How Many Holdings You Actually Need to Cut Portfolio Risk
Investors tend to believe one of two opposite myths: that owning more stocks always means less risk, or that ten stocks is basically as safe as five hundred. The mathematics of diversification shows both instincts are wrong, and reveals a specific, calculable number of holdings where most of the real benefit has already been captured.
Two kinds of risk inside every stock
The total volatility of any single stock can be split conceptually into two components. Firm-specific risk, sometimes called idiosyncratic or diversifiable risk, comes from events unique to that company: a product recall, a leadership change, a lost contract, a factory fire. Systematic risk, sometimes called market risk, comes from forces that hit essentially every company at once: rising interest rates, a recession, a broad shift in investor sentiment. When you combine stocks whose returns are not perfectly correlated with each other, the firm-specific pieces partly offset one another, since one company's bad quarter is statistically likely to be paired with another's ordinary or good one. Systematic risk cannot be cancelled this way, because by definition it moves every holding in the same direction at once, regardless of how many different companies you own.
This split is the entire logical foundation of diversification: it works, and works well, against one specific category of risk, and does essentially nothing against the other. Understanding which category a given risk falls into determines whether adding more holdings will actually help. A useful mental shortcut: firm-specific risk is the kind of news that shows up in a headline about one company, systematic risk is the kind of news that shows up in a headline about the economy, and only the first kind responds meaningfully to owning more, sufficiently different, companies.
The math of adding holdings
For a portfolio of n equally weighted stocks, portfolio variance can be written as: portfolio variance = (average variance / n) + ((n-1)/n) × average covariance. The first term, driven by firm-specific risk, shrinks toward zero as n grows large. The second term, driven by how the stocks move together, does not shrink to zero; it converges instead toward the average covariance itself as n becomes very large, which is the mathematical floor systematic risk imposes on any stock-only portfolio, however many names it holds.
Suppose the average stock in your candidate universe has a standard deviation of 30% (variance of 0.09), and the average pairwise correlation between stocks is 0.30, giving an average covariance of 0.30 × 0.30 × 0.30 = 0.027, since covariance between two equally volatile assets equals their correlation multiplied by both standard deviations. With a single stock, variance is simply 0.09, a standard deviation of 30%. With n = 20 equally weighted stocks: portfolio variance = (0.09/20) + (19/20) × 0.027 = 0.0045 + 0.02565 = 0.03015, a standard deviation of √0.03015 ≈ 17.4%, already a large reduction from the single-stock figure. Extending to n = 100: portfolio variance = (0.09/100) + (99/100) × 0.027 = 0.0009 + 0.02673 = 0.02763, a standard deviation of ≈ 16.6%, only about 0.8 percentage points lower than the 20-stock case despite holding five times as many names.
That comparison is the whole practical lesson in numbers: moving from 1 stock to 20 cut risk from 30% down to 17.4%, a reduction of roughly 12.6 points, while moving from 20 stocks all the way to 100 cut risk by less than a single additional point. Most of the diversification benefit available from adding names is captured within the first 20 to 30 holdings; beyond that range, an investor is mostly adding complexity and trading costs, not meaningfully cutting risk.
A second numeric case: lower average correlation
The average covariance figure in the diversification formula is itself a product of two things, average variance and average correlation, and improving diversification quality often means lowering the correlation input rather than simply adding more names. Return to the earlier example, an average stock variance of 0.09, but now assume a more genuinely diversified universe, spread across sectors and geographies, achieves an average correlation of 0.15 instead of 0.30, giving an average covariance of 0.15 × 0.30 × 0.30 = 0.0135, exactly half the earlier figure.
At n = 20 stocks with this lower correlation: portfolio variance = (0.09/20) + (19/20) × 0.0135 = 0.0045 + 0.012825 = 0.017325, a standard deviation of ≈ 13.2%, notably below the 17.4% computed earlier at the higher correlation assumption, using the identical number of holdings. This comparison isolates the two separate levers available to an investor: adding more names reduces risk along one curve, while improving the average correlation among the names already held shifts the entire curve downward, and the second lever, in this example, delivers a bigger risk reduction than doubling the stock count from 20 to a much larger number ever could, exactly the point made in the earlier comparison between 20 and 100 holdings at the higher correlation level.
What crisis periods reveal about correlation
The average covariance used in the formula above is not a fixed constant; it moves with market conditions, and it moves in the direction that matters least conveniently for investors. During calm periods, correlations across sectors and even across asset classes tend to run lower, giving diversification more room to work. During sharp market declines, correlations across most risky assets have historically tended to rise together, as broad, macro-driven selling pressure affects nearly everything at once regardless of company-specific fundamentals; this pattern showed up clearly during the 2008 financial crisis and again, briefly but sharply, during the initial 2020 selloff, when even historically defensive sectors declined alongside the broader market.
The practical consequence is sobering but important: a diversified portfolio's risk reduction is least reliable exactly when an investor needs it most, during a genuine crisis, and most reliable during calmer periods when the reduction matters less. This is not an argument against diversification, which still meaningfully outperforms concentration even during crises, but it is a reason not to expect a well-diversified stock portfolio to avoid a severe decline entirely during a true systemic event; it will still decline, just generally somewhat less than a concentrated alternative would have.
Building a genuinely diversified portfolio
Real diversification requires more than owning many different ticker symbols; it requires owning names whose underlying return drivers actually differ. Five stocks from the same industry, even five different companies with five different names, often share enough common exposure to interest rates, input costs, and sector sentiment that they behave closer to a single concentrated bet than to a genuinely diversified holding. Spreading across sectors, market capitalizations, and geographies moves the average covariance figure in the formula meaningfully lower, which is precisely the lever that improves the risk reduction available at any given number of holdings.
International diversification deserves specific mention here, since domestic and international equity markets have historically shown a correlation to each other that, while positive and rising over recent decades as markets have globalized, has generally remained below the correlation among domestic sectors to each other, offering incremental diversification benefit beyond what a domestic-only portfolio, however broad, can achieve on its own.
For a professional building a portfolio alongside a demanding career, a business owner or partner with equity tied up in the practice, or a physician whose specialty exposes them to region-specific reimbursement risk, is a useful reminder that the diversification math applies to human capital as well as to financial assets. A doctor whose household wealth already includes substantial exposure to a single hospital system, a single specialty's reimbursement trends, or a single local real estate market through a home and a practice building is already carrying a concentrated, undiversified bet before a single share of stock is purchased, and the investment portfolio is one of the few remaining levers available to offset that concentration, which argues for broader, not narrower, diversification in the liquid portfolio specifically for this kind of investor. The instinct to hold what feels familiar, shares of the employer whose paycheck already funds the household, or a fund concentrated in the same industry the investor works in, runs directly against this logic, even though it is one of the most common patterns observed among working professionals building an investment portfolio, and it is worth naming directly rather than assuming familiarity with a company automatically translates into a sound reason to hold more of it.
Actionable breakdown
- Aim for genuine breadth, not just a large count of holdings.
- Hold at least 20 to 30 names, or one broad index fund, for most of the benefit.
- Spread deliberately across sectors, sizes, and geographies.
- Recognize the limits of diversification honestly.
- Expect systematic risk to remain, however many stocks you add.
- Do not expect diversification to prevent losses during a genuine crisis.
- Watch for hidden concentration.
- Check overlapping top holdings across multiple funds you own.
- Avoid mistaking many tickers in one sector for real diversification.
- Add non-equity diversification for the risk stocks alone cannot remove.
- Bonds and cash address systematic equity risk, not just company risk.
Common pitfalls
Investors commonly believe five or six stocks from a single hot sector, semiconductor names during an AI-driven rally, for instance, constitute a diversified position simply because they are five different tickers, when their shared exposure to the same sector-wide catalysts leaves the average covariance, and therefore the risk, close to that of a single concentrated bet.
A second pitfall is overestimating how much extra safety comes from holding hundreds of stocks instead of a few dozen well-chosen ones; the math above shows the marginal benefit past roughly 30 holdings is small, and the added complexity, tracking, and trading costs rarely justify the pursuit of further reduction, and a broad index fund typically delivers this outcome in a single purchase far more cheaply than assembling and monitoring hundreds of individual positions by hand ever could.
A third pitfall is forgetting that correlations spike during exactly the periods when diversification is needed most, leading some investors to conclude, incorrectly, that diversification "failed" during a crisis when in fact it still reduced losses relative to a concentrated alternative, just by less than it typically does in calmer conditions. A fourth pitfall is confusing the number of funds held with the number of underlying holdings; an investor who owns four different actively managed large-cap growth funds may be holding hundreds of fund positions on paper while actually owning a small, highly overlapping set of the same dozen or so mega-cap names across all four, a form of hidden concentration that a quick look at fund count alone would never reveal.
The bottom line
Diversification eliminates firm-specific risk quickly, mostly within the first few dozen genuinely uncorrelated holdings, but it can never remove the market's own systematic risk, and it works least well exactly during the crises when investors most want it to.
Related reading: understanding investment risk, funds and ETFs, international diversification, combining two risky assets, why a broad index fund is hard to beat.