Portfolio diversification: reduce risk without sacrificing returns

Harry Markowitz won the Nobel Prize in Economics in 1990 for mathematically proving that diversification is 'the only free lunch in finance'. Combining assets with low or negative correlation reduces the total portfolio risk without necessarily reducing expected returns. But poor diversification can create a false sense of security.

Holding 50 stocks is not diversification if they all belong to the same sector or the same country. True diversification works across multiple dimensions simultaneously: sectors, geography, asset classes and market cap. Done well, it lets you sleep through corrections without sacrificing long-term growth.

Types of risk: systematic vs. specific

Before diversifying, it is essential to understand what type of risk diversification actually protects against:

The practical conclusion: with 15-20 stocks from different sectors, specific risk is nearly completely eliminated. Adding more stocks has diminishing marginal returns in terms of risk reduction. What you can never eliminate is market risk — when the S&P 500 falls 40%, almost everything falls.

The 4 dimensions of effective diversification

Correlation: the key concept of diversification

Correlation measures how two assets move relative to each other. It ranges from -1 to +1:

• •Correlation +1: both assets move exactly together. Adding the second provides zero risk reduction.
• •Correlation 0: independent movements. Adding the second does reduce risk.
• •Correlation -1: opposite movements. The perfect combination to reduce volatility (though this rarely exists in practice).

Approximate historical correlations: US stocks vs. US bonds: -0.2 to +0.2 (low, useful). US stocks vs. European stocks: +0.7 to +0.85 (high, limited protection). Stocks vs. gold: -0.1 to +0.2 (low, good crisis diversifier). Individual stock vs. its own sector: +0.8 to +0.95 (very high, almost no diversification benefit).

The most common mistake: illusory diversification

Owning 30 tech stocks is not diversification. In a tech correction (like 2000-2002 or the 2022 Nasdaq crash), they will all fall together with correlations near 1. Illusory diversification gives a false sense of security without the real protection.

Other examples of illusory diversification: holding European and American companies that do exactly the same business (high de-facto correlation), or mixing investment funds that actually hold the same 50 tech companies in their portfolios (position overlap).

How much to diversify? Markowitz's efficient frontier

Markowitz showed that there is an optimal combination of assets (the "efficient frontier") that maximizes expected return for each level of risk taken. In practice, for an individual investor this translates to:

• •Conservative portfolio (horizon < 5 years): 30-40% equities + 50-60% bonds + 10% alternatives (REITs, gold). Prioritizes capital preservation.
• •Balanced portfolio (5-10 year horizon): 60% equities + 30% bonds + 10% alternatives. Balance between growth and protection.
• •Aggressive portfolio (horizon > 10 years): 80-90% equities (global, diversified by sector and cap) + 10-20% bonds or alternatives. Maximizes long-term growth.

Correlation in crises: when everything falls together

One of the most frustrating phenomena for investors is that in major crises (2008, March 2020) correlations between risk assets spike sharply toward 1. Stocks across different sectors and countries fall together because panic and the search for liquidity affect everything indiscriminately.

Assets that historically maintain low or negative correlation in crises: high-quality Treasury bonds (flight to safety), gold (store of value), the US dollar (reserve currency) and cash. That is why a portfolio with some percentage in these assets cushions the falls, even if they seem "inefficient" in bull markets.

Frequently asked questions about diversification

How many stocks do I need to be well diversified?

Academic research shows that between 15 and 25 stocks from different sectors eliminates practically all specific risk. Beyond 25-30 stocks, the additional diversification benefit is marginal and adds monitoring complexity. What matters most is not the total number but that they are spread across sectors and geographies with low correlations between them.

Are global ETFs already well diversified on their own?

An MSCI World or S&P 500 ETF is well diversified at the individual company and sector level. However, it remains concentrated in equities (100%) and developed markets. For real diversification you should complement it with bonds or alternative assets to reduce risk in down years. A 100% S&P 500 portfolio can fall 50% in a severe crash.

Does diversification reduce returns?

In theory no: Markowitz showed that diversifying correctly can maintain expected return while reducing risk. In practice, adding bonds or defensive assets does reduce expected return in bull markets, but also reduces volatility and maximum drawdowns. The question is whether you can psychologically withstand seeing your portfolio fall 50% without selling. If not, diversification with bonds that reduces that drawdown to 25-30% is worth the cost in return terms.