Damiana Data
Treasury yields · public data · Sep 17, 2026

The longest inversion in this data

For over 2 years, the U.S. paid more to borrow short than long.

69-second video. Every figure in it comes from the checked story data.

What the video says

  1. For over 2 years, the U.S. paid more to borrow short than long.

    The longest run like it in this data, which starts in 1990. It has since ended.

  2. Every trading day since 1990: 9,183 yield curves.

    Each slice runs from short maturities in front to long ones behind. Height is the yield.

  3. Most of the time, lending for longer pays more.

    On 88.3% of these days, the 10‑year yield was above the 2‑year.

  4. Then the curve flipped.

    Jul 6, 2022 to Aug 26, 2024: 537 trading days inverted, the longest run in this data.

    Deepest: −1.08 points on Jul 3, 2023.

  5. The next longest since 1990: 220 days, in 2000.

    This one lasted 2.4× as long.

  6. Measured against the 3‑month bill, the same stretch ran 534 days and went deeper.

    Down to −1.89 points.

  7. On Sep 16, 2026, the 10‑year yield sits above the 2‑year again.

    10‑year: 5.01%. 2‑year: 4.74%.

  8. This data starts in 1990. Earlier inversions aren't in it.

    And nothing here tests what an inversion predicts. What inversions predict: MISSING. That test is next.

  9. Source: U.S. Department of the Treasury, Daily Treasury Par Yield Curve Rates.

    Par yields as published. Treasury changed its curve-fitting method on Dec 6, 2021. Not investment advice.

Evidence behind it

  • Confirmed: measured and verifiedPublished par yieldsCONFIRMED
  • Confirmed: measured and verifiedInversion episodesCONFIRMED
  • Missing: not available or not testedWhat inversions predictMISSING

Marks open only as wide as each input’s evidence class allows. See Method.

Method

Source
U.S. Department of the Treasury, Daily Treasury Par Yield Curve Rates (public domain)
spread 10y 2y
10 yr minus 2 yr par yield, percentage points
spread 10y 3m
10 yr minus 3 mo par yield, percentage points
inversion episode
consecutive observation days with the spread below zero; not merged across positive days
big move
one-day change of 0.5 percentage points or more in any maturity (kept, flagged)
Scope
longest means longest in data starting in the first year, never ever
Scope
no claim about what inversions predict or what followed them

Provenance

resolved story sha256
3fb7ff20a82097dab62e5e615a89522f286b57f89cc7b8b22a53beac2802f631
rendered video sha256
626f33efb32a515d5484d7a77a7c8e6f905dd6ca5f862a880faabe9f237ac3a6
web video sha256
351b2497114793a76cc044926df85916b7f136f5966d6954a245f89a63c583e5
curve sha256
ec58e2bc31a828b220d6f7e6db4e40d94c092e5893c33e1330ea9c83b61317bd
treasury snapshot
20260917T134036Z
treasury files · par-yield-curve-rates-1990-2023.csv
004461bcf87ec7b34daf035f05fc191d7cf2cd84986533df3e39d5dcdb0d62a4
treasury files · par-yield-curve-rates-2024.csv
6775840b76cd7fd41121f61d6f6c60184eca6b2ba49954ec7372c463cbeef34d
treasury files · par-yield-curve-rates-2025.csv
cf2fa3da7f160384b63d2ead698532b1d76cd9f789ce4911e2304db85b14e6d5
treasury files · par-yield-curve-rates-2026.csv
7dd1adf5bb157e17789c67ac94404b246bfb337919a8e10761be3e52e79a4822

Backtests and model estimates on historical data. Not investment advice.