Five Stocks Are Not Safer Than an Index Fund, Even If You Know Them

We recently posted a deliberately wrong hot take from our feed to see who would bite:

“Diversification is mostly an excuse for not doing enough research. If you genuinely understand a company, owning 30 stocks only dilutes your best ideas. Five high-conviction positions should be safer than an index fund because you actually know what you own.”

via @egr_investor (satire, though versions of it are posted in earnest every day)

The take works as bait because every clause contains a grain of something real. Skilled investors do sometimes concentrate. Thirty mediocre ideas can dilute three good ones. Knowing what you own is genuinely valuable. And yet the conclusion, that five researched stocks are safer than an index fund, is wrong in a way that a century of market data measures precisely. This guide takes the argument seriously enough to dismantle it piece by piece, steelman included, because the mistake it makes is one of the most expensive mistakes a DIY investor can make: treating conviction as if it were a risk reduction.

What research can remove, and what it cannot

Start with the tweet’s framing of diversification as a confession of laziness. Careful research can sharpen your estimates of a company’s competitive position, finances, management quality, and valuation. What it cannot do is remove the uncertainty that no amount of study reaches:

• competitors and technologies that do not exist yet
• fraud and accounting failures designed to survive scrutiny
• regulatory, legal, and political shocks
• capital-allocation mistakes by future management
• macroeconomic regimes nobody forecast
• the possibility that your valuation assumptions are simply wrong

Markowitz’s founding insight was aimed at exactly this residue. Portfolio risk is not the average of how risky each holding looks on its own; it depends on the covariances among holdings, which is why combining imperfectly correlated risks lowers portfolio variance without a proportional cut in expected return.¹ Diversification and research answer different questions. Research decides which risks you take on purpose. Diversification limits what any single mistake can do to the plan. A portfolio needs both, and neither substitutes for the other.

Knowing the company is different from knowing the price is wrong

The tweet’s second move is subtler: “if you genuinely understand a company.” Understanding a company is the entry ticket, and it is the wrong test. A wonderful business can be a poor investment when its future is already priced in; a mediocre business can be a great investment when the market assumes something even worse. Excess return requires being right about the future relative to what the current price already implies, which is why a defensible stock thesis has to answer five questions, and “is this a good business?” is none of them:

1. What does the current price appear to assume?
2. Which of those assumptions is materially wrong?
3. Why do you understand this better than the marginal buyer or seller, who may be a professional with better data?
4. How large is the mispricing after accounting for uncertainty?
5. What evidence would prove your thesis wrong?

Economics gives active research its due here. Grossman and Stiglitz showed that perfectly efficient prices are impossible: if prices already reflected all information, nobody would be paid to gather any, so nobody would, and prices would stop reflecting it.² Some research must earn a return. The catch is that this makes price discovery a competitive, costly contest whose rewards accrue at the margin, to whoever is fastest and best resourced. The existence of paid-for information is a fact about markets in aggregate. It is nothing like a promise that your research, on a stock followed by thousands of professionals, has produced an edge.

The arithmetic of five stocks

Now the central claim: five well-understood positions “should be safer than an index fund.” This is checkable, and it fails. Decompose a stock’s return into a market component and a company-specific residual:

For an equally weighted portfolio of $n$ stocks with independent residuals and average beta of one, the variance is approximately

The market component never diversifies away; the company-specific component shrinks with the number of holdings. Individual stocks carry far more volatility than the market they belong to; research on the decomposition of US stock risk finds firm-specific variance is the dominant share of a typical stock’s total variance.³ Using representative magnitudes, a typical stock at 45% volatility against a market at 18%, the arithmetic looks like this:

Five stocks carry roughly 43% more volatility than the market, and this is the generous version: it assumes equal weights, beta of one, and zero correlation among the residuals. Five technology companies, five banks, or five businesses exposed to the same economy violate that assumption and diversify one another less. Statman’s classic analysis put the threshold for a well-diversified stock portfolio at 30 to 40 randomly selected names, against the folk wisdom that ten did the job.⁴ Research into the companies changes none of this table. Volatility from fraud, disruption, and unforeseen events does not check whether the shareholder wrote a thesis memo first. Confidence changes how the risk feels; the arithmetic is indifferent to how it feels.

A few dozen companies carry the whole market

Volatility understates the case for breadth, because the deeper problem is which stocks you are likely to miss. Bessembinder’s landmark study found that 57.4% of US common stocks from 1926 to 2016 produced lifetime buy-and-hold returns below one-month Treasury bills, that the single most common lifetime outcome for an individual stock was a total loss, and that roughly 4% of firms, 1,092 companies, accounted for the entire $34.8 trillion of net stock market wealth creation above T-bills.⁵ His full-century update sharpens it: across 29,754 US stocks from 1926 through 2025, the average lifetime buy-and-hold return exceeded 30,000% while the median was negative 6.9%; nearly 60% of stocks reduced shareholder wealth over their listed lives; and just 46 firms account for half of the roughly $91 trillion of net wealth the market created.⁶

An average that enormous coexisting with a negative median is the signature of extreme positive skew. A stock can lose at most 100% and can gain without limit, so a handful of thousand-fold compounders pull the market’s average far above what the typical stock delivers. The index holds the winners by construction: broad ownership lets the rare extreme compounders grow into ever-larger weights while failures shrink toward irrelevance. A five-stock portfolio must instead pick from a universe where the typical member loses to T-bills, and the consequence shows up exactly where the tweet claims safety. Bessembinder’s 2026 study of “do-nothing” portfolios drawn from S&P 500 constituents found that narrow, randomly selected portfolios earn average returns similar to the index yet underperform it the majority of the time, and the disadvantage grows as portfolios get narrower and horizons get longer.⁷ Same average, worse typical outcome: the extra return of the lucky few concentrated portfolios is paid for by the majority that miss the winners. We walk through this skewness evidence, with a simulator calibrated to Bessembinder’s data, in Most Stocks Lose to T-Bills.

What concentrated households actually earn

If concentration reflected superior research, concentrated household portfolios should cluster among sophisticated investors and outperform. Mostly, the opposite. Goetzmann and Kumar, examining tens of thousands of brokerage accounts, found underdiversification most prevalent among younger, lower-income, less-educated, and less sophisticated investors, and associated with overconfidence, trend chasing, and local bias; it was costly for most, though a small subset did appear to hold concentrated positions on genuine information.⁸ Dimmock and coauthors tied underdiversification to probability weighting: investors who overweight small chances of huge payoffs hold fewer stocks and more lottery-like ones, at a measurable cost to risk-adjusted returns.⁹ That finding should sting anyone drawn to the tweet: a five-stock high-conviction portfolio is attractive partly because it offers a small chance of an enormous win, which is a preference, and preferences are not edges. Barber and Odean’s classic brokerage study rounds out the picture: the most active individual traders earned 11.4% annually while the market returned 17.9%, and the average household also trailed.¹⁰ Effort, activity, and confidence did not convert into results.

The steelman: when concentration is rational

An honest version of this argument has to concede what the research concedes. Anton, Cohen, and Polk’s long-running working paper found that fund managers’ single highest-conviction positions outperformed their other holdings by roughly 2.8 to 4.5 percentage points a year, suggesting managers hold diluted portfolios for commercial reasons while their best ideas carry real information.¹¹ Kacperczyk, Sialm, and Zheng found industry-concentrated mutual funds outperformed on average in their 1984–1999 sample.¹² Ivković, Sialm, and Weisbenner found concentrated households, particularly larger accounts holding local and less-followed stocks, earned excess returns consistent with an information advantage.¹³

Read carefully, all three results run in one direction: investors who already possess an edge sometimes express it through concentration. None of them shows the reverse, that concentrating creates an edge, and none shows that feeling conviction predicts having one. That ordering carries the entire argument. Skill can justify concentration. Concentration is no evidence of skill, and neither is conviction, which every overconfident investor also has in abundance.

Beyond demonstrated edge, two other cases are legitimate. Founders and owner-operators hold concentrated stakes because ownership buys control, aligned incentives, and the ability to affect the outcome, which is categorically different from holding five public stocks you cannot influence. And investors constrained by taxes or employment, low-basis legacy shares, lockups, vesting schedules, face a real question of whether the incremental risk justifies the cost of unwinding, which our concentration risk guide works through in detail. What none of these cases resembles is a retail investor choosing concentration voluntarily on the theory that familiarity is safety.

The zero-sum arithmetic of beating the market

Could you be in the skilled subset? The honest prior comes from arithmetic before it comes from evidence. Sharpe’s observation is that active and passive investors together hold the market, so before costs the average actively managed dollar earns exactly the market return, and after costs it earns less; every dollar of outperformance is another active investor’s underperformance.¹⁴ Fama and French found that few active mutual funds generate enough return to cover their costs, with evidence of genuine skill confined to the extreme tails.¹⁵ Berk and Green supplied the economic reason skill fails to reach investors even when it exists: money flows to successful managers until scale and fees absorb the advantage.¹⁶

The current scorecards show how hard the contest is for trained, resourced professionals: 79% of active US large-cap funds trailed the S&P 500 in 2025 alone.¹⁷ Persistence is the more damning statistic. Of the 164 large-cap funds in the top quartile in 2021, none remained continuously in the top quartile through 2025, and only 4.5% of above-median funds stayed above median for all five years, less than the 6.25% pure chance would produce.¹⁸ Mutual funds are not DIY investors, and these scorecards do not prove an individual cannot win; funds carry fees and scale problems you don’t. The narrower, correct lesson: identifying durable skill in advance is extremely difficult even for professionals evaluating professionals. A DIY investor evaluating themselves, with a small sample and no benchmark discipline, faces a harder version of the same problem, with overconfidence working the wrong way.

Who should not concentrate

Concentration is hardest to justify when the money matters and the edge is asserted rather than demonstrated. Treat five-stock investing as disqualified when:

• the money funds retirement, college, or a near-term purchase, and one failure would materially delay the goal
• the thesis is mostly “great company,” “huge market,” or “I use the product”
• you cannot state what the market is mispricing and why you know better than the marginal price-setter
• performance has never been measured against an investable benchmark, after taxes and trading costs, including the dead positions
• recent success is the main evidence of skill
• you would not calmly hold through the 50% to 80% drawdown that individual stocks routinely deliver
• the five positions share an industry or macro exposure despite having different tickers, a failure mode we dissect in The Tech Bro Portfolio

Employer stock deserves its own warning: your salary, unvested equity, and career prospects already concentrate you in one company, so adding shares stacks financial capital on top of human capital that fails in the same scenario. Concentration you did not choose is a reason to diversify elsewhere, never a foundation to build on.

Price the conviction

Concentration carries a hidden claim: that your picks will beat the market by enough to pay for the wider range of outcomes you accept by holding them. The calculator below makes the claim explicit. It computes the effective number of holdings from your position sizes, splits your volatility into market risk and diversifiable company-specific risk, prices the alpha a mean-variance investor would demand for that extra risk, and estimates how often the portfolio trails the index over your horizon if the claimed edge isn’t real. It ships with the tweet’s portfolio preloaded.

Two default results deserve emphasis. With everything in five typical stocks and no edge, about two-thirds of 20-year paths finish behind the index, echoing Bessembinder’s do-nothing result from actual data rather than a model. And at moderate risk aversion, the five-stock structure needs roughly five percentage points of annual alpha just to break even on a risk-adjusted basis, before taxes. Five points of durable annual alpha is a figure the best investors in history would respect. That is the bar the tweet clears with the phrase “should be safer.”

What we recommend

Index the money your plan depends on. A broad, cheap US and international core captures the equity premium and holds the rare extreme winners by default, without requiring you to identify them in advance. This is an acknowledgment that nobody, including the professionals losing to the index above, reliably knows which 46 companies will carry the next century.

If you want to pick stocks, and there are honest reasons to, from a real information advantage to the fact that it keeps you engaged enough to stay invested, run a core-and-satellite structure. Keep the active sleeve small enough that a total loss would not move any goal that matters; 5 to 10% of investable assets is a sensible guardrail for most households, a risk budget rather than an optimum. Five equal positions inside a 10% sleeve put 2% of your wealth in each idea, which is a bet. Five positions at 20% each is a career. Inside the sleeve, hold yourself to the standard the tweet skips: write down the variant view, the price-implied assumption you think is wrong, what would falsify it, and the exit rule, before buying. Then measure the sleeve after costs and taxes against an investable benchmark, factor-adjusted, dead positions included. Conviction that refuses measurement is a feeling wearing a process’s clothes.

And skip the 30-stock compromise. The tweet is right that thirty hand-picked names dilute your best ideas; the mistake is the conclusion drawn from it. Thirty positions recreate index-like exposure with more cost, more tax friction, more work, and less diversification than the real index. If you want the market, buy the market for three basis points. Concentrate only the part of the portfolio built to express an edge, and size it like the edge might not exist.

How Summitward helps

Summitward’s portfolio dashboard measures the exact quantities this guide argues about. The concentration panel computes your portfolio’s HHI and effective number of positions, the same statistic as the calculator above but on your real holdings, and its hold-versus-diversify projection runs percentile fans on concentrated positions. The factor analysis regresses your returns against market, size, value, profitability, and momentum factors and reports the alpha with significance, which answers the question self-assessment cannot: whether your stock picks beat the market after accounting for the risks they loaded on. A sleeve that outperformed because it was leveraged to momentum and small caps has demonstrated factor exposure, and factor exposure is available for basis points. The household view then puts employer equity and unvested compensation in the same picture, because a household’s true concentration includes the paycheck and the grants, whatever the brokerage account looks like on its own.

Frequently asked questions

Didn’t Buffett say diversification is for people who don’t know what they’re doing?

Buffett has described diversification as protection against ignorance, unnecessary for those who know what they are doing, and he is the strongest possible version of the steelman: a demonstrable, decades-long, benchmark-crushing edge, expressed through concentration, often with board seats and control. The quote describes his situation accurately. The trouble is that every overconfident investor also believes they know what they are doing, and the brokerage-account studies cited above find most concentrated investors resemble the overconfident group rather than the skilled one. Note also what Buffett tells investors without his advantages: his standing advice, repeated in his shareholder letters, is a low-cost S&P 500 index fund.

How many stocks do I actually need to be diversified?

Statman’s answer was 30 to 40 randomly selected stocks to capture most diversifiable-risk reduction, and later work argues even that understates it once you account for skewness, since a random 30-stock portfolio still probably misses the extreme winners that drive the market’s return. In practice the question is obsolete: total-market index funds deliver thousands of holdings at near-zero cost, so the minimum-stock threshold is a puzzle no DIY investor needs to solve by hand.

Is a small stock-picking account fine?

Yes, and we would rather you have one than pretend you never will. A 5 to 10% sleeve funds the learning, the engagement, and the occasional genuine insight, while capping the damage of the far more common outcome. The discipline is what converts it from entertainment to education: written theses, an investable benchmark, after-tax accounting, and dead positions included in the record.

My five stocks beat the index for three years. Do I have an edge?

Three years of outperformance is consistent with skill and heavily consistent with luck, factor exposure, or a sector cycle. Five concentrated positions produce enormous tracking error, so multi-year winning streaks arise by chance constantly; among professional funds, top-quartile performance shows essentially no persistence, with fewer funds staying above median than coin flipping predicts. Before crediting skill, regress the returns against factors, check whether one position or one theme explains everything, and ask whether the result survives after tax. Then keep the sleeve sized as if the answer might still be luck, because for a while, it might be.

Key takeaways

• Concentration amplifies whatever is present, skill or error; it cannot create an edge, and conviction is no evidence of one.
• Five typical stocks run roughly 26% volatility against the market’s 18%, and that assumes your picks are uncorrelated, which five same-theme stocks are not.
• Across a century, the median US stock lost 6.9% over its lifetime while 46 firms produced half of the market’s $91 trillion in net wealth; narrow portfolios trail the index most of the time because they usually miss the rare winners.
• Concentrated households are disproportionately overconfident rather than informed; the skilled subset exists and is small, and membership cannot be established by feeling knowledgeable.
• A five-stock all-in portfolio needs roughly five percentage points of annual alpha to justify its extra risk at moderate risk aversion; 79% of professional large-cap funds could not beat the index at all in 2025.
• Index the core, cap the active sleeve at a size you can afford to lose, document every thesis, and measure results after costs against a benchmark, factor-adjusted.

Related guides

• Most Stocks Lose to T-Bills. The Market Still Wins. the skewness evidence in depth, with a simulator calibrated to Bessembinder’s data.
• Compensated vs. Uncompensated Risk the theory behind the market/idiosyncratic split this guide’s calculator prices.
• Concentration Risk when to sell a concentrated position, HHI mechanics, and the tax math of unwinding.
• The Tech Bro Portfolio why VOO + QQQ + NVDA is five tickers and one bet.
• Can You Really Earn 20-30% a Year? the return expectations behind most concentration decisions, reality-checked.
• If Ben Felix Uses One Fund, Do You Need Ten? the opposite failure mode: diversification as clutter.

Sources

1. Markowitz, H. (1952). Portfolio Selection. Journal of Finance 7(1), 77–91.
2. Grossman, S. J., & Stiglitz, J. E. (1980). On the Impossibility of Informationally Efficient Markets. American Economic Review 70(3), 393–408.
3. Campbell, J. Y., Lettau, M., Malkiel, B. G., & Xu, Y. (2001). Have Individual Stocks Become More Volatile? Journal of Finance 56(1), 1–43. Cited for the decomposition of stock risk; the 45%/18% volatilities in the text are representative magnitudes.
4. Statman, M. (1987). How Many Stocks Make a Diversified Portfolio? Journal of Financial and Quantitative Analysis 22(3), 353–363.
5. Bessembinder, H. (2018). Do Stocks Outperform Treasury Bills? Journal of Financial Economics 129(3), 440–457.
6. Bessembinder, H. (2026). One Hundred Years in the U.S. Stock Markets. SSRN Working Paper 6438198.
7. Bessembinder, H. (2026). Returns to “Do-Nothing” Portfolios. SSRN Working Paper 7023959. S&P 500 constituents, 1971–2025.
8. Goetzmann, W. N., & Kumar, A. (2008). Equity Portfolio Diversification. Review of Finance 12(3), 433–463.
9. Dimmock, S. G., Kouwenberg, R., Mitchell, O. S., & Peijnenburg, K. (2021). Household Portfolio Underdiversification and Probability Weighting: Evidence from the Field. Review of Financial Studies 34(9), 4524–4563.
10. Barber, B. M., & Odean, T. (2000). Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors. Journal of Finance 55(2), 773–806.
11. Antón, M., Cohen, R. B., & Polk, C. (2021). Best Ideas. Harvard Business School Working Paper 21-004 (SSRN 1364827). Working paper, not a journal publication.
12. Kacperczyk, M., Sialm, C., & Zheng, L. (2005). On the Industry Concentration of Actively Managed Equity Mutual Funds. Journal of Finance 60(4), 1983–2011.
13. Ivković, Z., Sialm, C., & Weisbenner, S. (2008). Portfolio Concentration and the Performance of Individual Investors. Journal of Financial and Quantitative Analysis 43(3), 613–655.
14. Sharpe, W. F. (1991). The Arithmetic of Active Management. Financial Analysts Journal 47(1), 7–9.
15. Fama, E. F., & French, K. R. (2010). Luck versus Skill in the Cross-Section of Mutual Fund Returns. Journal of Finance 65(5), 1915–1947.
16. Berk, J. B., & Green, R. C. (2004). Mutual Fund Flows and Performance in Rational Markets. Journal of Political Economy 112(6), 1269–1295.
17. S&P Dow Jones Indices (2026, March). SPIVA U.S. Scorecard, Year-End 2025.
18. S&P Dow Jones Indices (2026). U.S. Persistence Scorecard, Year-End 2025.

Editor’s note

Educational content, not investment advice. The volatility table and calculator use stylized assumptions (representative volatilities, beta of one, lognormal returns, mean-variance utility) chosen to be generous to concentrated portfolios; real portfolios with correlated positions carry more risk than these figures show. The satirical tweet quoted at the top is our own. Facts verified against the cited journals, SSRN working papers, and S&P Dow Jones Indices scorecards as of July 2026; the two 2026 Bessembinder papers are working papers and figures may be revised.