In 30 seconds
- Diversifying cuts the risk specific to each company, not market risk: in 2008 the S&P 500 fell 38.5% and no stock portfolio escaped.
- What lowers risk is low correlation, not the number of stocks: five oil companies moved 28.9% a year between 2021 and 2025; five stocks from different sectors, 16.4%.
- Graham called for between ten and thirty holdings; Statman (1987) concluded that, if they're chosen at random, you need at least 30.
- Beta measures how much a stock moves with the S&P 500: it's a snapshot of recent years, not a fixed property of the company.
- Buffett backs concentration only if you truly understand the businesses you own; if you don't, he recommends an index fund.
Introduction
Between January 2021 and December 2025, a portfolio split equally among five US oil companies (Exxon Mobil, Chevron, ConocoPhillips, EOG and Occidental) had annual volatility of 28.9%. Five stocks, five different companies, and almost the same swings as any one of them. Over those same five years, another five-stock portfolio drawn from different sectors (Coca-Cola, JPMorgan, Nvidia, Duke Energy and Exxon Mobil) moved 16.4% a year, about as much as the S&P 500.
Same number of stocks, nearly half the risk. The difference isn't how many you own, but how they move relative to one another. That's what this lesson is about.
Explanation
Diversifying means spreading your money across investments that don't react the same way to the same news. If a company loses a big customer, sees a drug fail or gets caught cooking the books, your portfolio feels the drop, but the whole portfolio doesn't go down with it. It's the old advice about not putting all your eggs in one basket, with a condition people often forget: five baskets won't help much if they're all riding on the same truck.
The two risks in every stock
Every stock carries two kinds of risk. Specific risk belongs to the company itself: its product, its management, its balance sheet. Systematic, or market, risk is the one they all share: a recession, a rate hike, a panic. Diversification removes much of the first and doesn't touch the second. In 2008 the S&P 500 fell 38.5% (from 1,468.36 points on December 31, 2007, to 903.25 on December 31, 2008), and no stock portfolio escaped, however varied it was. Owning more stocks doesn't protect you from that risk; having less money in stocks, or a long horizon, does. You'll see why in the lesson on the economic cycle.
Correlation, the key piece
Correlation measures whether two assets move together. It runs from −1 (always in opposite directions) to +1 (always in step). At 0, one tells you nothing about the other. Harry Markowitz showed in 1952, in his paper "Portfolio Selection," that a portfolio's risk isn't the average risk of its parts: it depends on how they're correlated. The lower the correlation, the more the swings offset one another and the less the whole portfolio moves.
Beta is a close cousin: it measures how much a stock moves when the S&P 500 moves. A beta of 1 tracks the index; a beta of 0.5 moves half as much; a beta of 2, twice as much. It's a snapshot of the past (usually five years of monthly data) against one specific index, not a fixed property of the company: it changes when the business or the market changes.
How many stocks do you need?
In "The Intelligent Investor" (chapter 5), Benjamin Graham called for "adequate though not excessive diversification": a minimum of ten different issues and a maximum of about thirty. The empirical research landed on similar numbers. John Evans and Stephen Archer, in "Diversification and the Reduction of Dispersion: An Empirical Analysis" (The Journal of Finance, vol. 23, no. 5, December 1968, pp. 761-767), found with random portfolios that the reduction in risk ran out at around ten stocks. Meir Statman revised that figure in "How Many Stocks Make a Diversified Portfolio?" (Journal of Financial and Quantitative Analysis, vol. 22, no. 3, September 1987, pp. 353-363): with randomly chosen stocks you need at least 30, and 40 if you don't borrow to invest. Both studies are talking about random stocks from mixed sectors. Thirty banks don't count as thirty.
Formula
Risk of a two-stock portfolio (standard deviation σ):
σ portfolio = √(w1² × σ1² + w2² × σ2² + 2 × w1 × w2 × σ1 × σ2 × ρ)
w1, w2 = weight of each stock in the portfolio (they add up to 1)
σ1, σ2 = annual volatility of each stock
ρ = correlation between the two, from −1 to +1
Portfolio of n equally weighted stocks with average volatility σ and average correlation ρ:
σ portfolio = √(σ² / n + (1 − 1/n) × ρ × σ²)
Beta = covariance(stock, S&P 500) / variance(S&P 500)
Example
The table below shows the beta of five companies from different sectors: Coca-Cola, JPMorgan, Nvidia, Duke Energy and Exxon Mobil. Read it this way: a beta of 0.5 means that, on average, when the S&P 500 rises or falls 10%, that stock moves 5%. Keep in mind that it's a snapshot of recent years against a single index.
Before you look: which of the five do you think has the highest beta? And the lowest? Write your guesses down, then check.
Now for the other piece, correlation. We calculated it from monthly returns, dividends included, from January 2021 to December 2025 (60 months):
| Coca-Cola | JPMorgan | Nvidia | Duke | Exxon | |
|---|---|---|---|---|---|
| Coca-Cola | 1 | 0.15 | −0.07 | 0.58 | 0.20 |
| JPMorgan | 0.15 | 1 | 0.37 | 0.25 | 0.38 |
| Nvidia | −0.07 | 0.37 | 1 | 0.01 | 0.00 |
| Duke | 0.58 | 0.25 | 0.01 | 1 | 0.15 |
| Exxon | 0.20 | 0.38 | 0.00 | 0.15 | 1 |
Notice two things. Coca-Cola and Duke Energy, two defensive businesses, are the pair that look most alike (0.58). Nvidia looks like almost none of the others: its correlation with Coca-Cola between 2021 and 2025 was slightly negative.
Fill-in-the-blanks example: Coca-Cola and Nvidia at 50% each. Between 2021 and 2025, Coca-Cola's annual volatility was 16.8% and Nvidia's was 50.8%, with a correlation of −0.07.
1. If they moved in lockstep (ρ = 1), the portfolio's volatility would be the simple average: (16.8 + 50.8) / 2 = 33.8%.
2. With their actual correlation: √(0.25 × 0.168² + 0.25 × 0.508² + 2 × 0.25 × 0.168 × 0.508 × (−0.07)) = 26.2%.
3. Your turn: repeat the calculation with Coca-Cola and Duke Energy (volatility of 17.1%, correlation of 0.58). The simple average would be 17.0%. How far does the risk come down? (Answer: to 15.1%; it drops much less, because the two are so alike.)
And with more stocks? Use the second formula with an average volatility of 30% and an average correlation of 0.2, round numbers picked for illustration. With one stock, 30%. With five, 18%. With ten, 15.9%. With thirty, 14.3%. With a thousand, 13.4%. The first ten do almost all the work; past that point you gain tenths of a point. That 13.4% floor is market risk: you can't diversify it away.
Real-data example
| Company | Ticker | Beta |
|---|---|---|
| The Coca-Cola Company | KO | 0.3 |
| JPMorgan Chase & Co. | JPM | 1.0 |
| NVIDIA Corporation | NVDA | 2.2 |
| Duke Energy Corporation | DUK | 0.4 |
| Exxon Mobil Corporation | XOM | 0.2 |
How to read it
How to read it
Rough ranges for a portfolio of individual stocks; they aren't a Kaplio rule, except for the last row.
| What you check | Concentrated | In between | Diversified |
|---|---|---|---|
| Number of holdings | fewer than 10 | 10-20 | 20-30, from different sectors |
| Average correlation among them | 0.7 or more | 0.4-0.7 | below 0.4 |
| Concentration in your Kaplio portfolio | red: one position at 30% or more, the top three at 60%, or one sector at half the portfolio | amber: 20%, 45% or 35% | green: below those levels |
Going past thirty stocks rarely lowers your risk; it just makes the portfolio harder to keep track of. At that point an index fund does the same job with less work. By sector, high correlations are the norm within banking, oil, utilities and semiconductors: the companies share customers, raw materials or the same regulator.
Pitfalls and limitations
1. Fake diversification. Between 2021 and 2025, the five oil companies from the introduction had an average correlation of 0.80, and five big banks (JPMorgan, Bank of America, Citigroup, Wells Fargo and Goldman Sachs), 0.81. Splitting money equally among those five banks only brought volatility down from 27.4%, the average for each bank, to 25.3%. Five big tech companies (Microsoft, Apple, Nvidia, Alphabet and Meta) stayed at 25.7%. The five from different sectors came in at 16.4%.
2. Correlations rise when it hurts most. The figures above come from normal years. In a panic, almost everything gets sold at once, and the protection shrinks just when you need it most.
3. Beta looks backward. A low beta doesn't guarantee small drops: it's a five-year average, and a business can change faster than its beta does.
4. Diversifying doesn't fix a bad purchase. Spreading your money across thirty companies you don't understand just spreads the mistake around.
Case in point: twenty tech stocks in 2000
In March 2000, plenty of people held portfolios of fifteen or twenty stocks and believed they were diversified. Nearly all of them were tech or telecom names. The Nasdaq Composite closed at 5,048.62 points on March 10, 2000, and at 1,114.11 on October 9, 2002: a 77.9% drop. It didn't close above that peak again until April 23, 2015 (5,056.06 points). Cisco, the most valuable company in the world that spring, closed at $80.06 a share on March 27, 2000 (split-adjusted, excluding dividends) and didn't top that price until December 10, 2025, when it closed at $80.25. Twenty stocks, one bet: the "more stocks, less risk" rule breaks down when they all depend on the same thing.
The opposite case breaks the rule too. In his 1993 letter to Berkshire shareholders, Warren Buffett wrote that he and Charlie Munger would "now settle for one good idea a year" and that "a policy of portfolio concentration may well decrease risk if it raises, as it should, both the intensity with which an investor thinks about a business and the comfort-level he must feel with its economic characteristics before buying into it." In the same letter he spells out who that applies to. The "know-something investor," able to understand business economics and find five to ten sensibly priced companies with important long-term competitive advantages, can concentrate. The investor who can't value businesses should own a large number of stocks and space out their purchases, for example by investing periodically in an index fund. Concentrating requires you to recognize a durable competitive advantage and measure it with ROIC. If you're not sure you belong to the first group, you belong to the second.
And diversification doesn't stop at stocks: how the debt cycle and interest rates weigh on your portfolio is covered in how the economic machine works.
Practice on Kaplio
Frequently asked questions
How many stocks should I own to be diversified?
Graham suggested between ten and thirty. Studies using random stocks say something similar: Evans and Archer (1968) found the benefit ran out at around ten, and Statman (1987) raised it to 30 or 40. But the number matters less than correlation: thirty stocks from the same sector are still a single bet.
What is portfolio diversification and how does it reduce risk?
It's spreading your money across assets that don't move together, so a drop in one doesn't sink the whole portfolio. It reduces the risk specific to each company, but not market risk. Between 2021 and 2025, five stocks from different sectors moved 16.4% a year, against 28.9% for five oil companies.
Does diversification reduce systematic risk?
No. It removes much of the risk specific to each company, but not the risk all stocks share: a recession, a rate hike, a panic. In 2008 the S&P 500 fell 38.5% and no stock portfolio escaped. Against that risk, what helps is having less money in stocks or a long horizon.
How do you diversify a stock portfolio?
Pick companies from sectors with different drivers, keep any single position below 20% of the portfolio and check whether your stocks depend on the same thing: interest rates, oil, consumer spending. Look at correlation, not just the number of holdings. If you don't have time to analyze companies, an index fund diversifies for you.