Should You Lock Your Mortgage Rate or Wait? A Practical Decision Guide
Lock or Wait at 7.03%: What Each Path Costs on the Same Loan Over Five and Ten Years
The 30-year fixed average reached 7.03% in the week of September 24, 2026, a 52-week high. At that level the lock-or-float question stops being about a quarter point and becomes an arithmetic problem with four answers.
Updated September 25, 2026. Every rate, price and rule below was read that day from the agency or survey that published it, and each source is listed at the end with its as-of date. The scenarios are this article's own arithmetic on one stated loan, not a forecast and not personalised financial advice; mortgage rules cited are US federal and state practice varies.
Where the number actually is this week, and what it leaves out
Freddie Mac's Primary Mortgage Market Survey put the 30-year fixed average at 7.03% and the 15-year at 6.42% for the week of September 24, 2026. A year earlier the same survey read 6.30% and 5.49%. The 52-week low was 5.98%, on February 26, 2026, which means this week is the top of the range rather than a point inside it.
That figure is narrower than it looks. Since November 2022 the survey is no longer a poll of lenders; it is built from applications submitted to Freddie Mac through its automated underwriting system, covering conventional conforming purchase loans on owner-occupied single-family homes at 80% loan-to-value for borrowers with good to excellent credit. Freddie Mac also stopped publishing the accompanying fees and points. So 7.03% is a clean benchmark and not a quote: it describes a borrower with 20% down, conforming loan size and strong credit, with the cost of any points stripped out.
A two-week move that had nothing to do with the Fed
The clearest evidence about what drives this decision came from the calendar. Between September 3 and September 24, 2026 the 30-year average went 6.71%, 6.76%, 6.95%, 7.03% - a rise of 32 basis points. Across those same weeks the Federal Open Market Committee's target range did not move at all; it was set at 3.75% to 4.00% on September 16 and stayed there, with the effective federal funds rate at 3.88%.
What did move was the 10-year Treasury, from 4.79% on September 1 to 5.18% on September 24. The Federal Reserve's own explanatory material makes the mechanism explicit: "The rates charged on longer-term loans are related to expectations of how monetary policy and the broader economy will evolve over the duration of the loans, not just to the current level of the federal funds rate."
The spread between the mortgage average and the 10-year currently works out to 1.85 percentage points, which is this article's subtraction rather than a published statistic - no agency publishes a historical average for that spread, so any article quoting one has computed it, and you should ask over what period. The usable conclusion stands regardless: floating is a bet on the bond market, not on the Fed's next meeting.
Four paths on the same loan
Take the national median existing-home price of 429,100 dollars for August 2026 and put 20% down. The loan is 343,280 dollars over 30 years. Everything below is computed from the standard amortisation formula on that one loan, so the four rows differ only by the rate and by the upfront cost.
| Path | Monthly P and I | Interest, first 5 years | Interest, first 10 years | Balance after 10 years |
|---|---|---|---|---|
| A. Lock now at 7.03% | 2,290.77 | 117,405 | 226,397 | 294,784 |
| B. Float, and rates fall 50 bp to 6.53% | 2,176.54 | 108,771 | 209,143 | 291,238 |
| C. Float, and rates rise 50 bp to 7.53% | 2,407.32 | 126,058 | 243,745 | 298,146 |
| D. Lock at 7.03% and pay one point (3,433) for 6.78% | 2,233.36 | 113,085 + 3,433 upfront | 217,757 + 3,433 upfront | 293,034 |
Read the middle two columns against row A and the decision becomes symmetrical in a way most coverage hides. Being right about a 50 basis point fall saves about 8,600 dollars of interest over five years and about 17,300 over ten. Being wrong by the same 50 basis points costs about 8,700 over five years and about 17,300 over ten. The payoffs are near-identical in size, so floating is not a discount - it is a coin-flip with equal stakes on both faces, taken at the top of a 52-week range.
Row D is the one that surprises people. One point on this loan costs 3,433 dollars. If it buys 25 basis points, the net saving over five years is only about 890 dollars, and over ten years about 5,200. That is real money, but it is far smaller than the headline rate reduction suggests, and it is money you cannot get back if you sell or refinance early.
What a point has to buy before it pays for itself
Solving the same arithmetic backwards: on this loan, one point must reduce the rate by about 20 basis points to break even within five years, and by about 10 basis points to break even within ten. Anything less than that over your actual holding period is a transfer from you to the lender.
The regulator deliberately will not tell you what a point buys. The Consumer Financial Protection Bureau states only that "the amount that your interest rate is reduced depends on the specific lender, the kind of loan, and the overall mortgage market," and its April 2024 data spotlight puts it bluntly: "discount points have no fixed value in terms of the change in interest rate." That same analysis found the share of purchase borrowers paying points had risen from 30.5% in 2021 to 60.7% in 2023, with a median of one point among those who paid. When a majority of borrowers buy a product whose value the regulator refuses to quantify, the burden of arithmetic falls on the borrower, and the break-even above is the arithmetic to do.
The cost that is already sunk
February's 5.98% is worth calculating precisely, because it is the sum people carry around as regret and it is usually mis-sized. At 5.98% the same loan costs 2,053.73 a month against 2,290.77 today - a difference of 237.05 a month, or about 14,200 dollars over five years and 28,400 over ten.
That gap is genuinely large. It is also gone, and the only decision it should influence is the refinance plan, not the lock. Which leads to the awkward figure in the next section.
The house-price side of the same decision
Waiting is often framed as trading a higher rate for a lower purchase price. The published data does not currently support that trade. The FHFA House Price Index for the second quarter of 2026, released August 25, 2026, rose 2.1% year over year and 0.3% for the quarter, with the monthly seasonally adjusted index flat in June. The National Association of Realtors put the median existing-home price at 429,100 dollars in August 2026, up 1.6% year over year, with inventory at 4.9 months - which its chief economist called the highest in over ten years.
Prices are therefore rising at roughly a fifth of the borrowing rate. New construction is the one place prices are falling: Census and HUD reported a median new-house price of 393,700 dollars for August 2026, down 5.8% year over year, which is now below the existing-home median and reflects builder incentives. So the binding variable this autumn is the rate, not the price - and for anyone flexible on the type of house, the new-build side of the market is where the published discount actually is.
What the Fed's own projection says about waiting
The September 16, 2026 Summary of Economic Projections shows a median federal funds rate of 4.1% at the end of 2026 and 4.1% at the end of 2027, against a current target midpoint of 3.875%. On the committee's own median there is no easing in the forecast for either year.
That does not mean mortgage rates cannot fall - they follow the 10-year, and the same projections show PCE inflation dropping from 3.7% this year to 2.3% next. It does mean that "wait, the Fed will cut" is not a thesis the Fed's published numbers support. Anyone floating on that reasoning is relying on a rate path the policymakers themselves have not projected.
The costs that decide your break-even
Whatever you choose, the transaction costs set the horizon. Using the most recent official figures: the CFPB's analysis of 2023 mortgage data put median total loan costs for a home purchase at roughly 6,700 dollars, with median discount points around 3,000 dollars and more than 56% of originations paying some points. For a later refinance, the Federal Reserve's consumer guide - still its live publication, though dated August 27, 2008 - states that "it is not unusual to pay 3 percent to 6 percent of your outstanding principal in refinancing fees," and sets out the only break-even test that matters: closing costs divided by monthly savings equals months to recoup. Its worked example is 2,500 dollars over 91 dollars a month, or 27 months.
Apply that to path B above. If rates fall 50 basis points after you lock at 7.03%, refinancing saves 114.23 a month. At 3% of the balance, the fees on roughly 323,000 dollars would be about 9,700 dollars, which is 85 months to recoup - just over seven years. A 100 basis point fall roughly halves that. The practical implication is that the widely repeated "lock now, refinance later" advice only works on a large move, and a 50 basis point one is not large enough.
PMI, and the rules that end it
At 80% loan-to-value the scenario above avoids mortgage insurance entirely. Below that it applies, and the cancellation rules are federal and worth knowing exactly. The CFPB states you may request cancellation "on the date the principal balance of your mortgage is scheduled to fall to 80 percent of the original value of your home," that the servicer "must automatically terminate PMI" at 78%, and that it must end the month after the midpoint of the amortisation schedule. All three thresholds are measured against the original value, not a later appraisal, and they apply to mortgages closed on or after July 29, 1999.
No federal agency publishes an average PMI cost. The only official-side figure available is Fannie Mae's, which describes a typical range of 0.58% to 1.86% of the loan amount annually - explicitly a 2022 vintage. On a 343,280 dollar loan that spans roughly 1,990 to 6,385 dollars a year, a range wide enough that it deserves its own quote rather than an assumption.
Before you sign: what the lock does and does not promise
The CFPB's definition is precise and worth reading twice: your rate "won't change between the offer and closing, as long as you close within the specified time frame and there are no changes to your application." Locks "are typically available for 30, 45, or 60 days, and sometimes longer," extending one "may be expensive," and the agency states the downside plainly - a lock "may lock you out of a lower interest rate if rates fall after you get your loan offer." No official source publishes a typical extension fee or the cost of a float-down option, so both are questions to ask in writing before you agree.
One more contract term to check: the conforming limit. FHFA set the 2026 baseline at 832,750 dollars for a one-unit property, with a 1,249,125 dollar ceiling in high-cost areas, announced November 25, 2025. Crossing that line moves you into jumbo pricing, which the MBA survey put nine basis points above conforming this month.
Tip: A lock agreement, the loan estimate and the closing disclosure are the three documents you will need again at refinance time, so keep the originals together in a fireproof document bag rather than in a drawer. (These are Amazon Associate links - we may earn a small commission on qualifying purchases.)
Weighing it on the published numbers
Set the forecasts aside and only the arithmetic remains, and the arithmetic leans one way. The gain from floating and being right is almost exactly the same size as the loss from floating and being wrong, the current rate sits at the top of its 52-week range rather than the middle, the Fed's own median projects no policy easing through 2027, and the price side of the trade is appreciating at 1.6% to 2.1% a year, which does not repay a rate bet. Against that, one point is a poor instrument unless it buys more than 20 basis points and you intend to keep the loan past five years, and the refinance escape hatch needs a move closer to 100 basis points than 50 before its own break-even clears.
The defensible position on those numbers is to lock when your file is complete, to write the extension terms down before agreeing, to treat points as a separate calculation with your own holding period in it, and to get quotes from at least three lenders, which the CFPB still recommends even though it has dropped the dollar figure from its current page. Its archived 2018 estimate of what failing to shop costs - "an extra 300 dollars per year" and "9,000 dollars over a 30-year mortgage" - is old, but it is the same order of magnitude as the entire 50 basis point bet above, and unlike the bet it does not require being right about anything.
A lock-week checklist with numbers on it
- Get the rate and the APR on the same offer from three lenders on the same day. The CFPB defines the difference: the rate excludes fees, the APR "reflects the interest rate, any points, mortgage broker fees, and other charges."
- Ask each lender, in writing, how many basis points one point buys. Compare it with 20 basis points, the five-year break-even on a 343,280 dollar loan.
- Ask the lock length, the extension fee and whether any float-down exists. None of these are published anywhere official.
- Check your loan size against the 832,750 dollar conforming baseline before assuming conforming pricing.
- If you are below 20% down, ask when the balance is scheduled to hit 80% and 78% of the original value, and write both dates in your calendar.
- If a refinance is the plan, divide the expected closing costs by the expected monthly saving before you count on it. Anything past about 60 months is a plan, not a hedge.
Figures in this article are current as of September 25, 2026 and change weekly. It is general information, not personalised financial, tax or legal advice, and the deduction limits mentioned reflect IRS Publication 936 for the 2025 tax year, under which acquisition debt incurred after December 15, 2017 is limited to 750,000 dollars, and points are deductible in the year paid only when a specific set of conditions is met. Consult a licensed professional about your own transaction.
The agency and survey pages behind each figure
- Freddie Mac, Primary Mortgage Market Survey and its weekly archive - the 7.03% and 6.42% averages for September 24, 2026, the year-earlier readings and the February 26, 2026 low.
- PMMS methodology - the November 2022 change and the discontinuation of published points.
- Mortgage Bankers Association weekly applications survey - week ending September 18, 2026, contract rates and points.
- US Treasury daily yield curve and Federal Reserve H.15 - the 10-year yields and the effective funds rate.
- FOMC Summary of Economic Projections, September 16, 2026 - the 4.1% medians for 2026 and 2027.
- CFPB on rate locks, on discount points, on rate versus APR and on cancelling PMI.
- CFPB data spotlight on discount points, April 5, 2024, and the bureau's 2023 mortgage market report for median loan costs.
- Federal Reserve, A Consumer's Guide to Mortgage Refinancings - the 3% to 6% fee range and the break-even method.
- FHFA 2026 conforming loan limits and the House Price Index for the second quarter of 2026.
- NAR existing-home sales, August 2026, and Census and HUD new residential sales, released September 24, 2026.
- IRS Publication 936 - mortgage interest and points, 2025 revision.
Comments
Post a Comment