Twenty names can describe a broadly spread equity portfolio or a concentrated bet. Suppose one stock accounts for half the portfolio and the other nineteen share the remainder. The statement has twenty positions, but a severe loss in the largest holding can still dominate the result.
This is the practical question behind the research: how much of the portfolio depends on the same company, industry, country, financing condition or market exposure? Counting securities helps only when their weights and relationships are understood.
What the Research Can Establish
Classic studies investigated how risk changed as randomly selected stocks were added to portfolios. Their answers depended on the sample, weighting method, definition of risk and standard for adequate diversification. A result from one study is not a required holding count for every investor.
Benjelloun's 2010 study, revisiting Evans and Archer, found roughly forty to fifty stocks under its tested methods. That does not make forty a guarantee or twenty a failure. It means the method and adequacy criterion matter. A portfolio can reduce much company-specific risk while still leaving more than a particular investor can afford.
There is also a difference between reducing variance and eliminating catastrophic loss. Many holdings can fall together. Broad exposure does not protect against every market decline, and a low historical standard deviation does not establish that a portfolio can meet a spending obligation.
A Model That Makes the Assumptions Visible
Consider an illustrative one-factor model. Every stock has a beta of one to the same market factor. Assume the factor's annual standard deviation is 20% and each stock has independent company-specific volatility of 30%, uncorrelated with the factor and with other stocks. Ignore tax, fees and trading costs. These are assumed risk inputs, not estimates for actual securities.
Under those assumptions, portfolio variance is the market variance plus company-specific variance multiplied by the sum of squared portfolio weights. For equal weights across N stocks, the sum of squared weights is 1/N. Adding names reduces the independent component; the shared market component remains.
| Illustrative portfolio | Sum of squared weights | Modeled annual standard deviation |
|---|---|---|
| One stock | 1.0000 | 36.1% |
| Twenty equally weighted stocks | 0.0500 | 21.1% |
| One hundred equally weighted stocks | 0.0100 | 20.2% |
| Twenty stocks, one at 50% and nineteen sharing 50% | 0.2632 | 25.2% |
The twenty-stock equal-weight case retains 5% of the single-stock company-specific variance in this model. That is not a 95% reduction in total portfolio volatility. The common factor remains, so the standard deviation falls from about 36.1% to 21.1%, not nearly to zero.
The counterexample is in the last row. Twenty names with one half-weight position have much more modeled volatility than twenty equal weights. The stock count is identical. If company-specific returns are correlated rather than independent, the simple 1/N result also changes. During stress, relationships estimated from ordinary periods may be a poor guide to actual losses.
Keeping the other risk inputs unchanged, suppose the twenty equal-weight positions instead have assumed pairwise company-specific correlation of 0.50. The variance is 0.04 + 0.09 times [0.05 + 0.95 times 0.50], or 0.08725. The modeled standard deviation is about 29.5%, versus 21.1% with independent company-specific returns. This is another hypothetical assumption, not an estimate of a crisis correlation.
This calculation illustrates a mechanism. It does not estimate the risk of a client's twenty stocks or justify replacing their holdings with an equal-weight strategy. That recommendation would require actual exposures, risk estimates, constraints and costs.
Expected Return Is Not an Ending-Wealth Promise
A model can assign the same expected return to several portfolios while producing very different realized outcomes. An expectation is an average across possible results; it is not the amount each investor will receive. Arithmetic average returns and compounded wealth are different quantities.
For a simple example, start with $100. A 50% gain followed by a 50% loss leaves $75. Two years of zero return leave $100. Both sequences have a zero arithmetic average return, but their ending wealth differs. This is a return-path illustration, not a forecast or a complete comparison of concentrated and diversified strategies.
Bessembinder's research on U.S. CRSP common stocks documents pronounced skewness in outcomes and concentrated aggregate wealth creation relative to one-month Treasury bills. A concentrated investor may miss such stocks or hold a large winner. The evidence does not justify promising that a two-stock portfolio and a broad portfolio will arrive at the same destination.
In standard portfolio theory, investors are not generally expected to earn an additional premium merely for accepting diversifiable company-specific risk. That is a model-based expectation. It does not guarantee a premium for every systematic exposure, the same expected return for all stocks, or a positive realized market return over the client's horizon.
Sector concentration also changes more than company-specific risk. A technology fund may expose the investor to shared valuation, business-cycle or interest-rate influences. Adding several funds with the same dominant companies can preserve those exposures despite a larger number of fund names.
Look Through the Funds
A broad fund can provide extensive security diversification, but its role depends on the rest of the portfolio. A client holding a total-market fund, a large-cap growth fund and a technology fund may repeatedly own the same largest companies. Examine the combined weights rather than treating three fund labels as three independent sources of risk.
For each proposed allocation, identify major issuers, sectors, countries, currencies and common return drivers. Check whether a fund's mandate permits concentration, leverage, derivatives or substantial exposure to a single market. Holdings and index rules can change; a past count does not establish the future portfolio.
Asset classes add another layer. Equities, bonds, property securities and cash can have different drivers, but no fixed set of four category labels establishes a suitable plan. Weights, credit quality, duration, currency, valuation and liquidity affect the result. Exchange-traded REIT shares can decline alongside other equities. Non-traded and private REITs have different liquidity and valuation risks; a REIT label does not establish exchange trading.
A comparison of unspecified precious-metal and bond returns across loosely defined periods does not establish an allocation rule. Bullion, mining stocks and commodity futures are different exposures. Use specified total-return series, matching dates, currency and reinvestment conventions before comparing returns or volatility. Bonds can also lose value; diversification is not a promise that another asset will rise whenever stocks fall.
Fund the Transition and the Spending Need
A concentrated holding with a low tax basis presents an implementation decision, not merely an allocation defect. Selling can realize gain and require cash for tax. A proposed diversified portfolio must account for that cost and for any restrictions on selling, transferring or using particular funds.
Start feasible alternatives with the same investments and outside cash. Include taxes, transaction and ongoing fees, the use of sale proceeds and any assets retained to pay tax. Compare after-tax exposures and available resources at a common date. Do not count a reserve as extra money while also assuming the entire starting portfolio remains invested.
A phased transition, new contributions or different account placement may deserve analysis alongside an immediate sale. None is automatically preferable. Test execution rules, liquidity, risk retained while waiting and the client's objectives. A conditional comparison method does not establish a recommendation for a client whose tax and account facts are unknown.
A separate, simplified transition example makes the funding visible. Start with a concentrated holding worth $100,000, tax basis of $50,000 and no outside cash. Assume no distributions or other costs, a hypothetical 20% tax on every realized gain and complete liquidation at a common date one year later. Selling now creates a $50,000 gain and $10,000 tax, paid from the proceeds, leaving $90,000 for the diversified alternative. If that alternative gains an assumed 5%, its $94,500 value leaves $93,600 after tax on the $4,500 new gain.
Retaining the original holding uses the same starting resources. If it falls an assumed 20%, its $80,000 value leaves $74,000 after tax on the remaining $30,000 gain. If it instead rises an assumed 50%, its $150,000 value leaves $130,000 after tax. These invented, different return paths illustrate the range of outcomes and the cost of funding a transition. They do not establish that diversification earns 5%, that the concentrated holding follows either path or that selling is the right choice. Actual distributions, tax lots, rates and costs must be modeled separately.
The spending plan matters too. An investor can own many stocks and still lack cash for a near-term bill. Separate resources available for essential spending from assets that may have to be sold during a decline. Stress-test the amount, timing and accessibility of the funds rather than assuming a diversified portfolio can always supply the required withdrawal.
The Client Conversation
When a client asks why the plan does not hold only the recent winners, explain the trade-off directly: "A concentrated portfolio can finish far ahead or far behind. We cannot promise that diversification produces the same result with a smoother path. We are examining how much the plan depends on a few companies and whether the possible losses fit your goals."
Then make the exposure visible: "These three funds overlap in several large holdings. Adding another fund may not change that exposure much. We will look at the combined weights, the costs of changing them and the cash you need before deciding."
For an existing concentrated position, add the implementation question: "A sale may create tax. We will compare feasible ways to reduce concentration using the same starting assets and cash, including that tax and the risk you retain under each alternative. A larger holding count alone is not the objective."
These statements preserve the uncertainty that the analysis actually contains. They also make clear what the advisor must calculate rather than promising that every loss will be offset elsewhere.
Key Takeaways
- Look through fund and stock labels to combined weights, overlap and shared risk drivers.
- Attach the sample, weighting method and adequacy criterion to any research-based holding-count claim.
- Distinguish company-specific variance, total volatility, realized loss and the ability to fund spending.
- Test unequal weights and correlated losses before relying on a simple diversification example.
- Include tax, fees, account restrictions and reserve funding when comparing an actual transition.
- Explain that diversification changes the distribution of outcomes; it does not guarantee the same ending wealth or a market return.
The Advisor's Edge
The research and fund holdings are available to readers. The advisor's work is connecting those facts to a particular portfolio: identifying redundant exposures, testing losses against funded cash needs and comparing feasible changes after tax and costs.
The Certified Fund Specialist (CFS) program develops that analysis across two modules and twenty-one chapters. A useful fund recommendation considers holdings, risk, cost, tax and the client's circumstances together. Counting positions is one input, not the conclusion.
For the related measurement questions, see Standard Deviation, Beta, Alpha, and Sharpe Ratio Explained.
Sources and Notes: Taxable mechanics: IRS mutual-fund basis and distribution guidance. REIT trading distinctions: SEC investor bulletin. SEC asset allocation and diversification guide; Benjelloun, Evans and Archer: Forty Years Later, 2010; Bessembinder, Do Stocks Outperform Treasury Bills?. Risk and return calculations are hypothetical under the assumptions stated. This November 13, 2026 correction is local and has not been published.