Finvest · Portfolio
Part II · The building blocks · Chapter 5 of 13

Chapter 5: The bond sleeve

11 min read · Evidence current as of June 2026 · Updated June 12, 2026

Vanguard's capital-market model, the VCMM run of March 31, 2026, puts US aggregate bonds at 4.2–5.2% annualized over the coming decade. That is a model assumption, not a promise, and this guide labels it as one every time it appears. What the number retires is an excuse: through most of the 2010s, every case for bonds had to open with an apology for yields near nothing. Bonds pay again. The question left over is what, exactly, you are paying them to do.

Chapter 2 chose your stock-bond mix off the worst-year column, and Chapter 3 showed the weather that mix depends on. This chapter writes the bond share's job description, because a sleeve without named jobs is the first thing sold in a storm, usually at the worst possible price.

The three jobs

Job one is shock absorption. In 2008 the S&P 500's total return was −37.0% while the US aggregate bond index returned +5.2%; a 60/40 mix computed at those weights lost about 20.1%. The sleeve's first job is to shrink your worst year to a dollar figure you can stand, and Chapter 2 already taught you to size it: pick the worst-year number you can hold, read your mix off the column.

Job two is the spending floor. Dollars you must spend on known dates inside the next decade do not belong in stocks at any allocation; they belong in bonds that mature on or before those dates, the ladders and TIPS the Finvest Cash & Bonds Guide builds rung by rung. A floor is what lets the rest of the portfolio fall without anyone missing a payment.

Job three is rebalancing ammunition. Chapter 10 will write a rule that buys stocks after they fall. The purchase has to be paid from somewhere, and the sleeve is the somewhere: in a 2008-shaped year, the +5.2% side funds the purchase of the −37.0% side. Without a sleeve, the rebalancing rule is a sentence with no wallet.

Three jobs, one sleeve 1 · Shock absorber 2 · Spending floor 3 · Rebalancing ammunition 2008: stocks −37.0%, the aggregate index +5.2% Year 1 · matures first Year 2 · on its date Year 3 · on its date a dollar for every spending date, so storms force no sales sell what held buy what fell Chapter 10's written rule, paid for from the sleeve A bond dollar holding none of these jobs is a forecast about interest rates.
Figure 5.1. The job description. Shock absorption (2008 receipts shown), spending floors that mature on their dates, and the ammunition Chapter 10's rebalancing rule spends. Every bond dollar in your plan should be able to name its job.

Sizing it: the worst-year column again

Chapter 2's table does the sizing, so recall its spine: Vanguard's published model-allocation data (1926–2024) puts the worst calendar year at −43.1% for 100% stocks, −34.9% for 80/20, −26.6% for 60/40, −18.4% for 40/60, −14.4% for 20/80, and −13.1% for 100% bonds, with average years running from 10.5% down to 5.0%.

Run Dev's $140,000 through it. At his 80/20, the worst year on record takes about $48,860 and leaves $91,140. At 60/40 the worst year takes about $37,240 and leaves $102,760. Twenty extra points of sleeve would have bought him $11,620 of worst-year shallowness and cost about 0.9 points of average year (9.7% against 8.8%). That is the entire trade, in his dollars, and he keeps 80/20 because he can hold the $48,860 figure: his house fund lives outside this portfolio entirely, in bills, exactly so that no market year can touch it.

What a bigger sleeve buys Average year and worst calendar year by mix. Vanguard model-allocation data, 1926–2024. 100% bonds avg 5.0%/yr 20/80 avg 6.4%/yr 40/60 avg 7.7%/yr 60/40 avg 8.8%/yr 80/20 avg 9.7%/yr 100% stocks avg 10.5%/yr −13.1% −14.4% −18.4% −26.6% −34.9% −43.1% Read from the right: each added slice of sleeve trades some average for a shallower worst year.
Figure 5.2. The worst-year column drawn as depth (Vanguard model-allocation data, 1926–2024). Pick the worst year you can hold in dollars, then read your sleeve off the bar; that single trade is what the sleeve's size controls.

The caveat from Chapter 3, applied

The job description above has weather conditions, and 2022 printed them. That year stocks fell 18.1%, the aggregate bond index fell 13.0%, and a 60/40 computed at those weights lost about 16.1%, its worst calendar year since the 1930s. AQR's correlation work (Journal of Portfolio Management, 2023) explains the betrayal: stock-bond correlation, consistently negative through the first two decades of the 2000s, spiked to about +0.50 in the 2022 episode, the highest in their 1972–2022 sample, and the whole 1970s–90s stretch was a positive-correlation regime. Inflation uncertainty drives those regimes. Bonds hedge growth scares, not inflation shocks.

So design the sleeve against your storm list from plan line three. For growth scares, the classic intermediate sleeve absorbs, as the 2008 receipts show. For inflation shocks, the sleeve needs short duration, real yields through TIPS, and the unglamorous help of cash yields, because long nominal bonds fall with stocks in exactly that storm. A sleeve built of one bond type is a bet on one kind of weather.

Duration, in one paragraph

The Cash & Bonds Guide drew duration as a seesaw: a number in years, printed on every fund page, telling you how hard the fund swings when rates move, and 2022's aggregate-index fall of 13.0% was that dial doing what its printed number implied in a rate storm. The working rule travels in one sentence: keep the duration number smaller than the years until the money is needed. Floor dollars ride short seats matched to their dates; retirement dollars decades from spending can ride intermediate ones. This guide will not rebuild the dial. The Finvest Cash & Bonds Guide gives it three chapters, ladders and TIPS included, and any bond dollar with a date attached should pass through those chapters before it is placed.

When can the sleeve be small, and when must it be big?

It can be small when the money's horizon is measured in decades, a steady paycheck covers life so no market year forces a sale, and your worst-year dollar line from Chapter 2 stays bearable at stock-heavy mixes, ideally proven in a decline you actually lived through rather than imagined. It must be big when withdrawals begin within about a decade or have already started, when known spending dates need a floor of maturing bonds, or when Chapter 9's rehearsal shows your sell line sitting above the worst-year column for the mix you hold. Notice that age appears nowhere in that sentence except as a proxy: the sleeve is sized by the worst year you can hold and the dates you must pay, and a 40-year-old with a sabbatical next year needs more floor than a 70-year-old whose pension covers the groceries. The two sketches below are the extremes of the same logic.

Two sleeves, forty years apart

Quinn's 2065 target-date fund holds roughly 10% bonds, and she once asked whether a slice that thin was worth having. As an absorber it barely registers: at a 90/10 mix the worst calendar year since 1926 runs to −39.0% (Vanguard's data, 1926–2024), hardly shallower than all-stocks. Its real jobs are the other two. It is the ammunition her fund spends automatically when it rebalances into a fall, and it is a small floor under nothing, because her actual shock absorber is a nurse's paycheck and a forty-year horizon. Her crash duties fit in one line: keep contributing and let the fund do the buying.

Elaine's sleeve points the opposite direction. Her $700,000 must produce $5,000 a month, so the nearest dollars cannot be allowed a bad year at all: her ladder from the Cash & Bonds Guide already holds the first years of spending in rungs that mature on their dates, with durations measured in months. Behind the rungs sit quality bonds for the rest of the decade's spending, and equities behind those for the decades after; Chapter 12 assembles her full mix at 45/55. A formula would call her sleeve oversized. The jobs explain it: most of her bond dollars are floor, because her spending dates are now, and her worst-year line is counted in grocery months rather than percent.

Size the whole sleeve off the worst-year dollar line you wrote in Chapter 2, then give every bond dollar one of the three jobs and a duration shorter than its date. A bond dollar with no job and no date is not safety; it is a forecast about interest rates.

Where people go wrong

Firing the sleeve over 2022. That year was the inflation regime Chapter 3 named, the storm bonds were never built for. In the growth scare of 2008 the same sleeve returned +5.2% against −37.0%. Judging an all-weather tool by its one bad weather is how portfolios end up with no absorber in the next growth scare.

Reaching for yield on the wrong seat. A long-duration fund pays more most years and swings hardest when rates move. Putting dated money on a long seat converts a spending floor into a bet, and 2022 priced that bet in public.

Owning "safety" with no job attached. A vague bond allocation gets sold in the panic, which is the one move the sleeve can never repay. Jobs and dates, written down, are what hold it in place when the statement is red.

Reading 4.2–5.2% as a promise. The VCMM figures are model assumptions from the March 31, 2026 run, quoted as a range because that is all a model honestly offers. Chapter 2 already made this point about every forward number, and it goes double for the sleeve you are counting on.

YOUR PLAN SO FAR

1. Every goal, with its date and its bucket (Chapter 1).

2. Target mix: __% stocks / __% bonds, chosen from the worst-year column (Chapter 2).

3. The storm list: what protects me in each kind of crash (Chapter 3).

4. Inside stocks: __% US / __% international (Chapter 4).

5. Bonds: __%, duration matched to __. New this chapter. Dev's line reads "20%, matched to a retirement in the 2060s"; Elaine's reads "55%, matched to my spending dates, nearest rungs first."

Key takeaways

  • Bonds pay again on paper: Vanguard's VCMM (March 31, 2026 run) models US aggregate bonds at 4.2–5.2% annualized over ten years. Model assumptions, always labeled, never promises.
  • The sleeve has three jobs: shock absorption (2008: stocks −37.0%, the aggregate +5.2%, a 60/40 about −20.1%), spending floors that mature on their dates, and the ammunition Chapter 10's rebalancing rule spends.
  • Size it from the worst-year column (Vanguard, 1926–2024): on Dev's $140,000, moving from 80/20 to 60/40 trades about 0.9 points of average year for $11,620 less worst-year damage.
  • The 2022 caveat stands: stocks −18.1%, bonds −13.0%, the 60/40 about −16.1%, with stock-bond correlation near +0.50, the highest in AQR's 1972–2022 sample. Bonds hedge growth scares, not inflation shocks; inflation storms call for short duration, TIPS, and cash yields.
  • Keep every duration number smaller than the years until that money is needed; the full dial, with ladders and TIPS, lives in the Finvest Cash & Bonds Guide.

Sources: Vanguard return forecasts (VCMM) · Vanguard model allocation data · AQR on the stock-bond correlation · Finvest Cash & Bonds Guide