CompanyGraph
Why the right portfolio question is not “How many holdings?” but “Which uncertainties remain shared after I add another one?”
Two ways to be wrong
Suppose an investor owns one company and is wrong about its product, accounting, regulation, or financing. The portfolio absorbs that error almost in full. Add companies whose cash flows are genuinely different and the effect of one mistake falls. But if the new holdings depend on the same customer, credit market, energy price, or supplier, the ticker count rises without removing the central risk.
Concentration and diversification are therefore not moral positions. Concentration trades more outcome variance for more impact from a correct thesis. Diversification trades away some upside from a single exceptional holding in exchange for less damage from a single failure. The choice should be tied to the investor's ability to estimate the business and to survive being wrong.
What portfolio theory actually contributes
Harry Markowitz's 1952 portfolio theory formalized the importance of covariance: expected return and individual volatility are not enough; the way holdings move together changes the distribution of portfolio outcomes. The original paper provides the mathematical foundation for mean-variance analysis. It does not tell an investor which expected returns are correct, how stable correlations will be, or how much loss a person can tolerate.
| Question | Useful observation | Remaining uncertainty |
|---|---|---|
| How large is a position? | Its contribution to portfolio profit and loss | Whether the thesis is more accurate than the market's |
| How many positions? | How much idiosyncratic exposure can be diluted | Whether holdings share hidden drivers |
| How correlated are returns? | How holdings moved together in a sample | Whether stress changes the relationship |
| How much downside can be borne? | Financial and behavioral capacity for loss | Whether the investor will sell before the thesis is tested |
Concentration can reflect skill or an unexamined boundary
A manager who concentrates may have a narrow circle of competence, a documented process, and the ability to follow a company closely. But concentration can also arise from employer stock, tax constraints, inherited holdings, local familiarity, or overconfidence. The same portfolio shape can have different causes.
Research using household brokerage accounts found that more concentrated portfolios sometimes outperformed more diversified ones, particularly among investors with local or non-S&P 500 holdings. The result is evidence about one 1991-1996 sample, not proof that concentration creates skill. The authors discuss information advantages, but familiarity and selection effects remain alternative explanations.
Diversification can fail in a different way
Holding many names can reduce company-specific risk while leaving a portfolio exposed to a shared regime. A group of software companies may depend on the same cloud infrastructure and enterprise budgets. Several banks may depend on the same yield curve and commercial-property market. A global portfolio can still be concentrated in one currency or supply chain.
Correlation is also state-dependent. Assets that appear unrelated in calm periods can fall together when investors need cash, lenders tighten, or a physical disruption affects several operations. Historical correlation is an observation of a sample, not a permanent property of the assets.
The operational question beneath the portfolio
Every holding is a claim on an underlying business or asset. Before adding a second company, identify the real source of the first company's cash: customers, labor, machinery, software, licenses, inventory, and financing. Then ask whether the second company depends on the same source or can continue when the first is impaired.
This matters because a portfolio can be financially diversified but operationally coupled. Two businesses may have different products yet share a contract manufacturer, a critical input, or a regulated network. Conversely, two companies in the same industry may have different customers, balance sheets, and cost structures. Sector labels are a starting point, not a covariance estimate.
How to choose without a slogan
- Define the loss you are trying to limit. Is it permanent capital loss, temporary volatility, income interruption, or a forced sale?
- Document the information edge. A large position needs more than conviction; it needs evidence that the thesis is better tested than its alternatives.
- Map shared drivers. Group holdings by customer, funding source, geography, infrastructure, regulation, and technology as well as by industry.
- Test a stress case. Ask what happens when the common driver fails and whether cash, time, and authority remain to respond.
- Match size to uncertainty. A thesis with wide outcome ranges should not be made safe by a confident sentence about diversification.
The most defensible conclusion is conditional. Concentration is rational when the investor can judge a limited number of opportunities and survive the downside; diversification is rational when estimation error is high and genuinely different exposures can be combined. Neither label substitutes for understanding what the holdings actually depend on.