Stock picking: why individual stocks are so hard to get right

In depth9 min read · 15 sources

Stock picking means deliberately selecting individual stocks in order to do better than the market. It sounds like a matter of diligence, but it often fails because of statistics: most individual stocks return less than the money market over their lifetime, and a few exceptional winners carry the entire market. Here you will learn why this is so, how many stocks a portfolio needs, what private investors and fund managers actually achieve and how to measure stock-picking success honestly.

What stock picking means

A stock picker does not hold the whole index, but a selection ranging from a handful to a few dozen stocks. How far a portfolio deviates from the index is measured by active share. As a matter of principle, the arithmetic of Sharpe (1991) applies: before costs, all active investors together earn exactly the market return; after costs, less. Deviating from the market is therefore a zero-sum game: whoever wins with their selection wins against other active investors.

The skewness of individual stocks

The main reason why stock picking is hard lies in the distribution of returns. Bessembinder (2018) analyzed US stocks since 1926: the majority returned less than one-month Treasury bills over their entire time on the stock market. Measured by the wealth created, the best 4% of companies account for the entire net gain of the US market; all the others together roughly matched Treasury bills. The cause is positive skewness: monthly returns are right-skewed, and compounding amplifies this, because losses stop at −100% while gains do not.

The picture is similar worldwide. According to Bessembinder et al. (2023), from 1990 to 2020, 55.2% of US stocks and 57.4% of stocks outside the US returned less than one-month US Treasury bills. The best 2.4% of companies created the entire net wealth gain of 75.7 trillion US dollars; outside the US, 1.41% of firms accounted for 30.7 trillion. A stock picker has to catch these few big winners; an index fund holds them automatically.

Worked example: how often does a portfolio lag the index?

A simple model shows what skewness means for a portfolio. Suppose every stock performs over ten years like the market multiplied by its own factor. This factor is lognormally distributed, with a mean of 1 and a standard deviation of log returns of 1.0. That corresponds roughly to 32% company-specific volatility a year. The index then corresponds exactly to the mean. A portfolio of n randomly drawn, equally weighted stocks lags the index if the average of its factors is below 1.

The model is an illustration with round assumptions, not an estimate. But it shows the pattern found in the studies: the mean is high because a few stocks gain a great deal; the median is low.

How many stocks does a portfolio need?

Besides skewness, volatility matters. For n equally weighted stocks with the same volatility σ̄ and average correlation ρ:

Evans and Archer (1968) examined randomly assembled, equally weighted portfolios; this gave rise to the widespread rule of thumb that about ten stocks are enough. Statman (1987) disagreed: a well-diversified portfolio of randomly chosen stocks needs at least 30 stocks for investors who invest on credit, and 40 for investors who do not borrow. Skewness adds to this: volatility falls quickly, but the risk of missing the few big winners only falls slowly. More on this in the article correlation and diversification.

Concentration and conviction

One argument speaks for concentration: if you really know something, you should not dilute it. There is evidence for this. According to Kacperczyk et al. (2005), active US equity funds that concentrated on a few industries performed better on average from 1984 to 1999 after adjusting for risk and style; skill showed up more among managers with concentrated portfolios.

But concentration demands a high degree of accuracy. The fundamental law of active management according to Grinold (1989) links the information ratio to skill and breadth: IR ≈ IC × √BR. If you only make 20 independent bets a year, you need well over three times the skill of someone making 200 bets to achieve the same information ratio.

What the research shows

• Private investors: Barber and Odean (2000) analyzed 66,465 households at a US discount broker from 1991 to 1996. The households that traded most earned 11.4% a year, the market 17.9%; the average household turned over 75% of its portfolio a year. The authors name overconfidence as the cause.
• Disposition effect: using 10,000 accounts, Odean (1998) showed that investors are much keener to sell winners than losers (the disposition effect). Subsequent performance does not justify this, and in taxable accounts it costs after-tax return.
• Lottery stocks: according to Bali et al. (2011), stocks with the highest daily gain in the previous month subsequently deliver lower returns; the difference between the decile with the lowest values and the one with the highest was more than 1% a month. Investors apparently overpay for the small chance of a big hit (lottery stock).
• Professionals: according to Fama and French (2010), the aggregate portfolio of active US equity funds is close to the market portfolio, but the high costs pass through in full to investors as lower returns. Only a few funds earn enough excess return to cover their costs (luck versus skill).
• Europe: according to the SPIVA Europe (2026) report, over the ten years to the end of 2025, 97.02% of euro-denominated European equity funds lagged the S&P Europe 350; of the 1,040 funds at the start of the period, only 56.83% still existed at the end.

How fund managers select stocks

If you still want to pick individual stocks, you can take your cue from how professional managers work. The points describe common practice, not a proven formula for success:

• Bottom-up research: understand each company's business model, balance sheet and valuation. That is the craft of fundamental analysis. The key question is: what do I know, or weigh, differently from the market?
• Thesis and catalyst: an investment idea needs a testable thesis and an event that makes the market rethink, such as quarterly results, a new product or a change of management.
• Position size: weights follow conviction and risk, with caps per stock and per industry. No single total loss should be allowed to endanger the portfolio.
• Selling discipline: decide in advance to sell when the thesis is disproved or the price target is reached. This counteracts the disposition effect.
• Risk control: check which factors, industries and countries unintentionally shape the portfolio. Some stock pickers are in truth only betting on a style.

For advanced readers: measuring stock picking honestly

• Suitable benchmark: a portfolio of small US stocks should not be measured against the Dow. The benchmark should reflect the investable universe and include dividends.
• Adjust for risk: Jensen (1968) measured fund performance as the alpha of a regression against the market; Carhart (1997) added momentum to the Fama and French factors. A high raw result is often just a higher beta or a tilt toward small or cheap stocks.
• Statistical certainty: the t-statistic of an alpha is roughly information ratio × √years. A value of 2 is regarded as the minimum.

• Survivors: if you only analyze stocks that are still listed today or funds that still exist, you overestimate the results. Brown et al. (1992) describe this survivorship bias.
• Costs and taxes: trading costs, bid-ask spreads and taxes on realized gains are incurred more often with individual stocks than with an index fund.

Limits and common mistakes

• Hindsight bias: yesterday's winners look obvious in retrospect. A backtest with stocks that are known to have succeeded proves nothing.
• Too few stocks: five stocks are not diversification, neither for volatility nor for the chance of holding the rare big winners.
• Stories instead of prices: an exciting product is often already in the price; what matters is what the price already expects.
• Too much trading: in the studies, frequent reshuffling cost returns.
• Model limits: the worked example assumes independent, identically distributed stocks. In reality, industries, sizes and risks differ. The pattern of skewness remains, the exact percentages do not.

This article is not investment advice and deliberately names no individual stocks. Whether active or passive investing suits you better is explained in active or passive. How fraudsters exploit the appetite for a big hit is shown in pump and dump.

How Portfolio Lab calculates it

• Backtest with individual stocks: the securities database also contains stocks. In the Backtest, you can test a portfolio of individual stocks with reinvested dividends against a benchmark, such as an index ETF. The app only knows the stocks you select today; if you put well-known winners into the past, you are subject to survivorship bias (methodology).
• Factor analysis: the Factors page regresses your portfolio's monthly returns on market, size and value, in the five-factor model also on profitability and investment, and optionally on momentum, and shows the alpha with its t-statistic, with Newey-West standard errors if you wish (methodology).
• Style analysis: Style analysis explains returns with style indices following Sharpe and measures the selection return out of sample (methodology).
• Concentration: Look-through combines individual stocks and the positions of your funds and shows the share of the ten largest holdings and the effective number of positions, the reciprocal of the Herfindahl index (methodology).

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.