On October 14, the Bureau of Labor Statistics said prices rose 0.6% in September and 3.7% over the past year. That one number quietly rewrote the math on I bonds for November 2026, and most of the forecasts you’ll find online haven’t caught up. The widely quoted “about 4.9%” was built on a guess about September inflation. The real September number puts the likely new rate closer to 5.4%.
So the classic autumn advice, “buy I bonds before the rate resets on November 1,” is backwards this year. What you need to do before November is set up a login, because TreasuryDirect switches to ID.me only on October 28. And if this is money you’ll need back in a year, a plain CD still beats the bond, even at 5.4%.
The rate for I bonds in November 2026 is already mostly known
An I bond’s rate has two parts. The fixed rate is set when you buy and stays with the bond for up to 30 years. The inflation rate resets every six months based on CPI-U. Treasury’s published formula is fixed rate + (2 × semiannual inflation) + (fixed × semiannual inflation). Bonds bought from May through October 2026 earn 4.26%: a 0.90% fixed rate plus 3.34% of annualized inflation.
The inflation half of November’s rate is just arithmetic, and the inputs are public. It uses the change in CPI-U from March to September. March 2026 was 330.213, and the BLS September release put September at 336.940. That’s a 2.04% six-month increase, or 4.08% annualized. Treasury won’t announce it until the start of November, but it can’t come out any other way.
The fixed rate is the part nobody knows for sure. Treasury doesn’t publish its formula. David Enna of Tipswatch, who has tracked these resets for years, ties it to the 5-year TIPS real yield, which climbed from 1.46% on January 1 to 2.55% by mid-September. His best estimate is 1.30%, possibly 1.40%, and he puts the odds of anything below 1.30% near zero. Plug 1.30% and 4.08% into Treasury’s formula and you get 5.41%. At 1.40% it’s 5.51%.
The 4.9% figure is stale for a simple reason. Enna’s September forecast, and the Detroit Free Press column syndicated on October 16, assumed September inflation of about 0.35%, which gave an inflation component near 3.58%. September came in at 0.6%. Same fixed rate, half a point more inflation, and that’s the difference between “about 4.9%” and about 5.4%.
Why waiting two weeks beats buying in October
The fixed rate is the reason not to rush. Whatever it is on the day you buy, you keep it for as long as you hold the bond. October buyers lock in 0.90% forever. November buyers probably lock in 1.30%.
Take $10,000, the most one person can buy electronically in a calendar year, and run both dates.
If you buy in late October, your first six months earn 4.26%, which is $213. Your next six months use November’s inflation number with your 0.90% fixed rate, so 5.00%. On a balance of $10,213 that’s about $255. Year one comes to roughly $468.
If you buy in late November instead, your first six months earn about 5.41%, which is $270. For the next six months, assume inflation keeps the same pace (a big if, which I’ll come back to). You’d earn 5.41% again on $10,270, or about $278. Year one comes to roughly $548.
That’s about $80 more in the first year, and the gap doesn’t close. A 0.40-point fixed-rate advantage on $10,000 is about $40 a year for every year you hold, before compounding. If inflation drops to zero, the October bond pays 0.90% and the November bond pays 1.30%. Inflation hits both bonds the same way. The fixed rate is the only part you pick.
There’s also a small trick that works in your favor in November. Treasury credits interest from the first day of the month you buy. If you buy on November 24 instead of November 2, you get a full month of interest anyway and your cash sits in savings for three more weeks.
The deadline that matters is October 28, not October 31
This is the part I’d put on your calendar today. Starting October 28, 2026, TreasuryDirect drops its old username, password, and security image login. According to Treasury’s own email to account holders, ID.me becomes the only way in.
If you haven’t linked ID.me by then, your money isn’t frozen or seized. Treasury says that directly. But online access is restricted until you finish verifying, which means you can’t buy a November bond, redeem an old one, or download your 1099. Verifying your identity is a ten-minute chore on a quiet weeknight in October. It’s a much worse chore the week you’re trying to move $10,000.
A few details catch couples in particular. Each spouse needs a separate ID.me wallet tied to a separate email address. The name on the wallet has to match the TreasuryDirect account name exactly, so if you changed your name after marriage and only updated one of them, sort that out first. If you manage an account for a parent or a trust, you verify yourself first and then get one login for everything.
So the order I’d follow is: link ID.me this week, wait for the announcement in early November, then buy toward the end of the month.
When a 5% CD beats a 5.4% I bond
Most “I bonds are back” headlines skip this part. You can’t cash an I bond for 12 months. If you cash it before five years, you give up the last three months of interest. Treasury’s own example: cash out at 18 months and you get 15 months of interest.
Run the one-year version. The November bond earns about $548 over 12 months, but you’d lose roughly the last three months, about $139. That leaves around $409. Meanwhile, as of October 14, DepositQuest’s rate survey found one-year CDs at 5.00% APY at top credit unions and banks, which is $500 on $10,000. I bond interest is exempt from state income tax, so give the CD a 5% state tax haircut and it’s still about $475. For one-year money, the CD wins by around $65 and you can see the whole year’s rate on day one.
So the I bond is for a specific kind of money. Five-year money, something like the second half of an emergency fund or savings for a goal that’s a few years away, where you want protection if inflation spikes again. For the first few months of your emergency fund, you want instant access, and a high-yield savings account does that job better.
Two more caveats. First, the fixed rate is still a forecast until Treasury posts it. If it surprises low, the case for waiting gets weaker, but even at a 0.90% fixed rate a November bond starts at about 5.00%, which beats October’s 4.26%. Second, the same Tipswatch analysis notes that TIPS currently pay a higher real yield than the I bond’s likely fixed rate. TIPS are taxed every year outside a retirement account and their market prices move, so they’re a different tool, but if you’re doing this for the long haul it’s worth comparing.
I’m not a financial advisor, and none of this is a recommendation to buy anything. It’s the mechanism and the math, and for I bonds in November 2026 the math is lopsided: an October purchase locks in a 0.90% fixed rate for decades to save a couple of weeks. The deadline that can cost you is the October 28 login switch. Set up ID.me now, let the November 1 numbers post, and buy late in the month if this is money you can leave alone for five years.