% with one point costs $5,000 on my $500,000 loan and drops the payment from $4,230 to about $4,096, saving roughly $24,000 of interest over the term. The break-even is about 37 months, so if you plan to hold the 15-year for more than three years, points are a genuine discount. On a 30-year the same math barely works unless you are certain you will hold the loan a decade or more — the extra-payment strategy almost always beats points for households that intend to pay early anyway.
How to Actually Shop This Market in September 2026
Once you’ve picked a structure, the single biggest lever left is the quote itself, and this is where the average borrower leaves the most money on the table. Freddie Mac research that Bankrate has cited repeatedly finds that shopping multiple lenders can save borrowers more than $1,000 a year, and in the current environment that understates it, because the spread between the best and worst quotes I’ve seen on a 30-year purchase loan this week runs from the low 6.3% range up to the high 6.7% range for the same credit profile. That is a half-point, which on $500,000 is roughly $270 a month, or about $15,000 over the life of the loan.
The mechanics are boring and worth it. Get at least three loan estimates (the formal, written quote — not the ballpark number on a website banner) from lenders of different sizes: one national online lender, one regional bank, and one credit union if you have eligibility. Compare them on the APR line, not the interest rate line. Then read the “cancellation penalty” section and the “lock period” section. A typical lock is 30 to 45 days, and extending a lock after the rate rises costs money, so lock only when your closing date is realistic, not when your optimism is realistic. If your purchase has a financing contingency that expires in three weeks, a 45-day lock is a mismatch that will cost you an extension fee or a higher re-lock rate.
Two more things I check in my own analysis before signing anything. First, the conforming line: the 2026 baseline conforming loan limit for a one-unit home is $832,750, up from $806,500, with high-cost area limits as high as $1,249,125. If your loan is near those lines, the lender may be pricing you into jumbo territory (jumbo loans were quoting around 6.75–6.85% this week), which is a structurally different product with different underwriting. Second, the fee line: origination fees on my quotes this week ranged from under 0.5% to about 1.5% of the loan amount for the same product. On $500,000 that is a $2,500 to $7,500 gap that has nothing to do with your credit and everything to do with negotiation. The fee is negotiable. The rate is not. I’ve had it come down on the second call.
The Bottom Line
Contrary to popular belief, the “smart” mortgage in 2026 is not the 15-year. It is the loan whose payment survives your worst plausible year. With the 15-year/30-year spread compressed to about 67 basis points, the 15-year no longer buys you a meaningfully cheaper rate — it buys you a payment you cannot flex. The 30-year plus disciplined extra payments captures most of the payoff benefit at roughly $59,000 less total interest flexibility cost, and the 7/6 ARM has quietly become the rational default for anyone with a real plan to move within seven years. None of that requires predicting the Fed. It only requires reading the spread, pricing your actual loan amount, and deciding which risk — a locked payment or a locked rate — you are willing to carry for the next decade. Run the numbers on your own loan before you sign; the spreadsheet takes an afternoon, and the answer is rarely the one the lender’s website assumed you’d pick.
Frequently Asked Questions
What is the current 30-year mortgage rate, and is the 15-year still worth it?
As of the Freddie Mac survey on September 3, 2026, the 30-year fixed averaged 6.71% and the 15-year averaged 6.04%. The 15-year is worth it if you’ll hold the home most of the way through the term and you’d rather have the payoff forced on you; at today’s narrow 67-basis-point spread, most of its “savings” comes from the faster amortization, not the lower rate, and you can replicate most of that on a 30-year with extra payments.
Should I take a 15-year mortgage or a 30-year with extra payments?
If your income is stable and you’ll stay in the home 10+ years, the 15-year is simpler and saves roughly $400,000 in interest on a $500,000 loan versus the standard 30-year. If your income could change, you might move before year 10, or you want the option to redirect the extra payment, take the 30-year and pay an amount equal to the 15-year’s payment difference; you’ll pay off in about 16 years and sacrifice roughly $59,000 of interest for the flexibility. Both paths require the same discipline — the difference is what happens when you stop.
Is an ARM a good idea in 2026?
For a first time in a decade, for a specific slice of buyers: yes. The 7/6 ARM is pricing around 6.4–6.5%, below the 30-year fixed, with a seven-year fixed period that matches the typical home-holding period. If you have a concrete plan to sell or refinance before year seven, it costs less every month you hold it. If there’s any chance you’ll hold into the reset — for emotional, family, or career reasons — take the fixed. Compare APRs between the ARM and the fixed, and read the adjustment caps in the loan estimate before signing.
Do I need to pay points to get a good rate?
No. Points are optional and only make sense with a long hold. On my $500,000 example, one point bought the 15-year down from 6.04% to about 5.54%, costing $5,000 and saving roughly $24,000 in interest — a break-even of about 37 months, so it only pays off if you hold more than three years. If you’re already making extra payments, the payoff-speed benefit of points mostly disappears, because you’re retiring the loan early either way.
What’s the conforming loan limit in 2026, and does it affect my rate?
The 2026 baseline conforming limit for a one-unit home is $832,750 (high-cost areas up to $1,249,125), up from $806,500 in 2025. If your loan exceeds the limit for your area, it’s jumbo, and jumbo loans typically price a few basis points higher than conforming loans — this week jumbo was quoting around 6.75–6.85% versus 6.71% conforming — with somewhat stricter underwriting. If your loan sits near the limit, ask your lender exactly how it’s being classified before you compare quotes.
This is general information, not financial advice. Mortgage rates change daily, and the figures above reflect survey data as of early September 2026; your actual rate will depend on your credit, loan amount, location, and lender. Consult a licensed mortgage professional and a tax advisor before making decisions.
#MortgageRates
#30YearVs15Year
#ARMvsFixed
#HomeBuying2026
#PersonalFinance