Market data · The 10-year minus 2-year spread since June 1976
Yield curve inversion, 8 times since 1976
The 10-year Treasury note yields +0.41 percentage points more than the 2-year note at the 8 September 2026 close, so the curve is not inverted. A yield curve inversion is that spread closing below zero, and it has happened 8 times since June 1976. A recession followed 5 of them, between 10 to 23 months later.
12,564 daily closes since 1 June 1976 2,049 of them below zero, or 16.3% Rebuilt from FRED, retrieved 2026-09-09
10-year minus 2-year Treasury spread
+0.41Percentage points at the 8 September 2026 close. Below zero is an inversion
Where that sits on 50 years
Read this before the chart
Eight inversions in 50 years is a small sample, and five of them were followed by a recession. Anything computed on five observations describes what happened rather than what is likely, which is why the limits section is longer on this page than on most.
The numbers
Is the yield curve inverted right now? No
Every figure here is computed at build time from three files that do not share an end date, so none of them is presented as though they do. The spread runs to 8 September 2026, the recession indicator to August 2026 and the London gold closes to August 2026. Nothing on this page pairs a figure from one series with a date from another.
10-year minus 2-year Treasury spread
+0.41 pts8 September 2026 close
The curve is not inverted, because a 10-year Treasury note pays +0.41 points more than a 2-year one. An inversion is this number closing below zero.
Inversions since June 1976
82,049 of 12,564 trading days below zero
Episodes rather than days, because the 2,049 days below zero came from 8 runs and months inside one run share a cause. The rule that separates them is stated in the method section.
Followed by a recession
5 of 8within 24 months of the curve coming back
Recession dates come from the National Bureau of Economic Research, which is the body that dates them, and this page reads them off the indicator rather than typing them in. The March 1990 episode pointed at a downturn an earlier inversion had already signalled, so it is not counted twice.
Lead time to the recession
10 to 23 monthsmedian 17 months, on 5 observations
Measured from the first day the curve closed below zero to the month the recession began. The spread between the shortest and the longest is 13 months, which is the reason this signal cannot be traded on a calendar.
Gold across an inversion
−0.3% median3 of 8 episodes finished higher
London gold closes from the month the curve inverted to the month it came back. The median is close to nothing, and the two large moves either side of it are the same 1980 spike measured from opposite ends.
Sources: Federal Reserve Bank of St. Louis, 10-Year minus 2-Year Treasury Constant Maturity spread. Federal Reserve Bank of St. Louis, NBER based recession indicator for the United States. LBMA Gold Price (PM London auction, USD/oz). Retrieved 2026-09-09
The long record
Yield curve inversion chart since 1976
Every daily close of the 10-year minus 2-year spread since 1 June 1976, on a linear axis with a solid rule drawn at zero. The rule is the whole chart: everything below it is an inversion. Turn on the shading to see the 6 recessions the National Bureau of Economic Research has dated inside this window.
10-year minus 2-year Treasury constant maturity spread, daily
The spread and every US recession since 1976
8 September 2026
+0.41Not inverted
The dashed rule is the +0.85 average across all 12,564 closes and the solid rule is zero. The deepest close in the record is −2.41 on 20 March 1980, during the inversion that ran into the January 1980 recession, and the steepest is +2.91 on 4 February 2011.
Inversions and recessions on one axis
Source: Federal Reserve Bank of St. Louis, 10-Year minus 2-Year Treasury Constant Maturity spread and Federal Reserve Bank of St. Louis, NBER based recession indicator for the United States. Retrieved 2026-09-09
Download the full dataset as CSVDeepest close inside each inversion, percentage points
How deep did each yield curve inversion go?
Depth and consequence do not line up the way most accounts of this indicator imply. The −0.14 dip of March 1990 lasted 22 days and the −0.19 inversion that began in January 2006 was the shallowest of the three that preceded a recession, while the deepest reading in the record came 17 months before the January 1980 downturn. Red bars are the episodes a recession followed.
What it means
What does a yield curve inversion mean?
An inversion is a statement about interest rates rather than about the economy, and the step from one to the other is where most of the confidence in this indicator quietly comes from.
Lending the US government money for ten years normally pays more than lending it for two, because a decade is longer to wait and more can go wrong inside it. When that reverses, the bond market is saying it expects short-term rates to be meaningfully lower in a few years than they are today. Short-term rates are set by the Federal Reserve, and the Federal Reserve cuts when growth and employment are weakening, so the inversion is really a forecast of rate cuts that has been read for fifty years as a forecast of the slowdown that causes them.
That reading has a record behind it. Since June 1976 the spread has closed below zero in 8 separate stretches, and a recession followed 5 of them within 24 months of the curve returning above zero. Turned around, 5 of the 6 recessions dated inside this window had an inversion in front of them.
The exception is worth naming, because the usual version of this claim says every recession. February 2020 arrived with no qualifying inversion in front of it on this series. What preceded it was the August 2019 episode, three closes at four basis points, which is below the threshold this page uses and is set out in full further down. Anyone who counts those three days gets six from six, and the difference between those two records is a rule rather than a fact.
Whether any of this should change what your retirement account holds depends on your own timeline rather than on the chart. See which gold IRA companies fit your account
Where it stops being useful is the timing. Measured from the first day the curve closed below zero to the month the recession began, the lead ran 10 months at the short end and 23 months at the long end, with a median of 17 months. Since a normal expansion runs several years, being told that a recession is more likely at some point in the next two years is a weaker claim than it sounds, and it is the strongest claim this record supports.
The un-inversion is worth watching for a different reason. Measured from the day the spread closed back above zero, the same 5 recessions arrived anywhere from 12 months before the curve came back to 8 months after it. Twice the downturn had already started while the curve was still inverted, because the Federal Reserve was already cutting by then and cutting steepens the curve. A curve that has just un-inverted is not evidence that the risk has passed.
The reason this page carries a threshold
August 2019 does not count as an inversion here
The 2019 inversion led the financial news for a week and on this series it was three closes. The spread finished at −0.04, −0.03 and −0.03 on 27, 28 and 29 August and never went below zero again that year. Under the rule this page uses, which needs 5 consecutive closes below zero to open an episode, it does not qualify. Other spreads tell a different story, and the 10-year minus 3-month curve was inverted for months either side of it, which is a fair reason to prefer that measure and not a reason to report three days as an inversion of this one.
The record
All 8 yield curve inversions since 1976
One row per episode, computed from the daily closes rather than assembled from a list. The recession column is derived from the National Bureau of Economic Research indicator, and the gold columns are London closes matched by calendar month. An n/a means that window has no second end yet, which is a different thing from a zero.
| Inverted | Days | Deepest | Recession began | Lead | Gold during | Gold in recession |
|---|---|---|---|---|---|---|
| 18 August 1978 to 2 May 1980 | 623 | −2.41 Mar 1980 | January 1980 | 17 mo | +156.6% | −5.9% |
| 12 September 1980 to 19 July 1982 | 675 | −1.70 Dec 1980 | July 1981 | 10 mo | −48.6% | +7.4% |
| 13 December 1988 to 7 November 1989 | 329 | −0.45 Mar 1989 | July 1990 | 19 mo | −0.5% | −4.5% |
| 8 March 1990 to 30 March 1990 | 22 | −0.14 Mar 1990 | July 1990, counted above | n/a | +0.0% | n/a |
| 15 June 1998 to 28 July 1998 | 43 | −0.07 Jun 1998 | None dated | n/a | −2.5% | n/a |
| 2 February 2000 to 29 December 2000 | 331 | −0.52 Apr 2000 | March 2001 | 13 mo | −6.5% | +6.9% |
| 31 January 2006 to 6 June 2007 | 491 | −0.19 Nov 2006 | December 2007 | 23 mo | +14.4% | +12.1% |
| 6 July 2022 to 6 September 2024 | 793 | −1.08 Jul 2023 | None dated | n/a | +50.0% | n/a |
Lead is measured from the first day below zero to the month the recession began. Gold during runs from the month the curve inverted to the month it came back, and gold in recession from the month the recession began to the month it ended. Sources: Federal Reserve Bank of St. Louis, 10-Year minus 2-Year Treasury Constant Maturity spread. Federal Reserve Bank of St. Louis, NBER based recession indicator for the United States. LBMA Gold Price (PM London auction, USD/oz). Retrieved 2026-09-09
Median change in the London gold price, three windows
Gold's median move across an inversion was −0.3%
Three windows rather than one, because a single column invites the wrong reading. Across the 8 inversions themselves the median is −0.3% and 3 finished higher, which is close enough to nothing that the two extremes of +156.6% and −48.6% turn out to be the same 1980 spike measured from either side of its top. Through the 5 recessions that followed, gold's median was +6.9% inside a range of −5.9% to +12.1%.
London gold closes matched to each window by calendar month, never by an exact-date join against the daily spread. Source: LBMA Gold Price (PM London auction, USD/oz). Retrieved 2026-09-09
Longest inversion on record
793 days6 July 2022 to 6 September 2024, bottoming at −1.08 on 3 July 2023
Deepest inversion on record
−2.41 pts20 March 1980, inside the episode that began 18 August 1978 and preceded the January 1980 recession
Since the last inversion ended
24 monthsThe curve closed back above zero on 6 September 2024 and no recession has been dated since. The longest any previous inversion waited after coming back is 8 months
Method
How this page counted 8 yield curve inversions
Three series, one rule, and a threshold that changes the answer. All three are named below, and the threshold is published with its sensitivity rather than presented as the only way to count.
The spread is FRED series T10Y2Y, the 10-year Treasury constant maturity yield minus the 2-year, published every trading day since 1 June 1976. Constant maturity means the Treasury interpolates a yield for an exact 10-year and 2-year point rather than quoting whatever bond happens to be nearest, so the series is comparable across fifty years of changing issuance. No smoothing, no averaging and no weekly sampling is applied here.
An inversion opens on the 5th consecutive close below zero and closes on the 63rd consecutive close at or above it, which is a trading week to open and a trading quarter to shut. The two thresholds differ on purpose. A curve that spends three days below zero has not inverted, and a curve that pops above zero for a fortnight in the middle of a two-year inversion has not un-inverted. The 2006 to 2007 episode crossed back above zero six times, once for ten weeks, and a symmetric rule reports it as five inversions rather than the one every account of it describes.
Recession dates come from FRED series USREC, which carries the National Bureau of Economic Research chronology as a monthly indicator. The indicator turns on for the period following the peak, so the peak month is the month before its first reading of 1, and this page derives both peak and trough from the file rather than typing eight dates in. Reading that convention backwards would shift every peak by a month and produce a page that looked entirely normal, which is why the test suite checks the derived chronology against the committee's published one.
Gold is the London monthly close, matched to each window by calendar month. The spread is daily and gold is monthly, and joining two series of different frequency on an exact date string keeps only the month-ends that land on a trading day, which drops roughly 40% of the record and looks like a valid narrow window rather than an error. Nothing on this page performs that join.
| Rule | Closes to open | Closes to shut | Inversions found |
|---|---|---|---|
| 2 in, 63 out | 2 | 63 | 9 |
| 5 in, 5 out | 5 | 5 | 15 |
| 5 in, 21 out | 5 | 21 | 12 |
| Used on this page | 5 | 63 | 8 |
| 5 in, 126 out | 5 | 126 | 6 |
| 21 in, 63 out | 21 | 63 | 6 |
The count ranges from 6 to 15 across these six settings, on the same 12,564 daily closes. Most of that spread is 2006 and 2007 splitting apart and reassembling. Anyone quoting a different number of inversions is probably using a different rule rather than different data, and the honest version of this table is the one that shows both.
Limits
What this yield curve inversion page does not show
Eight things worth knowing before any figure above gets used for a decision. Every one of them is a real constraint on the data or on the metal rather than a disclaimer.
-
The sample is eight
8 inversions in 50 years, of which 5 were followed by a recession. The 2,049 days below zero look like a large sample and they are not, because days inside one inversion share a cause and a starting point. Every median on this page rests on 5 to 8 observations.
-
The lead time is too wide to act on
10 months at the short end and 23 months at the long end, a spread of 13 months. Measured from the un-inversion instead, the recession arrived anywhere from 12 months before the curve came back to 8 months after it. A signal whose timing varies by more than a year tells you about a period, not a date.
-
An inversion is not a trade
Gold's median change across the 8 inversions was −0.3%, and 3 of them finished higher. The extremes of +156.6% and −48.6% are the 1980 spike measured from either side of its top rather than two separate lessons.
-
It has missed once, and one verdict is open
June 1998 inverted for 43 days and the next recession was not dated for another 32 months. The inversion that ended 6 September 2024 has now gone 24 months without one, which is longer than any gap in the record, where the recession arrived at most 8 months after the curve came back.
-
Recession dates arrive late
The National Bureau of Economic Research dates a peak after the fact, and it announced the February 2020 peak four months after it happened. A recession beginning this quarter would not appear in the indicator, or in the table above, for the better part of a year.
-
One curve, and only this curve
The 10-year minus 3-month spread inverts on a different schedule and would produce a different table, including a 2019 episode this one excludes. Neither spread says anything about why the curve moved, and a steepening driven by 10-year yields rising is a different event from one driven by 2-year yields falling.
-
Every gold figure prices spot metal
The gold columns are London closes, so they describe an ounce and not an account. A gold ETF tracks that line minus an expense ratio and a mining stock does not track it at all. Bullion needs a self-directed IRA, a custodian and a depository, and the metal has to clear the 99.5% purity rule before a custodian will accept it.
-
No fee, premium or tax is deducted
A dealer premium over spot runs 5% to 12% on IRA-eligible coins, and a setup fee, an annual custodian fee and a depository storage fee all sit on top of it. On a $50,000 purchase the gap between those two premiums alone is $3,500, which is larger than the median move in any gold column above.
How this page is built
Every figure is computed at build time from three committed data files, so the numbers are in the HTML before a script runs and the charts render with JavaScript turned off. Nothing on this page is a hand-typed constant, including the inversion dates, the recession dates and the gold returns, and a rebuild of the underlying series moves the prose with the table.
The test suite checks the derived recession chronology against the one the National Bureau of Economic Research publishes, rather than against another copy of itself. A suite that only compares a derivation to its own output agrees with a wrong input as readily as with a right one, which is how a single bad monthly figure once reached nine pages and a thousand passing tests.
The spread, the recession indicator and the gold closes were last updated from their publishers on 9 September 2026, and the spread itself is checked through 8 September 2026. Written by the Gold IRA Digest Editorial Team and last reviewed 9 September 2026. Corrections go to the editorial team and are logged publicly.
Sources
- FRED series T10Y2Y, 10-year minus 2-year Treasury constant maturity spread, daily. Retrieved 2026-09-09
- FRED series USREC, NBER recession indicator, monthly. Retrieved 2026-09-09
- National Bureau of Economic Research, US business cycle expansions and contractions
- LBMA precious metal prices, monthly gold closes. Retrieved 2026-09-09
- FRED series T10Y3M, the 10-year minus 3-month spread this page does not use
- Board of Governors of the Federal Reserve System, H.15 selected interest rates, the constant maturity series T10Y2Y is derived from
- US Treasury daily par yield curve rates, which the H.15 constant maturity yields are built from
Full dataset available as CSV, one row per trading day with the spread, a flag for whether that day sat inside a qualifying inversion, and a flag for whether the month was inside a recession.
The Digest Perspective
The most recent recession arrived with no inversion in front of it. February 2020 followed the August 2019 dip, three closes at four basis points, below the bar this page uses. That makes it 5 of 6 rather than all 6.
The open question is what came next. The curve came back above zero on 6 September 2024 after 793 days below it, the longest inversion on record, and on the 5 occasions a recession did follow it arrived 12 months before the curve came back to 8 months after it. We are 24 months past that now with nothing dated.
The Federal Reserve
Wants to cut without restarting inflation, and cutting is what steepens a curve. The un-inversion above is partly the Fed's doing rather than an independent all-clear, which is why twice the downturn began before that steepening appeared.
The Treasury
Wants to fund deficits, and heavy issuance at the long end steepens the same curve for reasons unrelated to growth. A steepening driven by 10-year yields rising is a different event from one driven by 2-year yields falling, which one number cannot separate.
Where we are probably wrong
8 episodes is not a sample to lean on, and the lead has stretched from 10 to 23 months across the 5 that landed. The committee that dates recessions also works well behind the fact, so the 24 months above is a floor rather than a verdict. Anyone presenting this curve as a clock is selling something.
We think the curve has done the one thing it does reliably, which is tell you the cycle turned. It has never said when, and 13 months between the shortest and longest lead is the honest width of that.
Which leaves the question that is yours rather than ours. Not whether a recession is coming, but whether the account would still be where you need it if one arrived at a time nobody can name. That turns on how close you are to withdrawals and what the rest of the account holds.
Opinion, dated and kept separate from the record above. Written by the Gold IRA Digest Editorial Team, which earns a commission when a company we match you with opens an account, changing neither what you pay nor what we publish.
Common questions
Yield curve inversion questions
- Is the yield curve inverted right now?
- No. The 10-year Treasury note yielded +0.41 percentage points more than the 2-year note at the 8 September 2026 close, and an inversion is that spread sitting below zero. The last inversion ran from 6 July 2022 to 6 September 2024, which is 793 days and the longest in the 50 years since this series began. Across the whole record the spread has closed below zero on 2,049 of 12,564 trading days, or 16.3% of them.
- What does a yield curve inversion mean?
- It means the bond market is paying less to lend the government money for ten years than for two, which only makes sense if traders expect short-term rates to be lower in a few years than they are now. Short rates fall when the Federal Reserve is cutting, and the Federal Reserve cuts into a slowing economy, so the inversion is a forecast of easing rather than a forecast of a recession directly. That distinction matters because the easing sometimes arrives without the recession, which happened after the June 1998 inversion.
- How long after a yield curve inversion does a recession start?
- Somewhere from 10 to 23 months on the 5 occasions it happened, measured from the first day the curve closed below zero to the month the National Bureau of Economic Research later dated as the start. The median is 17 months. Measured instead from the day the curve came back above zero, the recession arrived anywhere from 12 months before the curve came back to 8 months after it, because twice it had already begun while the curve was still inverted.
- Has a yield curve inversion ever been wrong?
- Once cleanly, out of the 8 episodes since June 1976. June 1998 is the miss: the curve spent 43 days below zero and the next recession was not dated for another 32 months. A second episode, March 1990, carries no recession in the table only because the downturn that followed it had already been credited to the inversion before it, which is bookkeeping rather than a failed signal. The most recent inversion ended on 6 September 2024 and nothing has been dated in the 24 months since, which is already longer than any previous inversion waited. A recession that began this year would not be dated for months yet, so treat that 24 months as a floor rather than a verdict.
- What does gold do after a yield curve inversion?
- Very little while the curve is inverted, and somewhat more once the recession arrives. Across the 8 inversions the median change in the London gold price from the month of the inversion to the month it ended was −0.3%, with 3 of them finishing higher. Through the 5 recessions that followed, measured from the month each began to the month it ended, the median was +6.9% and 3 finished higher. Both medians rest on fewer than ten observations and neither carries a fee, a dealer premium or a storage cost.
Gold IRA Digest is an independent publisher. This page is information, not investment advice, and nothing on it is a recommendation to buy or sell gold, Treasuries or any fund. Past performance does not predict future results. Some links on this site earn us a commission, which never changes what we publish about a company.