Key takeaways
- The 2s10s spread closed at +45bp on 30 July 2026, after +35bp on 10 July. It has been continuously positive since 6 September 2024 — the un-inversion is nearly two years old, not news.
- The curve flattened 44bp between 30 January and 30 June, then steepened 15bp through 30 July. Both legs were bear moves: every maturity sold off, and only the order changed.
- At 4.23% on 30 July, the 2-year sat 48bp above the top of the Fed's 3.50-3.75% target range. The FOMC held that range on 29 July by 9 votes to 3, all three dissents favouring an immediate hike.
- The 30-year at 5.21% pays 53bp over the 10-year at 4.68%, but at a modified duration of about 15.1 that entire pickup is erased by a 3.5bp rise in the 30-year yield.
The curve un-inverted in 2024, and nothing since has changed that
The 10-year minus 2-year Treasury spread last closed negative on 5 September 2024, at -2bp. It has printed positive on every trading day since 6 September 2024. Any account of the un-inversion as a 2026 event is describing something that happened in September 2024.
Every yield here is a constant-maturity close as at 30 July 2026. The spread stood at +45bp, on a 2-year of 4.23%, a 10-year of 4.68% and a 30-year of 5.21%. The 10-year TIPS yield was 2.41%, putting the breakeven at 2.27%. Three weeks earlier, at the 10 July close, the spread was +35bp.
It flattened for five months and then steepened in one
The chart plots month-end 2s10s closes through 2026. From 74bp on 30 January the spread compressed to 30bp on 30 June — 44bp of flattening in five months, with an intraseries low of 27bp on 18 June — and then widened back out to 45bp by 30 July.
Decompose both legs, because the spread hides which one moved. Between 30 January and 30 June the 2-year yield rose 62bp (3.52% to 4.14%), the 10-year rose 18bp (4.26% to 4.44%) and the 30-year rose 4bp (4.87% to 4.91%). Every point sold off and the front sold off hardest: a bear flattener. Between 30 June and 30 July the order reversed — the 2-year rose 9bp, the 10-year 24bp and the 30-year 30bp — which is a bear steepener. What did not change is the sign of any leg. All three maturities ended July higher than they stood on 30 January.
The 2-year is pricing tightening the Fed has not delivered
The federal funds target range stood at 3.50-3.75% on 30 July 2026, where it has been since 11 December 2025. At the 30 July close the 2-year note yielded 4.23% — 48bp above the top of the range, and 73bp above the bottom, against 41bp above the top on 9 July. A 2-year yield materially above the policy rate is a market expecting the average overnight rate across the next two years to exceed today's rate. Read literally, the front end has priced out easing and begun pricing the reverse.
The committee is no longer unanimous about that either. At its meeting on 29 July 2026 the FOMC held the range at 3.50-3.75% by 9 votes to 3, with Beth Hammack, Neel Kashkari and Lorie Logan each dissenting in favour of raising it by a quarter point at that meeting. Three dissents in one direction is a formal record that part of the committee now reads the data the way the front end does.
This is where the shape becomes legible. A bull flattener — long yields falling toward stable short yields — is a growth-scare signal. A bear flattener, with the front end leading a general sell-off, is an inflation-and-policy signal. The two look identical in a spread chart and mean opposite things, which is why the spread alone is an inadequate summary and why the level and direction of each leg has to be read separately. July was a third case again: the front end steadied near its highs while the long end kept selling off, which is more consistent with a term-premium move than with a change in the expected policy path — though daily yield data cannot separate the two.
Carry does not compensate for duration at this curve shape
The long end pays, in headline terms. Extending from the 10-year at 4.68% to the 30-year at 5.21% picks up 53bp of yield. But a par 30-year at 5.21% carries a modified duration of about 15.1 (a par 10-year at 4.68% is about 7.9, and a par 2-year at 4.23% about 1.9). Fifty-three basis points of extra yield is therefore wiped out in price by a 3.5bp rise in the 30-year yield — 0.53 divided by 15.1.
That is the sizing arithmetic in one line, and July supplied the demonstration. The 30-year yield rose 30bp in that month alone, roughly nine times the buffer its entire yield advantage provides, and 34bp from its 30 January close. The extra 53bp buys very little in risk terms; what it buys is convexity in a rally, and exposure to roughly twice the 10-year's duration in a sell-off.
What this framework cannot tell you
The 2s10s spread is the difference between two constant-maturity par yields. It is not a forward rate, it embeds a term premium that is not separately observable, and its reputation as a recession indicator rests entirely on episodes of inversion — of which there is currently none, so the indicator is silent rather than reassuring. The duration figures above assume par bonds at 30 July yields, not any specific security. The 2.27% breakeven includes an inflation risk premium; it is not a measurement of expected inflation. If the front end reverses and the 2-year falls back below the funds rate, the bear reading in this piece is wrong and the curve's shape becomes a growth signal instead.
The level to watch is not the 2s10s at all. It is the 2-year against the top of the target range. That gap — 48bp at the 30 July close, against 41bp on 9 July — is the cleanest available statement of how far the market has moved away from the Fed, and it will invert this argument long before the spread chart does. It is also the mirror image of what has happened in the euro area, where the central bank has already hiked in response to an energy shock, and it is the input that quietly reprices every long-duration asset, from AI semiconductors downward.