This blog covers the general topic of financial markets.


The 60/40 Portfolio: A Regime-Dependent Average, Not a Law of Nature

first posted: 2026-05-26 20:56:02.690836

What if the classic 60/40 portfolio isn't timeless wisdom, but just one regime among many?

Most investors hold something close to 60% equities and 40% bonds. The logic is simple: equities provide growth, bonds provide safety, and the mix has delivered decent returns with manageable drawdowns. But when we look under the hood across nearly a century of data, a highly unstable picture emerges.

All returns are monthly excess returns (above the risk-free rate), annualized. Drawdowns are maximum peak-to-trough. Weights are unconstrained optimal allocations. Data from Shiller, FRED, CMO (gold), and Bloomberg (oil).

Note: All allocations are unconstrained mean-variance optimal. Real-world portfolios would apply leverage, concentration, and shorting limits. Extreme weights illustrate regime instability rather than actionable advice.


The Full-Sample Case for 60/40

Over the entire 1934–2026 period, the optimizer agrees: 55% equities, 45% bonds.

Stocks & Bonds (1934–2026)

img

AssetExcess ReturnVolatilitySharpeMax DrawdownWeight
Stocks8.1%12.9%0.63−53.1%55%
Bonds1.4%5.7%0.25−49.2%45%
Portfolio8.8%7.6%1.15−23.8%

Risk-free rate: 3.46% | Kelly leverage (K): 8.7

This looks reasonable. A 1.15 Sharpe, a −24% worst drawdown. It's the bedrock of modern portfolio theory. But this full-sample average hides enormous variation.


Adding Gold: The Fiat Era Picture

Gold was price-suppressed before 1971 (dollar pegged, not freely convertible). Using CMO data from 1971 onwards, the picture shifts:

Stocks, Bonds & Gold (1971–2026)

AssetExcess ReturnVolatilitySharpeMax DrawdownWeight
Stocks7.0%12.7%0.55−53.1%40%
Bonds1.9%6.9%0.27−40.5%38%
Gold5.9%16.5%0.36−90.9%22%
Portfolio9.5%6.6%1.45−17.0%

Risk-free rate: 4.45% | Kelly leverage (K): 10.8

Adding gold improves the portfolio: higher returns, lower drawdowns, and the optimizer gives it a 22% weight despite a near-total wipeout (−91%) at one point. The diversification benefit is real—gold is uncorrelated with both stocks (−0.02) and bonds (−0.11).


Adding Oil: More Diversification, More Volatility

Oil total return data from Bloomberg, 1991 onwards:

Stocks, Bonds, Gold & Oil (1991–2026)

AssetExcess ReturnVolatilitySharpeMax DrawdownWeight
Stocks8.7%12.4%0.70−53.1%30%
Bonds2.4%6.2%0.39−29.8%44%
Gold5.8%12.2%0.48−55.2%19%
Oil9.2%25.8%0.36−90.0%7%
Portfolio8.1%5.1%1.60−15.0%

Risk-free rate: 2.58% | Kelly leverage (K): 20.6

Oil earns a small allocation (7%) because its volatility is extreme—even holding the 6-month contract, oil suffered a −90% drawdown (and the 1-month contract briefly went negative in 2020). The portfolio improves modestly, but the marginal benefit of each new asset is shrinking.


The Problem: Regime Instability

The full-sample weights look sensible. But investors don't live in the full sample. They live through individual decades. And when we split the data into roughly 10-year periods, the optimal weights become dramatically different.

Stocks & Bonds by Decade (since 1934)

PeriodStock WeightBond WeightPortfolio SharpeMax DDKelly K
1934–440%100%5.20−2.4%806
1944–5411%89%2.32−2.6%100
1954–6478%22%1.51−16.0%14
1964–74Short 22%Short 78%1.46−13.8%−7
1975–85305%Short 205%0.97−60.1%2
1985–9535%65%1.99−8.5%16
1995–0533%67%1.52−5.5%15
2005–1530%70%1.22−9.3%19
2016–26142%Short 42%1.15−30.1%6

The "optimal" stock weight ranges from 0% to 305%, with bonds as high as 100% long to 78% short. The portfolio required shorting bonds in the 1970s and 2010s, and levering equities 3-to-1 in the 1980s. No real-world investor would follow these weights—and that's precisely the point. The data is telling us that the optimal stock/bond mix is completely regime-dependent.

Adding Gold (since 1971)

PeriodStock WeightBond WeightGold WeightPortfolio SharpeMax DDKelly K
1971–82142%Short 1,594%1,552%—*−100%0.2
1982–9336%102%Short 38%2.10−15.9%11
1993–0440%72%Short 12%1.50−8.4%13
2004–1526%53%22%1.54−11.6%25
2015–2659%Short 28%69%1.65−15.1%13

The 1971–82 portfolio had absurd weights driven by gold's 19% excess return during the inflationary crisis. Realistic constraints would be necessary.

Gold's weight swings from −38% to +1,552%. The 1970s make gold essential; the 1980s–90s make it a portfolio drag. The 2015–26 period gives gold a 69% weight as bonds turn negative.

Adding Oil (since 1991)

PeriodStock WeightBond WeightGold WeightOil WeightPortfolio SharpeMax DDKelly K
1991–0331%77%Short 25%17%1.89−12.5%17
2003–1425%45%24%6%1.66−13.1%25
2014–2653%Short 8%55%0%1.57−14.1%14

Even with four assets, the weights remain unstable. The 2014–26 period shorts bonds, gives gold 55%, and nearly eliminates oil. Every decade demands a different portfolio.


The Core Insight

The 60/40 portfolio is not a law of nature. It is the average of many incompatible regimes:

RegimeEraRequired Portfolio
Financial repression1934–54Nearly 100% bonds
Post-war growth1954–6478% stocks
Inflation crisis1964–82Short bonds, long gold
Great moderation1982–00Balanced 60/40
ZIRP/QE era2000–15Heavy bonds
Post-QE inflation2015–26Long gold, short bonds

A static allocation works by accident when the future resembles the average past. When it doesn't—as in the 1970s or the 2020s—it fails.


What Comes Next?

The instability of long-only weights points toward a need for tactical signals—something that can adjust exposure based on the current regime rather than hoping the long-run average holds. In the next section, we examine carry-based strategies that do exactly that.

Data sources: Robert Shiller (equity, bond total returns), FRED (CPI, risk-free rate), Commodity Markets Outlook (gold before 1971), Bloomberg (oil, gold futures since 1991). Kelly leverage K = μ/σ² represents the optimal fraction of capital to bet given the risk-return profile.*

Updated Futures Comparison Table (as of late May 2026)


Standard & Mini/Micro Contracts

AssetContractSizeNotional Value (approx.)Tick ValueInitial Margin (approx.)Daily Volume (typical)Open InterestNotes
EquityES (E-mini)$50 × S&P 500 Index~$377,000 (at ~7540)$12.50$24,000 – $26,5001.0 – 2.0+ millionVery HighBenchmark contract
EquityMES (Micro)$5 × S&P 500 Index (1/10th)~$37,700$1.25$2,400 – $2,700Hundreds of thousandsHighExcellent for smaller sizing
NotesZN (10y)$100,000 face value~$110,000 (at ~110)$15.625$2,000 – $2,300800k – 1.5+ millionVery HighHighest liquidity Treasury future
GoldGC100 troy oz~$452,000 (at ~$4520/oz)$10.00$30,000 – $36,000100k – 300kHighGlobal benchmark
GoldMGC (Micro)10 troy oz (1/10th)~$45,200$1.00$3,000 – $3,500Moderate (tens of k)GoodPopular smaller alternative
OilCL (WTI)1,000 barrels~$92,000 (at ~$92/bbl)$10.00$12,000 – $13,000300k – 1M+250k – 400kMost liquid energy contract
OilMCL (Micro)100 barrels (1/10th)~$9,200$1.00$1,200 – $1,500Tens of thousandsModerateGood for precise/smaller risk
CopperHG25,000 pounds~$162,500 (at ~$6.50/lb)$12.50$12,000 – $14,00030k – 60k+140k – 180kMajor industrial metal benchmark
CopperMHG (Micro)2,500 pounds (1/10th)~$16,250$1.25$1,200 – $1,500Low to moderate (thousands)GrowingGood for smaller accounts / precision hedging

Margin notes:

  • Margins are exchange-set (CME) and fluctuate with volatility. Figures above are approximate initial (speculative) margins. Maintenance margins are usually 10-20% lower.
  • Many brokers offer lower day-trading margins (e.g., $400–$500 for ES/MES, $300 for ZN).
  • ZN has exceptionally low margin relative to notional — one of the most capital-efficient contracts.

Oil Rolling Practices (Most Delicate Among the Four)

Oil (CL) is by far the most sensitive to rolling mechanics due to its strong tendency toward backwardation (near-term prices > longer-dated) or contango, and because physical delivery/ storage costs create meaningful roll yield.

Common professional approaches:

  • Front-month roll (most common for signals/backtests): Roll from the expiring front month into the next active contract 5–10 days before expiry (or first notice day) to avoid delivery risk and liquidity drop-off.
  • Constant maturity / continuous contract: Many systematic traders use a weighted roll (e.g., gradually shifting from CL1 to CL2 over several days) or a fixed tenor like "CL 2nd month" to reduce noise.
  • Calendar spread awareness: Monitor the CL1-CL2 or CL6 spread closely. In backwardation (common in tight supply), rolling long positions captures positive roll yield. In contango, it creates drag.
  • Avoid last few days: Liquidity dries up and basis can become erratic near expiry.
  • Micro (MCL): Same logic but with smaller impact per contract — useful when scaling positions precisely around roll dates.