Why Are Treasury Yields Falling? Thursday’s 16bp 2-Year Drop

FOMC Alerts · October 23, 2026

Why are Treasury yields falling?

Why are Treasury yields falling? On Thursday the short answer was stocks. A tech-led selloff, with the Nasdaq down 2.60%, sent money into government bonds, and yields dropped across the curve. The 2-year Treasury yield fell 16bp to 4.65% and the 10-year fell 17bp to 4.96%, its first close under 5% since Sep 22 (U.S. Treasury par curve).

Here is the odd part. The 2-year is the maturity that usually tracks where traders think the Fed is headed, as Reuters noted in its coverage. Yet fed funds futures for the next two meetings barely moved: the implied rate for the end of December slipped only about 4bp. Bonds priced a lower rate path for 2027, not a different decision next Wednesday. Below: the curve by maturity, the gap between bonds and Fed odds, what drove the bid, and the oil signal hiding underneath.

Thursday’s curve, maturity by maturity

Treasury par yield curve, closing levels. Source: U.S. Treasury.

The move was close to parallel. Everything from the 2-year to the 10-year fell 16-17bp, so the 2s10s spread barely changed (32bp to 31bp). The 10-year’s 4.96% was its lowest close since Sep 22; its 2026 closing high was 5.31% on Oct 5. The 2-year’s 4.65% was its lowest since Sep 14. Only the bill end held firm: the 3-month fell just 3bp, consistent with a Fed that isn’t expected to move on Oct 28. Some of it came back Friday: at 11:50 AM ET the 2-year traded at about 4.70% and the 10-year near 4.99% (Tradeweb, via CNBC).

The gap: 16bp in the 2-year, about 4bp in Fed pricing

By our calculation using CME’s method on Thursday’s fed funds futures settlements, the chance of a hike on Oct 28 fell to about 5.5%, from about 13% on Wednesday’s settles. The chance that the target range sits above today’s 3.75-4.00% after the December meeting fell to about 80%, from 89%, after dipping to about 72% around 2 PM. The implied end-December rate went from 4.133% to 4.092%. Even the January 2027 contract moved only about 5.5bp. (How we turn futures prices into odds, and why different readings disagree, is in our guide to fed funds futures probability.) On Friday, using delayed quotes as of about 11:40 AM ET, the same calculation put an October hike at about 10% and December above the range at about 81%.

Prediction markets told the same story late Thursday. On Kalshi, an Oct 28 hike traded at 5-6 cents (11 cents the day before) and a December hike at 71-72 cents. Polymarket showed nearly identical prices. By about 11:50 AM ET Friday, both prediction markets had an October hike at 7-8 cents and a 25bp December hike at 71-75 cents. On Thursday, CME’s FedWatch tool put October hike odds at 10.5%, down from 27.7% a week earlier and 55.4% a month earlier, according to Reuters. Different times and methods, same direction: October was already mostly off the table, and December mostly held.

So the 2-year’s 16bp is not a bet on the next meeting. It’s the market trimming how high, and for how long, rates stay through 2027. On a typical FOMC day the 2-year moves about 4bp (median, our calculation, 93 meetings since 2015). Thursday, with no Fed decision, moved it four times that.

What drove the bid: AI spending and a flight to quality

The selloff centered on the cost of the AI buildout. The S&P 500 fell 1.21% to 7,663.43 and the Nasdaq dropped 2.60%, its worst day since Jun 5, while the Dow finished flat. Meta fell 6.98% after its third-quarter results. Its release showed revenue of $64.60 billion, up 26%, but free cash flow was negative $2.12 billion as capital spending, including finance-lease principal, reached $34.91 billion in the quarter. Meta narrowed its 2026 capex outlook to $135-145 billion, from $130-145 billion (Meta).

Equipment makers flagged limits on how fast the buildout can go. On its Oct 21 earnings call, Lam Research said: “As we move into calendar year 2027, we see clean room space availability and supply chain constraints for certain components continuing to limit chip makers and equipment companies’ ability to fully meet end demand.” (Lam Research)

Bond buyers took the other side of the stock selling. “The flight-to-quality trade is winning in the Treasury space,” Thomas Urano, co-chief investment officer at Sage Advisory, told Reuters, after a stretch in which energy and inflation worries had competed with safe-haven demand. That’s the textbook hedge working, which hasn’t been a given this year; we covered how bonds stopped hedging stocks in 2026.

It wasn’t a data story. Weekly jobless claims fell 6,000 to 194,000, slightly better than expected (Labor Department). There was no weak labor number behind the drop; equities did the work.

Growth fear in real yields, oil in breakevens

Split the 5-year yield into its parts and the day looks different. The 5-year real yield (TIPS) fell 22bp to 2.33%, more than the 17bp drop in the nominal 5-year. That means the 5-year breakeven, the market’s inflation compensation, rose to 2.45% from 2.40% (FRED). The 10-year breakeven went the other way, down to 2.32% from 2.37%. Households lean the other way too: in Friday’s final October survey from the University of Michigan, year-ahead inflation expectations rose to 4.7% from 4.6% and the 5-10 year outlook to 3.5% from 3.4%, both the highest since May.

Read it as two forces at once. The real-yield drop is the growth scare. The firmer 5-year breakeven fits an oil market that hasn’t backed off: December WTI settled at $99.47 on Thursday, up 2.4% (Yahoo Finance). The New York Times reported that the U.S. and Israel are preparing for renewed strikes on Iran after the midterm elections, per a syndicated summary on Investing.com. That’s a report, not a confirmed plan. Urano’s point to Reuters was that the safe-haven side won this round. The oil side is still there.

How rare is a 16bp drop in the 2-year?

Thursday’s move was the 2-year’s biggest one-day drop since Aug 1, 2025 (-25bp), and the 10-year’s biggest since Aug 2, 2024 (-19bp), by our calculation from Treasury data. Since 2022 there have been 27 days when the 2-year fell 15bp or more; only five of them came since the start of 2024.

Big drops don’t decide the next Fed meeting. On Mar 13, 2023, during the bank failures, the 2-year fell 57bp in a day. The Fed still raised rates by 25bp nine days later, on Mar 22, 2023. The sample is small, and one day of bond buying is not a policy signal.

What to watch

• Wed Oct 28, 2:00 PM ET: FOMC statement; the chair’s press conference at 2:30 PM ET. No new projections at this meeting. The Fed raised rates to 3.75-4.00% on Sep 16.
• Thu Oct 29, 8:30 AM ET: Q3 GDP and September PCE inflation, the day after the Fed.
• Dec 8-9: the next FOMC meeting, with projections. That’s where most of the hike pricing sits.
• Tue Nov 10, 8:30 AM ET: October CPI.
• The 5-year breakeven and December WTI: if oil keeps pushing inflation compensation up while real yields fall, the bond rally has a ceiling.

What it means for 0DTE

• Energy: Thursday was earnings and positioning, not scheduled event risk. The scheduled energy is Oct 28 at 2:00 PM ET and GDP/PCE the next morning.
• Premium: the VIX closed at 19.03, up 14% from 16.68, after touching 20.31. Here’s how the VIX curve prices the Fed, GDP and Big Tech week; let it set what premium costs.
• Structure: a 2.6% Nasdaq day on no economic data is a reminder that ranges can widen without a calendar event. Keep max loss known before entry and size for gaps.
• Plan, don’t predict: prepare for two scenarios into Oct 28, a bond bid that keeps leaning on rate odds and an oil-driven rebound in breakevens. Neither is a direction to bet.

Related reading

• The Fed Sets the Short End. Who Sets the 10-Year?: why the 10-year was rising before Thursday.
• The Fed Hiked. Your Savings Account Didn’t Notice.: what the hike path means for CD and savings rates.
• FOMC Alerts hub: the Fed calendar and all our coverage in one place.
• All FOMC Alerts posts: the full series, newest first.

Trade it with structure

Observer is a six-week trial and training program: train with Coach for $17/week. You get everything a full coaching member gets, every week, and the six weeks are designed to install pro trading habits: courses on strategy, process and risk management, live daily coaching, a Discord trade room with hundreds of experienced traders, and the platform (heatmap, GEX tools, risk graph, position analyzer and journal). Start Observer.

Educational content, not investment advice. Options involve risk and are not suitable for all investors.