Hedging with bonds
Why investors hold Treasuries beside their stocks, when the cushion worked, and when it failed.
From the Nasdaq Composite’s high to its low in each fall (price only), and a new 10-year note bought on the high and priced on the low, coupons included · FRED, NASDAQCOM and DGS10Four falls in stocks, and a 10-year note in each · FRED
00 read this with your agent
Read this with your agent first.
Copy one into any AI. It teaches from this page only. In a browser with WebMCP, an agent can call the page’s own tools, including get_formula, and work through the numbers with you.
Builds on note 1.6, Reading the yield today. Every term is defined where it first appears, and the words this note defines are collected at the end.
1Set up your AI optional, two minutes, once
2Then copy a prompt
01 what a hedge is
What a hedge is.
Something you hold because it tends to rise when the rest falls.
Start with home insurance. Most years you pay for it and get nothing back. In the year of a fire, it pays. You don’t buy it to make money. You buy it to lose less in a bad year.
Investors do the same with what they own. Everything an investor holds is a portfolio. A hedge is the part of it that tends to gain when the rest loses. The most common pair is stocks for growth and US government bonds, Treasuries, as the cushion.
| home insurance | Treasuries beside stocks | |
|---|---|---|
| in a calm year | you pay, and get nothing back | the bonds usually earn less than the stocks |
| in a bad year | it pays for the damage | the bonds rise while the stocks fall |
| what it costs | the yearly payment | the return given up in good years |
| when it fails | the damage isn’t covered | the bonds fall with the stocks |
Two questions decide whether the cushion works: why would a bond rise when stocks fall, and does it always?
02 why treasuries
Why Treasuries rise in a scare.
Two parts of the yield fall at once.
A bond’s payments are fixed, so its price moves opposite to its yield: a falling yield means a rising price (note 1.1). And the 10-year yield is the expected path (the Fed’s rate as lenders expect it to average over ten years) plus the term premium (extra pay for locking money up that long) (note 1.3). When a scare comes from a weakening economy, both parts tend to fall:
| step | what happens | what it does to the 10-year |
|---|---|---|
| one | bad news about the economy: layoffs, falling sales | – |
| two | lenders expect the Fed to cut its rate to protect jobs | the expected path falls |
| three | investors sell stocks and buy the safest thing they can find: a flight to safety | more buyers for Treasuries, so their price rises |
| four | the yield falls | every note already held is worth more |
Why Treasuries, and not any bond?
Because the US government is about the safest borrower there is (note 1.2), and Treasuries trade every working day, so they can be sold at once. A company’s bond can fall in a scare, because the company may not repay.
What the Fed did in each of the last four big falls in stocks, from the day the Nasdaq peaked to the day it hit bottom:1
| fall | the Fed’s rate (upper target) | the 10-year | the Fed |
|---|---|---|---|
| 2000–02 | 5.75% → 1.75% | 6.39% → 3.61% | cut |
| 2007–09 | 4.5% → 0.25% | 4.48% → 2.89% | cut |
| 2020 | 1.75% → 0.25% | 1.56% → 0.76% | cut |
| 2021–22 | 0.25% → 4.5% | 1.54% → 3.88% | hiked, to fight inflation |
Three times the Fed cut and the 10-year fell. The fourth time it hiked, and the 10-year rose. That one difference decides the rest of this note.
03 when it worked
When the cushion worked, and when it failed.
Four falls in stocks. Three cushioned, one not.
Each fall runs from the Nasdaq Composite’s high to its low. (The Nasdaq is an index of the stocks listed on the Nasdaq exchange, heavy in technology: note 1.4.) Beside it, a new 10-year note bought on the day of the high.
How is the note’s return worked out?
| step | what we do |
|---|---|
| one | buy a new 10-year note on the day of the high, for $1,000. A new note pays that day’s yield, so that is its coupon. |
| two | collect its yearly interest until the day of the low |
| three | on the day of the low, price what is left of it at that day’s yield, the way note 1.1 priced a bond |
| four | add two and three, and compare with the $1,000 paid |
The result in money. The Nasdaq’s level on the day of its high and the day of its low, and what $1,000 put into each on the high was worth on the low:17
| fall | the Nasdaq, high → low | $1,000 in stocks became | $1,000 in a new 10-year note became |
|---|---|---|---|
| 2000–0210 Mar 2000 – 9 Oct 2002 | 5,049 → 1,114 | $221 (−77.9%) | $1,343 (+34.3%) |
| 2007–0931 Oct 2007 – 9 Mar 2009 | 2,859 → 1,269 | $444 (−55.6%) | $1,181 (+18.1%) |
| 202019 Feb – 23 Mar 2020 | 9,817 → 6,861 | $699 (−30.1%) | $1,078 (+7.8%) |
| 2021–2219 Nov 2021 – 28 Dec 2022 | 16,057 → 10,213 | $636 (−36.4%) | $844 (−15.6%) |
In 2007–09 the stocks lost more than half, and over the same 16 months the note gained $181: a gain while everything else fell. In 2021–22 both lost.
The note gained in three falls and lost in the fourth. FIG 1 draws each fall week by week, with 3-month bills beside them; the last button is a year when stocks rose.
Why did 2021–22 fail?
One cause hit both. Inflation reached 9.1%, so the Fed hiked. Higher rates pushed stocks down, and the same higher rates pushed bond prices down. In the other three falls the cause was a weakening economy: it pushed stocks down and pulled yields down, so bonds rose. Note 1.5 called these the growth kind and the strain kind:
| a growth scare | an inflation scare | |
|---|---|---|
| what goes wrong first | jobs and spending | prices |
| what the Fed does | cuts | hikes |
| the 10-year | falls | rises |
| Treasuries | rise | fall |
| stocks | fall | fall |
| the cushion | works | fails |
| here | 2000–02, 2007–09, 2020 | 2021–22 |
So a Treasury is a hedge against growth scares, not against inflation scares. Which kind is more common depends on the era: researchers tie the switch to whether inflation or growth news leads, and to how central banks respond.17 Note 1.5’s FIG 2 shows the eras.
04 which bond
The length of the bond is a dial.
Longer bonds cushion more in a growth scare, and lose more in an inflation scare.
A one-point move in yields changes a bond’s price by about its duration in percent. Duration is roughly the average wait, in years, for a bond’s payments, and longer bonds have longer durations (note 1.1). So a hedger chooses how much swing to hold. The shortest is 3-month bills rolled over: each bill replaced with a new one as it comes due. The same falls, by length:1
| fall | 3-month bills | 2-year note | 10-year note | 30-year bond |
|---|---|---|---|---|
| 2000–02 | +10.2% | +14.1% | +34.3% | +34.3% |
| 2007–09 | +2.0% | +7.2% | +18.1% | +26.8% |
| 2020 | +0.1% | +2.3% | +7.8% | +16.9% |
| 2021–22 | +2.1% | −2.7% | −15.6% | −33.1% |
In the 2021–22 fall the 30-year bond lost about as much as the Nasdaq did. Bills never lost, but in a sudden fall they earned almost nothing: 0.1% in 2020. They protect against losing money, not against stocks falling.
The trade-off: the more a bond cushions in a growth scare, the more it loses in an inflation scare. In 2000–02 the 30-year gained no more than the 10-year: its yield fell less, from 6.19% to 4.95%, and its longer duration only made up the difference.
05 one portfolio
One portfolio, worked.
$10,000 all in stocks, or 60% in stocks and 40% in 10-year notes.
A 60/40 portfolio holds 60% in stocks and 40% in bonds: a common split for investors who want a cushion. Take $10,000 into the 2007–09 fall, bought on the day of the high and left alone until the low, one way or the other:
| 2007–09 | all in stocks | 60/40 |
|---|---|---|
| put in on 31 Oct 2007 | $10,000 in stocks | $6,000 in stocks, $4,000 in notes |
| the stocks: −55.6% | $10,000 → $4,440 | $6,000 → $2,664 |
| the notes: +18.1% | – | $4,000 → $4,724 |
| left on 9 Mar 2009 | $4,440 | $7,388 |
The 60/40 ended $2,948 ahead: the notes grew while the stocks shrank.
The same sum for every fall, just the totals:
| fall | all in stocks, at the low | 60/40, at the low | the 60/40 kept more |
|---|---|---|---|
| 2000–02 | $2,210 | $6,698 | $4,488 |
| 2007–09 | $4,440 | $7,388 | $2,948 |
| 2020 | $6,990 | $8,506 | $1,516 |
| 2021–22 | $6,360 | $7,192 | $832 |
In the three growth scares the notes gained and roughly halved the loss. In 2021–22 the 60/40 still lost less, but only because the notes fell less than the stocks: both fell.
And in a good year?
That is the cost of the cushion. In the year to 30 September 2026 stocks rose and the 10-year note fell, because its yield rose from 4.16% to 5.29%:
| 30 Sep 2025 – 30 Sep 2026 | the return | $10,000 becomes |
|---|---|---|
| all in stocks (the Nasdaq) | +18.5% | $11,850 |
| a new 10-year note | −3.8% | – |
| 60/40 | +9.6% | $10,958 |
| what the cushion cost | $892 |
Like insurance, the cushion is paid for in the years nothing goes wrong.
06 other tools
Other tools, and what each protects against.
Each one covers one risk, not all of them.
If the cushion fails in an inflation scare, why not hold TIPS?
A TIPS’s face value rises with prices (note 1.3), so it protects against inflation. But its price still moves with its own yield, the real yield: what it pays on top of inflation. In 2022 inflation was high, and the real yield also jumped. A 10-year TIPS bought at the start of 2022 paid the lowest coupon the Treasury sets, 0.125%:2116
| 2022, the calendar year | an ordinary 10-year note | a 10-year TIPS |
|---|---|---|
| its yield | 1.52% → 3.88% | the real yield −1.04% → 1.58%, up 2.62 points |
| its price, per $100 of face value | – | $112.35 → $87.88: −21.8% |
| inflation added to the face value | none | +6.5% |
| the year’s return | −16.1% | −16.6% |
The TIPS was paid for the inflation, but the jump in the real yield cut its price by more. A TIPS hedges inflation, not a rise in the real yield. In 2022 both came at once.
Two more tools are about time, not yields. Matching means holding a bond that comes due when the money is needed. Its price swings on the way, but the loss is only locked in if it is sold (note 1.1). A ladder matches many dates at once: bonds that come due one after another, say one a year for ten years. Pension funds match bonds to the pensions they must pay (note 1.4).
| tool | protects against | does not protect against | when it went wrong |
|---|---|---|---|
| long Treasuries | a growth scare | an inflation scare | 2021–22: down with stocks |
| bills | losing money | stocks falling: they earn only the short rate | – |
| TIPS | inflation | a jump in the real yield | 2022 |
| matching, a ladder | price swings, if held to the end | having to sell early | the UK, 2022: pension funds had borrowed against their bonds and were forced to sell (note 1.4); Silicon Valley Bank, 2023: long bonds against deposits that left (note 1.5)1416 |
Both failures were a mismatch: the bonds were long, but the money could be called back at once.
07 what a hedger watches
What a hedger watches.
Signs of which kind of scare is more likely next.
| gauge | what it tells a hedger | on 30 Sep 2026 |
|---|---|---|
| do stocks and yields move together? (note 1.5, FIG 2) | above zero: growth news leads, and bonds have been cushioning; below zero: inflation and rate news lead, and they have not | −0.17, below zero since April |
| the breakeven: the inflation the market expects (note 1.3) | rising: inflation fears, the setting where the cushion fails | 2.36%, flat for a year |
| the Fed’s direction (note 1.2) | cutting into a slowdown: the cushion works; hiking into inflation: it fails | hiked on 16 September, to 3.75–4.00% |
| the term premium: extra pay for lending long (note 1.3) | rising: lenders want more to hold long bonds, which then swing more | 0.50 → 1.02 in a year |
On 30 September the gauges were mixed: expected inflation flat, but stocks and yields moving opposite and the Fed hiking, the setting in which the cushion has been weaker. That is a reading of the gauges, not a forecast.
This note explains how investors use bonds. It does not say what anyone should hold.
08 the words
The words in this note.
Every term this note defines, in the order it appears.
| word | what it means |
|---|---|
| flight to safety | Investors selling riskier things, such as stocks, in a scare and buying the safest things they can find, such as Treasuries. |
| 3-month bills, rolled over | Owning a 3-month Treasury bill and replacing it with a new one each time it comes due: it earns the short rate and barely moves in price. |
| 60/40 portfolio | A portfolio with 60% in stocks and 40% in bonds, a common split for investors who want a cushion. |
| matching | Holding a bond that comes due on the date the money is needed, so its price swings on the way don’t matter unless it is sold. |
| ladder | Bonds that come due one after another, say one each year, so money comes back on a schedule. |
Every term in the notebook, with the note that explains it, is in the notebook’s glossary.
sources read 2026-10-01
hover, tap or focus · every number above is in here
Sources · 7 references
Every series was downloaded from FRED on 1 October 2026 (unadjusted CPI on 2 October), and the filings from EDGAR the same day; both are kept as a snapshot with the post. Every number on the page is computed from that snapshot or quoted from the sources below. Each source is marked peer-reviewed or not. Nothing here is advice: the post explains; it does not recommend.
- Board of Governors of the Federal Reserve System (2026). Selected Interest Rates (H.15): market yields on US Treasury securities at constant maturity, 1-month to 30-year (DGS1MO … DGS30), 10-year TIPS (DFII10), the federal funds target range (DFEDTARL, DFEDTARU; DFEDTAR before 2008) and the effective rate (FEDFUNDS); the broad dollar index (DTWEXBGS, H.10). Retrieved from FRED, Federal Reserve Bank of St. Louis, 1 October 2026. Data, not peer-reviewed. fred.stlouisfed.org/series/DGS10
- US Bureau of Labor Statistics (2026). Consumer price index for all urban consumers, not seasonally adjusted (CPIAUCNS, retrieved 2 October 2026), the unemployment rate (UNRATE) and nonfarm payrolls (PAYEMS, retrieved 1 October 2026), from FRED. Inflation is computed here as the change from the same month a year earlier, as BLS reports it; October 2025 has no index, because no prices were collected during the shutdown. Data, not peer-reviewed. fred.stlouisfed.org/series/CPIAUCNS
- Nasdaq OMX Group and S&P Dow Jones Indices (2026). Nasdaq Composite (NASDAQCOM) and S&P 500 (SP500), daily closes. Retrieved from FRED, 1 October 2026. Data, not peer-reviewed. fred.stlouisfed.org/series/NASDAQCOM
- Bank of England (2022). Bank of England announces gilt market operation, 28 September 2022; and Financial stability buy/sell tools: a gilt market case study, Quarterly Bulletin 2023. The 130-basis-point rise in 30-year gilt yields over three days of trading, the forced selling by LDI funds, and the purchases. Not peer-reviewed. bankofengland.co.uk/quarterly-bulletin/2023
- Board of Governors of the Federal Reserve System (2023). Review of the Federal Reserve’s supervision and regulation of Silicon Valley Bank, 28 April 2023. Not peer-reviewed. federalreserve.gov/publications/files/svb-review-20230428.pdf
- Campbell, J.Y., Pflueger, C. and Viceira, L.M. (2020). Macroeconomic drivers of bond and equity risks. Journal of Political Economy 128(8), 3148–3185. Why the stock–bond correlation changes sign between regimes, and the role of monetary policy in it. Peer-reviewed. doi.org/10.1086/707766
- US Department of the Treasury (2026). Treasury Inflation-Protected Securities (TIPS). TreasuryDirect. The interest rate “is fixed at auction and is never less than 0.125%”, and the principal rises and falls with the BLS consumer price index (checked 10 October 2026). Not peer-reviewed. treasurydirect.gov/marketable-securities/tips
To verify before publication: Costco’s 30 September close against Koyfin. Checked against their own pages on 2 October 2026: the FOMC statement of 16 September 2026, the BLS releases for August 2026, the Treasury’s statements of 2 August and 1 November 2023, the Bank of England’s case study, the Federal Reserve History essay, and every series against a fresh recomputation. Not sourced, and so not stated as numbers: the market-implied odds of a hike, and the average maturity of the government’s debt.
documentation
- FRED: the 10-year Treasury yield (DGS10): The series the post is about, daily since 1962, from the Fed's H.15 release.
- TreasuryDirect: marketable securities: What bills, notes and bonds are, and how long each lends for.
- the Federal Open Market Committee: Where the Fed sets the federal funds rate: its meetings, statements and projections.
- TreasuryDirect: TIPS: Treasury Inflation-Protected Securities, whose yield is a real yield.
- the Kim–Wright term structure model: The Fed's page for the model and its published term premium estimates.