In normal times, longer-term bonds pay more than shorter-term bonds. You're lending money for longer, so you demand more interest. The chart of yields by maturity slopes up.
When it doesn't — when 2-year Treasuries pay more than 10-year Treasuries — that's an inversion. And for the last 50 years, inversions have preceded every U.S. recession.
This is one of the few macro signals that actually has predictive value. Worth understanding.
What the curve actually shows
The yield curve plots interest rates against maturity. On any given day:
- 3-month Treasury bill — closest to the Fed funds rate
- 2-year Treasury note — market expectation of where Fed funds will be over the next 2 years
- 10-year Treasury note — long-term growth + inflation expectations
- 30-year Treasury bond — very long-term outlook
The most-watched spread is 10s minus 2s (also written "10Y - 2Y" or "T10Y2Y"). When that number is positive, the curve is normal. When it's negative, the curve is inverted.
Why inversion predicts recessions
Here's the logic:
The short end of the curve is dominated by Fed policy. If the Fed is fighting inflation by hiking rates aggressively, short-term yields go up fast.
The long end is dominated by future growth and inflation expectations. If markets expect growth to slow (because of those rate hikes), long-term yields don't rise as much — and may even fall if recession is expected.
Net result: short rates above long rates. The curve inverts.
The signal: the bond market is collectively saying "the Fed is going to break something with these rate hikes, and we expect lower rates and slower growth ahead."
That collective opinion has been right about every U.S. recession since 1968. Not perfect — but a remarkably consistent leading indicator.
The lag
Here's the critical detail most retail traders miss: inversion is not a "sell stocks now" signal.
The historical pattern:
- Curve inverts
- Stocks often continue rising for 6-18 months afterward
- Then recession arrives, stocks drop hard
Average lag from initial inversion to recession start: roughly 14-18 months. The market often makes new highs during this window.
If you panic-sold the day of the 2022 inversion, you missed substantial upside before the actual slowdown materialized.
Variations of the curve to watch
Different spreads tell different stories:
| Spread | What it tells you |
|---|---|
| 10Y - 2Y | The classic recession indicator |
| 10Y - 3M | More forward-looking; the Fed's preferred measure |
| 30Y - 5Y | Long-term inflation expectations |
| 5Y - 2Y | Short-term Fed expectation |
The 10Y-3M is actually more reliable than 10Y-2Y in some academic research. When BOTH are inverted, the recession signal is most robust.
What it doesn't tell you
The curve gives you direction and rough timing. It doesn't tell you:
- How severe the recession will be
- When exactly stocks will peak
- Which sectors will be hit hardest
- How long the recession will last
You use other tools for those questions. Inversion is a heads-up, not a playbook.
How to actually trade around it
When the curve inverts:
- Don't sell everything. History says markets often rip first, recess later.
- Slowly tilt defensive. Increase staples, healthcare, utilities allocation by 5-10%.
- Watch credit spreads. If high-yield bond spreads start widening simultaneously, the timeline accelerates.
- Keep more cash than usual. 10-20% cash is reasonable in late-cycle. Buying opportunities in recessions are the best buying opportunities, period.
- Tighten stop losses. Don't let a 5% loser become a 25% loser when conditions are deteriorating.
When the curve "uninverts"
Almost as important: when the curve goes back to normal (10Y rises above 2Y again).
Counter-intuitively, this is often closer to the actual recession start than the inversion itself. The Fed has typically started cutting rates by then — which steepens the curve — and the cuts are usually in response to deteriorating conditions.
The uninversion is the warning that the slowdown is no longer hypothetical.
Two big caveats
- The 2022-2023 inversion took longer than expected. Markets pushed back on the signal for months. This led some commentators to declare "this time is different." It usually isn't, but the timing can be off by 6+ months.
- Fed manipulation distorts the curve. Massive QE/QT programs change the relationship between Fed expectations and long rates. The signal still works but is noisier when the Fed is heavily active in the bond market.
What to actually do with this
Open T10Y2Y on FRED (Federal Reserve Economic Data) and bookmark it. Check it once a month. When it crosses below zero, take note. When it crosses back above, take notes again.
That's it. You're not trying to predict the exact day of recession. You're maintaining situational awareness so you're not the last person at the party when the music stops.