Diversification: why spreading your investments reduces risk

Diversification means spreading money across investments that don't all rise and fall together, so that a disaster in any one of them can't sink the whole portfolio. It is the closest thing investing has to a free lunch: done well, it lowers the ups and downs of a portfolio without necessarily lowering its expected return.

The idea is old, but why it works is often misunderstood. Owning many things is not enough. Twenty technology stocks behave much like one big technology stock. What matters is how the holdings move relative to each other, which investors measure with correlation. This guide explains correlation, works through the math of a two-asset portfolio with real numbers, shows which risks diversification removes and which it can't, and uses five years of market data to show what correlations actually look like.

What diversification does

Imagine two companies: one sells umbrellas, the other sunscreen. In a rainy year the umbrella maker thrives and the sunscreen maker struggles; in a sunny year the reverse. Each stock alone is a coin flip on the weather. Own both and the weather mostly stops mattering: whatever happens, one business is having a good year. The portfolio's return is steadier than either holding's.

Real companies are never that neatly opposite, but the principle holds whenever their fortunes are not identical. A share of any one stock carries two kinds of risk.

• Unsystematic risk (also called idiosyncratic, specific or diversifiable risk) comes from things that affect one company or industry: a failed drug trial, a product recall, a fraud, a lost lawsuit, a new competitor. These events are largely unrelated from company to company, so in a portfolio of many companies they tend to cancel out.
• Systematic risk (market risk) comes from things that affect nearly everything at once: recessions, interest-rate changes, inflation, financial crises, pandemics. No number of stocks removes it, because all of them are exposed.

Diversification is the tool for the first kind. Asset allocation, spreading money across stocks, bonds and cash, is the main tool for softening the second (see Asset allocation: how stock, bond and cash mixes have behaved).

Correlation, the key ingredient

Correlation measures how consistently two investments move together, on a scale from −1 to +1:

• +1: perfect lockstep. When one rises, the other always rises in proportion. Combining them reduces nothing.
• 0: no consistent relationship. One's moves tell you nothing about the other's.
• −1: perfect opposites. Combined in the right proportions, their swings cancel completely.

Most stocks are positively correlated with each other, often between about 0.2 and 0.7, because they share the same economy. That is still far enough below +1 to help.

The two-asset math

Volatility, the standard deviation of returns, is the usual yardstick for how much an investment swings. For a portfolio of two assets with weights w₁ and w₂, volatilities σ₁ and σ₂, and correlation ρ:

The return is a simple weighted average. The volatility is not: the last term shrinks as the correlation falls. That asymmetry is the whole point.

With real numbers

Over the five years of weekly data in the chart below, Apple (AAPL), Microsoft (MSFT) and Exxon Mobil (XOM) each had an annualized volatility of about 28% (27.5%, 27.8% and 27.8%). Their correlations were very different:

How many holdings are enough?

If each additional stock brings some company-specific risk that partly cancels the rest, how far does that go? A simple model answers it. For an equal-weighted portfolio of N stocks that each have volatility σ and an average pairwise correlation ρ:

With σ = 30% and ρ = 0.25, one stock has 30% volatility, ten stocks about 17.1%, thirty about 15.7%, and no number of stocks gets below 15% (30% × √0.25). Two lessons follow. The first handful of holdings do most of the work. And the floor is set by the correlation, which is market risk: adding a fiftieth similar stock barely helps, while adding something that behaves differently, such as bonds, can.

The model is a simplification. Real stocks differ in volatility and correlation, and correlations rise in crises. But the shape, a steep early drop and then a long flat tail, shows up in actual portfolios too.

What real correlations look like

The matrix covers five years of weekly price moves for the S&P 500 ETF (SPY), three large stocks, a gold fund (GLD) and two bond funds (AGG and TLT). A few patterns stand out:

• Single stocks versus the market. Apple and Microsoft each had a correlation of 0.67 with SPY. They are big parts of the index, and the market's moves explain a large share of theirs. Exxon's correlation with SPY was only 0.16; energy stocks followed oil prices more than the broad market in this period.
• Same sector is not the same stock. Even Apple and Microsoft, both giant technology companies, had a correlation of just 0.40 with each other. Their company-specific news differed enough to matter.
• Bonds and gold. AGG's correlation with SPY was 0.27, TLT's 0.11 and gold's 0.17. All three moved largely independently of stocks week to week.
• Two versions of the same thing. AGG and TLT, both Treasury-heavy bond funds, correlated at 0.92. Owning both adds little diversification.

Volatility differed too: AGG swung about 5.8% a year, SPY 16.3%, and each single stock about 28%.

Ways to diversify

• Across companies: many stocks instead of a few. A broad index fund does this in one purchase (see Index investing: passive vs. active funds and what the data shows and What is an ETF? How exchange-traded funds work).
• Across sectors and industries: technology, health care, energy, financials, consumer goods. Sector concentration is the most common hidden risk in a "diversified" stock list.
• Across company sizes and countries: large and small, US and international.
• Across asset classes: stocks, bonds, cash and, for some investors, real assets. This is asset allocation.
• Across time: investing gradually, and holding for years, so one bad entry date doesn't define the outcome.

Pitfalls

Diversifying by count, not by behavior. Ten funds that all track large US stocks are one holding with ten tickers. Check what each owns.

Owning your employer's stock on top of your job. Your salary and your shares then depend on the same company: a doubly concentrated position.

Assuming correlations are fixed. They drift, and in market panics many risky assets tend to fall together. Build in room for that.

Over-diversifying into expensive overlap. Owning dozens of overlapping funds raises costs and complexity without reducing risk further.

Confusing diversification with protection from loss. A diversified stock portfolio still falls in a bear market. Diversification removes company-specific risk, not market risk.

How Finabase helps

The heatmap shows the whole market sized by market capitalization and colored by the day's move, which makes concentration visible: on most days whole sectors turn the same color together. The comparison tool charts several symbols rebased to the same start, a quick way to see whether two holdings really behave differently, and portfolios show what each holding weighs in an allocation view. Finabase is a research tool; it does not recommend portfolios.

Frequently asked questions

How many stocks do I need to be diversified?

There's no single number. In the simple model above most of the benefit arrived within the first 10 to 20 similar stocks, but real diversification depends on how different the holdings are, not just how many. One broad index fund holds hundreds or thousands.

Does diversification guarantee I won't lose money?

No. It reduces the damage any single holding can do, but market-wide declines still hit a diversified portfolio. In 2008, an S&P 500 fund holding 500 companies lost about 37% with dividends reinvested.

What is a good correlation for diversification?

The lower the better, all else equal. Anything well below +1 helps; near zero helps a lot; negative correlation helps most but is rare and often unstable among assets that are expected to earn a positive return.

Is an S&P 500 fund diversified?

Across large US companies and sectors, yes. It is not diversified across company sizes, countries or asset classes, and because it is weighted by market value, its largest companies carry a large share of the fund.

Summary

Diversification lowers risk because investments that don't move in lockstep partly cancel each other's swings, so the portfolio's volatility is below the average of its parts while its expected return stays the weighted average. Correlation is the lever: at zero correlation two 20%-volatility assets make a 14% portfolio, and real data from 2021 to 2026 showed a technology-plus-energy pair diversifying better than two technology giants. Owning more stocks removes company-specific risk quickly and then stalls at the market's own risk, which only different asset classes can soften, and even those can fall together in a year like 2022.