ResourcesGuide

Barbell Investing Strategy: How It Works, With Worked Examples

A barbell puts most money somewhere very safe and a small slice into bets that can lose everything or multiply, with nothing in the middle. Here is the arithmetic of an 85/15 barbell, how the bond-maturity barbell differs, and where both break down.

Published Updated 9 min readHow we research

Key takeaways

  • A barbell combines a very safe end (Treasury bills, insured cash) with a small speculative end and skips medium-risk assets. Nassim Taleb’s examples use roughly 85–90% safe and 10–15% speculative.
  • Worked example, $100,000 at 85/15 with T-bills at 4.22% (Oct 22, 2026): if the 15% goes to zero, the total is $88,587 (−11.4%); if it triples, $133,587 (+33.6%).
  • The cap holds for one year, not for life: refilling the sleeve after five total losses in a row would leave $54,557 of $100,000.
  • The safe end grows slowly: T-bills returned 3.4% a year from 1928 to 2025 and 0.3% after inflation (Exent calculation on NYU Stern data).
  • The bond barbell is a different idea: short plus long Treasuries instead of one middle maturity. On Oct 22, 2026 a duration-matched version yielded 4.77% vs 4.96% for a 10-year note.

Market risk reading, cross‑asset context and an events calendar: see Exent Market intel →

What is the barbell investing strategy?

It is a portfolio built at two extremes: most of the money in assets that are very unlikely to lose value, and a small slice in high-risk, high-payoff bets, with nothing in between. The aim is a bounded loss: a one-period loss bound is possible only if the safe sleeve preserves principal and the speculative positions cannot create liabilities beyond their purchase cost (no margin, no uncovered options). In Antifragile, Nassim Taleb describes 90% in boring cash and 10% in maximally risky securities, so that you “cannot possibly lose more than 10 percent” while keeping the upside; in The Black Swan he gives the range as 85 to 90 percent in very safe instruments such as Treasury bills and 10 to 15 percent in very speculative bets.

Scope. The idea, its arithmetic and its history for US investors, plus the unrelated bond barbell that shares its name. Not covered: which speculative assets to pick (this page names none on purpose), options strategies, or tax treatment. All mixes are hypothetical illustrations, not advice.

Two kinds of barbell
Taleb barbell
  • Safe end85–90%: T-bills, insured cash
  • Middleleft empty
  • Speculative end10–15%: can go to zero
  • Safe end → Short maturities · both rely on short bills
  • Speculative end → Long maturities · different risk: rates, not ruin
Bond barbell
  • Short maturitiesbills, 1–2 years
  • Middle maturitiesleft out
  • Long maturities20–30 years
Illustration. The Taleb barbell caps losses; the bond barbell reshapes interest-rate risk.

What happens to an 85/15 barbell if the risky part goes to zero or triples?

Hypothetical $100,000 for one year: $85,000 in Treasury bills at the 4.22% 3-month yield of October 22, 2026 (FRED), assumed to hold for the year, and $15,000 in a speculative sleeve. The safe end ends at $85,000 × 1.0422 = $88,587 in every case.

Speculative sleeveArithmeticTotal after 1 yearChange
Goes to zero$88,587 + $0$88,587−11.4%
Halves$88,587 + $7,500$96,087−3.9%
Flat$88,587 + $15,000$103,587+3.6%
Triples$88,587 + $45,000$133,587+33.6%
  • Barbell 85/15, sleeve triples+33.6%
  • Barbell 85/15, sleeve goes to zero−11.4%
  • 60/40 stocks/Treasuries, worst year (1931)−27.3%
  • 100% S&P 500, worst year (1931)−43.8%

With fully paid speculative positions, the barbell’s worst year is set by its own design, −11.4% here, while a 60/40 stock/Treasury mix lost 27.3% in 1931 and 18.0% in 2022 (Exent calculation on NYU Stern data). That is the whole appeal. The catch is the speculative end: it has to be able to pay several times its cost for the structure to beat ordinary portfolios, and most single speculative bets do not. How far such bets can fall is shown in how much of a portfolio to put in speculative stocks, where popular growth names lost 81% to 98% in 2022.

Is the barbell’s maximum loss really capped?

Only per period. If the sleeve is refilled to 15% every year and lost in full each time, the losses compound:

Total losses in a rowEach yearLeft of $100,000
1× (0.85 × 1.0422) = × 0.8859$88,587 (−11.4%)
2× 0.8859 again$78,477 (−21.5%)
3$69,520 (−30.5%)
5$54,557 (−45.4%)

Five losing years in a row leave a barbell investor worse off than a 60/40 investor in its worst fall since 1928 (−40.1% in year-end value, 1928–32). The cap is real only with a rule decided in advance: how often the sleeve is refilled, and from what. Taleb’s own framing stresses many small bets with limited downside, not one large bet, and protecting the safe end from ruin first.

How did a barbell-style mix do historically?

No long data series exists for “speculative bets”, so the nearest honest test swaps the speculative end for the S&P 500, which is far less risky than a true barbell sleeve. Annual data, rebalanced every January, no fees or taxes:

Hypothetical mixReturn a year, 1928–2025After inflationWorst yearReturn a year, 1972–2025Worst year since 1972
85% T-bills / 15% S&P 5004.6%1.5%−5.1% (1937)5.6%−4.3% (2008)
90% T-bills / 10% S&P 5004.2%1.1%−3.3% (1937)5.2%−2.4% (2008)
30% S&P 500 / 70% 10-yr Treasuries6.6%3.5%−17.9% (2022)7.8%−17.9% (2022)
60% S&P 500 / 40% 10-yr Treasuries8.3%5.1%−27.3% (1931)9.4%−18.0% (2022)
  • 85% T-bills / 15% S&P 500
    Return a year, 1928–2025
    4.6%
    After inflation
    1.5%
    Worst year
    −5.1% (1937)
    Return a year, 1972–2025
    5.6%
    Worst year since 1972
    −4.3% (2008)
  • 90% T-bills / 10% S&P 500
    Return a year, 1928–2025
    4.2%
    After inflation
    1.1%
    Worst year
    −3.3% (1937)
    Return a year, 1972–2025
    5.2%
    Worst year since 1972
    −2.4% (2008)
  • 30% S&P 500 / 70% 10-yr Treasuries
    Return a year, 1928–2025
    6.6%
    After inflation
    3.5%
    Worst year
    −17.9% (2022)
    Return a year, 1972–2025
    7.8%
    Worst year since 1972
    −17.9% (2022)
  • 60% S&P 500 / 40% 10-yr Treasuries
    Return a year, 1928–2025
    8.3%
    After inflation
    5.1%
    Worst year
    −27.3% (1931)
    Return a year, 1972–2025
    9.4%
    Worst year since 1972
    −18.0% (2022)

Exent calculation. The proxy barbell had shallow worst years but earned little after inflation: the safe end itself, 3-month T-bills, returned 3.4% a year from 1928 to 2025 and only 0.3% after inflation, with 40 of 98 years losing purchasing power. A real barbell needs the speculative end to do the heavy lifting, which is exactly the part history cannot test. Read the table as historical results for a bill-heavy stock proxy, not as a forecast.

What is a barbell strategy in bonds?

A different tool with the same name: holding short-term and long-term bonds instead of one middle maturity (a “bullet”), or a ladder spread evenly across maturities. It changes how the portfolio reacts to rate moves; it does not cap losses the way Taleb’s version does. Treasury yields on October 22, 2026 (Federal Reserve via FRED): 3-month 4.22%, 2-year 4.65%, 10-year 4.96%, 30-year 5.29%.

Hypothetical holdingYieldDuration (years)ConvexityIf all yields +1 pointIf all yields −1 point
Bullet: 10-year Treasury4.96%7.874−7.4%+8.2%
Barbell: 48.6% 3-month bills + 51.4% 30-year bonds4.77%7.8172−7.0%+8.7%
Hypothetical holdingBullet: 10-year TreasuryBarbell: 48.6% 3-month bills + 51.4% 30-year bonds
Yield4.96%4.77%
Duration (years)7.87.8
Convexity74172
If all yields +1 point−7.4%−7.0%
If all yields −1 point+8.2%+8.7%

Arithmetic on par bonds, weights chosen so both have the same duration (sensitivity to rates; see bond duration). The barbell gives up 0.19 points of yield today for more convexity, so it lost less in a 1-point rise and gained more in a 1-point fall. That edge only holds when short and long yields move together. If long yields rise more than the 10-year yield, the barbell loses more than the bullet; it also needs the bills rolled every few months at whatever rate prevails. For a simpler comparison of ladders and funds, see bond ladder vs bond fund.

A third use of the word describes stock portfolios split between high-beta growth stocks and low-volatility defensive ones. That is a stock-only tilt with no safe end; see balancing growth and defensive stocks.

When does a barbell strategy not work?

When the safe end is not safe or the speculative end is not small. Counterexample: an investor calls 80% “safe” but holds it in a long-term bond fund, and puts 20% in one hot stock. In 2022 the 10-year Treasury lost 17.8% and a long-term Treasury fund (TLT) lost 31.2% (Exent market data), while speculative growth stocks fell 81% to 98%: no part of that portfolio provided the bound the barbell design aims for promises. The same goes for a sleeve that grows after a good year and is never trimmed: a 15% slice that triples becomes about 34% of the total ($45,000 ÷ $133,587), and the cap quietly disappears.

  • The safe end is only as safe as its holdings: Treasury bills held to maturity or insured deposits within FDIC limits, not long bond funds.
  • Inflation erodes the safe end: T-bills earned 0.3% a year after inflation from 1928 to 2025.
  • The speculative end needs a written rule for size, refills and trimming; without it, the loss cap does not exist.
  • Limitations: the historical rows use the S&P 500 as a stand-in for speculative bets; calendar-year data, no fees or taxes; bond figures use one day’s yields and a parallel-shift approximation.
  • Everything here is a hypothetical illustration, not a recommendation; for hedging tools, see tail-risk hedging and the cost of hedging.

Related: tail-risk hedging, what hedging costs, asymmetric risk and reward and the core-satellite portfolio, which keeps a diversified core instead of an empty middle.

Add market context with Exent Market intel

Market intel provides market context, cross‑asset analysis, an events calendar and regular written notes. These can inform your review of an existing plan; they do not reliably predict crashes or guarantee better outcomes.

  1. Regular written notes that read cross‑asset ratios, including high‑beta vs low‑volatility stocks and stocks vs bonds.
  2. A Risk ON, Neutral or Risk OFF reading of US market risk built from four pillars, with an email when the reading changes.
  3. An events calendar and 26 Market Intel 101 lessons on rates, bonds and market structure.

Exent is our product. Educational content, not investment advice.

Frequently asked questions

Hypothetical: $85,000 in Treasury bills and $15,000 in speculative bets. With T-bills at 4.22% for the year, the two illustrated outcomes are $88,587 if the sleeve is lost and $133,587 if it triples.

Taleb’s examples use about 85–90% very safe and 10–15% very speculative. The split is a choice, not a rule; the safe side sets the worst case.

It can cap losses if the safe end is truly safe and the sleeve stays small. The weak points are low real returns on the safe end (0.3% a year for T-bills after inflation, 1928–2025) and refilling a sleeve that keeps failing.

Holding short and long bonds instead of middle maturities. On Oct 22, 2026, a mix of 3-month bills and 30-year bonds with the same duration as a 10-year note yielded 4.77% versus 4.96%, with more convexity.

A ladder spreads money evenly across maturities so some matures every year; a barbell concentrates it at the shortest and longest ends and leaves the middle empty.

In one period, not if the safe end holds. Over several periods, yes: refilling a 15% sleeve that is lost five years in a row would turn $100,000 into $54,557.

The terms are used for splitting stocks between high-growth and defensive names. It has no safe end, so it does not cap losses; in 2022 many high-growth stocks lost most of their value.

Sources

  1. Antifragile: Things That Gain from Disorder — Random House (Nassim Nicholas Taleb), Nov 27, 2012
  2. The Black Swan: The Impact of the Highly Improbable — Random House (Nassim Nicholas Taleb), Apr 17, 2007
  3. Historical Returns on Stocks, Bonds and Bills: 1928–2025 (histretSP) — NYU Stern School of Business, Aswath Damodaran, Aug 24, 2026
  4. Market Yield on U.S. Treasury Securities at 3-Month Constant Maturity (DGS3MO) — Federal Reserve Board via FRED, Oct 22, 2026
  5. Market Yield on U.S. Treasury Securities at 2-Year Constant Maturity (DGS2) — Federal Reserve Board via FRED, Oct 22, 2026
  6. Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity (DGS10) — Federal Reserve Board via FRED, Oct 22, 2026
  7. Market Yield on U.S. Treasury Securities at 30-Year Constant Maturity (DGS30) — Federal Reserve Board via FRED, Oct 22, 2026

Add market context to your plan.

Market intel provides a Risk ON / Neutral / Risk OFF reading from four pillars, cross-asset analysis, an events calendar and regular written notes. Context for reviewing a plan, not a crash forecast.

See what Market intel includesLimited offer67% off your first month

Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.

DISCLAIMER: Exent AI is an informational research platform and does not provide investment advice, financial advice, or act as an investment adviser, broker-dealer, or fiduciary. All content, analytics, and model outputs are provided for educational and informational purposes only and should not be relied upon as a recommendation to buy, sell, or hold any financial instrument. Investing involves risk, including the possible loss of principal. Exent AI does not execute trades, hold customer funds, or provide brokerage services. Past performance and hypothetical results are not indicative of future results. Third-party data and AI-generated outputs may be incomplete or inaccurate.