Benjamin Graham's defensive investor: the 50/50 stock-bond rule

Benjamin Graham told the defensive investor to keep between 25% and 75% of his funds in common stocks and the rest in high-grade bonds, with an equal 50/50 split as the standard, in Chapter 4 of The Intelligent Investor (fourth revised edition, 1973). In our backtest on US data from 1928 to 2025, a 50/50 mix of stocks and 10-year Treasuries, rebalanced each year, compounded 7.8% a year against 10.0% for 100% stocks.

Its worst fall in year-end value was 32.5%, half the 64.8% of stocks. In 2022 the 50/50 mix lost 17.9%, almost as much as stocks, because bonds fell too.

Key takeaways

1. 50/50 exchanged some growth for smaller losses. In our 1928-2025 US test it compounded at 7.8% a year versus 10.0% for shares, with a worst year-end drawdown of 32.5% versus 64.8%, before costs and taxes.
2. Bond protection depends on the shock. The 50/50 mix lost 17.9% in 2022 when both shares and 10-year Treasuries fell; inspect duration, credit and currency before relying on the bond label.
3. The rule is more transferable than the weights. Write a target, a review trigger and a spending reserve before comparing allocations; the euro ETF example below has different bonds and currency exposure from the historical test.

What did Benjamin Graham say about stocks and bonds for the defensive investor?

Graham told the defensive investor to split his money between high-grade bonds and high-grade common stocks, never holding less than 25% or more than 75% in stocks. The bond share is the inverse, between 75% and 25%, and Graham presented an equal division, 50/50, as the standard implied by that range.

The passage sits in Chapter 4, "General Portfolio Policy: The Defensive Investor", of the fourth revised edition of The Intelligent Investor (1973), the last Graham revised himself; the 2003 edition reprints it with commentary by Jason Zweig. We paraphrase it here because we could not obtain the book's text; the Indianapolis Business Journal (3 January 2018) quotes the 25% to 75% sentence.

Graham's defensive investor is defined by what he wants to avoid. As the same column summarises him, he puts the chief emphasis on avoiding serious mistakes or losses, while the enterprising investor is willing to spend time and care selecting securities. The book was first published in 1949; Graham died in 1976.

Graham's bonds were high-grade bonds. High-yield bonds are a different asset: they can suffer large losses inside a crisis, as our comparison of high-yield and government bonds in crashes shows.

Is Graham's 50/50 rule a fixed allocation?

No: 50/50 is the starting point, and Graham allowed the stock share to move anywhere between 25% and 75% on the investor's judgement of market prices.

In the same Chapter 4 passage, he described the traditional reason for raising the stock share as the bargain prices left by a protracted bear market, and called for going below 50% in stocks when the market had become dangerously high. The IBJ column reads the rule the same way: closer to 25% stocks when they are highly priced, towards 75% when they sell at a discount.

The band is about prices, not age. Age-based rules come from other writers; Jack Bogle's rule of holding your age in bonds, for example, which he called a rule of thumb. A defensive investor who keeps to 50/50 must still sell some of whichever asset has risen to restore the split, which means buying stocks after they have fallen.

How did a 50/50 portfolio compare with 100% stocks from 1928 to 2025?

In our backtest, the 50/50 mix compounded 7.8% a year against 10.0% for 100% stocks, and its worst year-end drawdown was 32.5% against 64.8%. We used Aswath Damodaran's annual US returns: the S&P 500 with dividends reinvested for stocks, and a 10-year US Treasury bond for high-grade bonds. Each mix is reset to its target at every year end, with no costs, taxes or fees.

1928 to 2025, in USD, CAGR / CAGR after inflation / worst year-end drawdown: 100% stocks 10.0% / 6.8% / -64.8%; 75% stocks 9.1% / 5.8% / -50.3%; 50/50 7.8% / 4.6% / -32.5%; 25% stocks 6.3% / 3.2% / -17.9%. Source: our calculation from Damodaran, Historical Returns on Stocks, Bonds and Bills (January 2026 update, sheets "Returns by year" and "Inflation Rate"); chart data (CSV); annual backtest data (CSV). Year-end values miss falls within a year, so drawdowns measured more often could be deeper.

Each step towards bonds lowered both the compounded return and the worst fall. $100 invested at the start of 1928 grew to $1,157,009 in stocks and $158,972 in the 50/50 mix by the end of 2025, a 7.3-fold difference before costs and taxes.

The gap between average and compounded returns shows why the path matters. Stocks averaged 11.9% a year but compounded 10.0%, because large losses cut the base for later gains; the 50/50 mix averaged 8.3% and compounded 7.8%. Its annual returns swung about half as much as stocks', a standard deviation of 10.5% against 19.4%.

The 1928 start includes the Great Crash. From 1973, the year of Graham's last edition, to 2025, stocks compounded 10.9% a year with a worst year-end drawdown of 37.4% (2000-2002), and the 50/50 mix 8.9% with a worst drawdown of 17.9% (2022).

After inflation, both mixes spent more than a decade below a previous peak. In real terms the 50/50 mix lost 31.7% from the end of 1972 to the end of 1974 and did not regain its 1972 level until 1985; stocks stayed below their 1999 real peak until 2013.

Did bonds protect the 50/50 portfolio in every crash?

No: Treasuries rose in 1929-1932, 2000-2002 and 2008, but in 2022 they fell almost as much as stocks. The cushion came from holding half the money outside stocks, and from bond prices that happened to rise when stocks fell.

Cumulative return, 100% stocks then 50/50: 1929-1932 -64.8% and -32.5%; 1973-1974 -36.5% and -16.6%; 2000-2002 -37.4% and -2.9%; 2008 -36.6% and -8.2%; 2022 -18.0% and -17.9%. Source: our calculation from Damodaran's annual returns; chart data (CSV). 10-year Treasuries alone gained 15.5% over 1929-1932, 41.8% over 2000-2002 and 20.1% in 2008, and lost 17.8% in 2022.

In 1973-1974 the 50/50 mix lost 16.6% before inflation, but inflation was high, so its real loss was larger. 2022 is the clearest counterexample: rising yields pushed down bonds and stocks in the same year. Our article on why bonds protected in 2008 but failed in 2022 explains the mechanism.

What do people get wrong about Graham's stock-bond rule?

The common misreadings turn a flexible band into a fixed formula, or confuse it with Graham's other work.

Graham's influence ran through his stock analysis as much as his portfolio policy. Seth Klarman named his 1991 book, Margin of Safety, after the principle Graham put at the centre of value investing, as our study of Klarman's cash allocation at Baupost notes, and Charlie Munger's Daily Journal portfolio shows how far his partner moved towards concentration.

What did the checker return for the ETF illustration?

A European 50/50 ETF illustration received C overall in copio's checker on 3 November 2026. This is a separate investment from the US stock/10-year Treasury backtest above: its global government bonds are hedged to euros and its US equity fund is unhedged.

The score describes these inputs under the checker's assumptions; it does not grade Graham's rule or its historical return. These are recorded model outputs with data limitations: equivalent S&P 500 ETFs returned different diversification grades in our checker comparison. Use the figures to inspect assumptions, not to rank the named investor. The selected UCITS funds are:

Fees: A, 0.16% a year. That is €16 a year for every €10,000 invested in the two funds; the S&P 500 fund charges 0.07% and the bond fund 0.25%.

Long-term assumptions: C, −1.68 points a year. Under copio's long-term capital market assumptions, the illustration's expected compounded growth is 6.14% a year, 1.68 points below the 7.82% of the model's growth-optimal portfolio, before inflation. The model assigns lower expected growth to the government-bond half; this is an assumption about the selected funds, not a forecast from the 1928-2025 backtest.

Historical efficiency: F, −5.9 points a year. Over the past ten years, MSCI World combined with euro money-market funds at the same measured risk earned 5.9 points a year more than the illustration's two funds. The hedged bond fund lost 10.3% over the ten years to October 2026, including 14.73% in the year to October 2022, by our compounding of DWS's yearly returns.

Crises: B, −19.77%. In a 40% fall in US equities, passed on to each asset class through its long-term sensitivity to US stocks, the checker estimates the illustration would lose 19.77%. Half the portfolio sits outside stocks, so the estimated loss is about half the shock.

Diversification: D, 2 growth sources. If US large caps stopped growing, the illustration's expected growth above cash would fall from 4.3% to 0.0% a year. The model assigns almost no growth above cash to government bonds in this scenario. This describes its assumptions about growth, not a proof that the bonds provide no diversification or liquidity benefit.

The illustration differs from our backtest in three ways. Its bond fund holds developed-market government bonds of all maturities, hedged to euros and led by US Treasury notes, where the backtest uses only 10-year Treasuries in dollars; the illustration uses it because it is the checker's reference fund for government bonds outside the euro area.

The S&P 500 fund is unhedged, so a euro investor also carries dollar risk on the stock half. And Graham's moves within his 25% to 75% band, on his judgement of market prices, are not modelled. You can enter the same illustration, or your own portfolio, in the free portfolio checker.

What does copio think of this portfolio?

Graham's useful legacy is a rule for controlling exposure before markets move. In our annual US data, 50/50 reduced the worst year-end drawdown from 64.8% to 32.5%, while growth fell from 10.0% to 7.8% a year before costs and taxes.

That is a measured trade-off, not proof that bonds fail to diversify or that 50/50 is right for every household. The 2022 result shows why the type of bond matters: both halves fell almost equally. Copio's question is whether a portfolio can sustain its owner's spending and losses while continuing to compound.

Three checks for your portfolio

1. Write a rebalancing rule you could follow during a fall. Record your target equity share, the deviation that would trigger a review, and whether new contributions can restore it before you sell anything. Separate scheduled rebalancing from a decision to change the target; Graham's market-valuation judgement is not reproduced by our fixed-weight backtest.
2. Read the bond holdings before treating them as protection. Record duration, government versus corporate exposure, currency and hedging from each fund's factsheet. Replay both a recessionary crash and an inflation/rate shock such as 2022; compare the whole portfolio's losses and recovery, after fees. A euro-hedged global bond fund and a US 10-year Treasury series are different tests.
3. Turn a percentage drawdown into a spending decision. Apply the historical 32.5% year-end fall to the amount you plan to invest, then list withdrawals that would still need funding during that fall. Keep that scenario separate from a loss limit: intra-year falls can be deeper and future losses can exceed the sample. Compare a different equity weight, a spending reserve and an explicit hedge on the same horizon, including hedge costs in quiet years.

The portfolio checker can help explore your own fund mix; the allocation guide sets out the comparisons. Copio provides educational analysis and does not choose an allocation or place trades for you.