You have heard that owning one stock is reckless and owning the whole market is safe. Somewhere between those is a number, and nobody ever tells you what it is. Five? Twenty? A hundred? Researchers have been arguing about this since the 1960s, and the reason they disagree teaches you more than any single number would.
Two Kinds of Risk, and Only One of Them Goes Away
Every stock carries two separate dangers. The first is company-specific risk: the factory burns down, the CEO is indicted, a rival launches a better product. The second is market risk: a recession, a rate shock, a war, something that drags nearly every stock down at once.
Diversification works on the first one only. Add a second company and the odds that both factories burn down in the same year are tiny. Add a fiftieth and there is almost nothing left to cancel out, because the risk still standing is the kind that hits all fifty together.
That is why a finite number exists at all. The risk you can remove shrinks toward zero as holdings pile up. The risk you cannot remove sits underneath as a floor. The whole question is how quickly you reach that floor.
The Classic Answer: Somewhere Between Ten and Forty
John Evans and Stephen Archer published the first widely cited attempt in 1968, in the Journal of Finance. They built randomly chosen portfolios of increasing size and measured how much the spread of returns fell. It fell fast, then stopped falling. They wrote that their results raised doubts about the justification for going beyond about ten securities.
The arithmetic is plain. Put $10,000 into one stock and a total failure costs you $10,000. Spread it across 10 stocks and the same failure costs $1,000. Spread it across 30 and it costs $333. Going from 1 holding to 10 saved you $9,000. Going from 10 to 30 saved another $667.
Meir Statman challenged the ten-stock figure in 1987, in the Journal of Financial and Quantitative Analysis. He weighed the cost of buying more stocks against the risk each one removed. His conclusion was at least 30 stocks for an investor who borrows to invest, and 40 for one who also holds safe bonds.
Why Later Research Puts the Number Higher
Those studies measured one thing: how bumpy your returns are. Later work asked harder questions and got bigger answers.
In 2001, John Campbell, Martin Lettau, Burton Malkiel and Yexiao Xu studied US stock volatility from 1962 to 1997. Individual companies had grown more jumpy relative to the market as a whole, so it took more of them to get the same smoothing. The diversification that 20 stocks bought in the 1960s took roughly 50 stocks by the late 1990s. That trend has not run one way since. Firm-level volatility spiked in the dot-com bubble, fell through the 2000s, then spiked again in 2008 and in 2020. The number moves with the era.
Research by Dale Domian, David Louton and Marie Racine in 2007 measured something else: the chance of finishing a 20-year holding period below a target, which they call shortfall risk. By that yardstick, risk kept falling as portfolios grew well past 100 holdings, long after the bumpiness of the returns had flattened out.
Both sides are right about what they measured. Ten to thirty holdings does most of the smoothing. Landing near the market's return is a different goal, and it takes more.
The Skew Problem: A Few Giant Winners Carry the Index
Hendrik Bessembinder's 2018 paper in the Journal of Financial Economics shows why. He examined more than 25,000 US common stocks from 1926 to 2016. Four out of every seven returned less over their lifetimes than one-month Treasury bills would have. The best-performing 4% of companies account for the entire net gain of the US stock market across that span, because the rest, taken together, merely matched Treasury bills. The top 86 companies alone, about a third of one percent of the sample, account for roughly half of that gain.
Stock returns are not scattered evenly around an average. Most are mediocre or worse, and a few are extraordinary. Pick 20 at random and you will probably smooth out your volatility. You will also probably miss every one of the huge winners. Missing them is how a basket of real companies ends up far behind the index built from the same companies.
Twenty Stocks in One Sector Is Not Twenty Stocks
Counting holdings assumes each one is a separate bet. Often they are not. Correlation is the word for how closely two investments move together, and that, rather than the count, is what actually matters.
Own 20 regional banks and you own one bet on interest rates wearing 20 name tags. Own 20 oil producers and a drop in the crude price marks all 20 down in the same week. You diversified away the company-specific risk and replaced it with sector risk, which acts like market risk inside your portfolio.
The same logic runs outward through sectors, asset classes and countries. One axis gets almost no attention: time. A hundred companies all bought on the same Tuesday still leaves you holding a single purchase price. Buying at intervals across years gives you many entry prices instead of one. No holding count will ever show that.
What Happens When One Winner Takes Over
Say you put $10,000 into 10 stocks, $1,000 each. Over ten years, nine go nowhere and one grows tenfold to $10,000. The portfolio is worth $19,000 and that single company is 53% of it. You still own ten stocks. You are no longer diversified, and the count never said a word.
Here the trade-off is real and has no clean answer. Selling part of the winner and spreading the money back restores the balance, at a price. You are trimming the one position that worked, and in a taxable account the sale triggers a capital gains bill. Leaving it alone keeps you in the winner and quietly turns your savings into a bet on one company. Bessembinder's finding cuts both ways here: the giant winners are what produces the market's return.
Position sizing means deciding in advance what share of the portfolio any one holding may reach, then checking. Checking is the skipped step, which is how portfolios drift into concentration with nobody ever deciding anything.
The Limit Is How Many You Can Actually Follow
Holdings carry a cost no risk model prices. Each company you own individually is homework: results four times a year, an annual report, news that might undo your reason for buying. Twenty companies means 80 earnings reports a year.
A beginner picking individual stocks is caught between two bad options. Own 8 and company-specific risk is still live. Own 40 and you cannot know 40 businesses, so you end up trading on headlines.
The way out is not clever. A single broad index fund holds hundreds or thousands of companies and asks nothing of you afterward. So the how-many question is mostly a question for people who have chosen to pick stocks themselves.
When Diversification Is Only a Feeling
You can also overshoot, and that failure looks like success.
Owning several funds feels like several decisions. Frequently it is one. A total US market fund, a large-company fund and a technology fund all hold the same giant businesses near the top. Buying all three concentrates you in those businesses instead of spreading you out. Every fund publishes its holdings. Line up the top ten from each fund you own and the overlap stops being invisible.
Holding 200 individual stocks has the opposite problem. You have rebuilt an index fund by hand, with 200 positions to track and more trading costs, and no advantage to show for the effort.
