Most Stocks Lose to T-Bills. The Market Still Wins.

Two facts that look like contradictions: the U.S. stock market created roughly $35 trillion of net wealth above Treasury bills between 1926 and 2016, and the median individual U.S. stock during that same period had a lifetime buy-and-hold return that lost to T-bills. Both are true. Bessembinder’s 2026 update through a full century (1926-2025) widens the gap: $91 trillion of net wealth created, and the median stock still finished its life slightly below where it started. The reconciliation explains why broad, low-cost, market-cap-weighted index investing isn’t a settling-for-mediocrity choice. It’s the only way to mechanically guarantee you own the small handful of stocks that drive the entire asset class’s wealth creation.

The puzzle

If most individual stocks underperform the safest possible benchmark (one-month Treasury bills), and the asset class as a whole still creates trillions in wealth, what’s actually happening?

The answer is positive skewness. A small number of extreme winners (Apple, Microsoft, ExxonMobil, Amazon, Alphabet) pull the market average dramatically up. The typical stock disappoints. The market index works because it owns enough stocks to capture the rare winners by construction, not because the typical stock is a good bet.

This research is from finance professor Hendrik Bessembinder at Arizona State’s W.P. Carey School. It’s now the most-cited modern academic argument for broad diversification, and it reframes what investors should optimize for. Stop trying to identify the next big winner. Make sure you own enough stocks that the next big winner is in your portfolio whether you saw it coming or not.

The U.S. evidence: 1926-2017

Bessembinder studied every U.S. common stock that traded on the NYSE, NYSE American, or Nasdaq from 1926 through 2016 (later updated through 2017) using the CRSP database, about 26,000 stocks in total. He computed the lifetime buy-and-hold return for each stock and compared it to one-month T-bills compounded over the same period. The results:^1,2

• 58% of stocks had lifetime buy-and-hold returns below one-month T-bills.
• 38% of stocks beat T-bills, but only modestly.
• Just over 4% (1,092 of ~26,000 stocks) accounted for all $35 trillion of net wealth created above T-bills.
• 86 stocks alone produced $16 trillion, half of the market’s entire excess wealth creation over 90 years.
• The median U.S. common stock’s entire lifespan in the CRSP database was just 7.5 years. Most stocks delist quickly via bankruptcy, acquisition, or going private.

The headline isn’t that stocks are bad. It’s that individual stock outcomes are wildly uneven, and the market’s wealth comes from a tiny right tail.

The 2026 update: one hundred years, 1926-2025

In 2026 Bessembinder extended the U.S. study to a full century, covering 29,754 common stocks from January 1926 through December 2025. The extra decade of data pushed every headline number further in the same direction:³

• The cross-stock mean lifetime buy-and-hold return was over 30,000%, while the median was −6.9%. The average stock looks spectacular; the typical stock lost money over its life.
• About 60% of firms had negative total returns relative to one-month T-bills.
• Shareholders’ net wealth rose by $91 trillion over the century, even though most individual firms left their long-term holders worse off than one-month T-bills would have.
• Just 46 firms account for half of that $91 trillion, and 1,082 firms (3.7%) account for all of it. The rest, in aggregate, were a wash against the risk-free rate.
• Apple and Nvidia alone produced more than 10% of all net wealth created in U.S. markets over the entire century. The top five (Apple, Nvidia, Microsoft, Alphabet, Amazon) account for about21%.

The concentration is tightening over time, not loosening. In the 1926-2016 window, half of the market’s $35 trillion in net wealth traced to 86 stocks. In the 1926-2025 window, half of a far larger $91 trillion traces to just 46 firms. Fewer names are responsible for a bigger share of the total, which is the signature of a distribution whose winners compound faster than its losers can drag it down.

The global evidence: 64,000 stocks, 1990-2020

The skeptical response to the U.S. study is “maybe that’s a quirk of the U.S. market or a long-ago era.” Bessembinder and coauthors answered this in a 2023 follow-up that looked at over 64,000 global common stocks across 42 countries from January 1990 through December 2020. The findings replicated and extended the U.S. result:⁴

• 55.2% of U.S. stocks and 57.4% of non-U.S. stocks underperformed one-month U.S. Treasury bills (USD-denominated) over matched holding horizons.
• Global equities still created $75.7 trillion of net wealth above T-bills.
• The best-performing 2.39% of firms accounted for all of that net global wealth creation.

A precision note: the global study uses one-month U.S. T-bills (in USD) as the common cross-country benchmark, not each country’s local cash rate. The lesson holds because the comparison is apples to apples across countries, but be careful when reading sloppy summaries that say “most stocks underperform cash” without specifying which cash.

Three claims people treat as one

Most of the confusion around this research comes from swapping three different statements that sound identical but measure different things. A stock can satisfy one without satisfying the others. Keeping them separate is what stops the finding from being overstated in either direction.

The three line up closely in the data, but they are not the same claim. “Most stocks have a negative return” is about absolute dollars. “Most stocks underperform T-bills” is about the risk-free hurdle. “Most firms reduce shareholder wealth” is about aggregate dollars created or destroyed. A careful summary names which one it means.

The math: why long-horizon stock returns are positively skewed

The skewness isn’t a behavioral or institutional accident. It’s structural, and it gets stronger the longer you hold. Three forces drive it:

1. Limited liability caps the downside at −100%. Equity holders can lose everything they invested but no more. Worst case: zero.

2. The upside is uncapped. Apple stock returned more than 250,000% from its 1980 IPO through 2020 (split-adjusted). There is no symmetric catastrophe equivalent on the loser side because of rule #1.

3. Multiplicative compounding amplifies the asymmetry. Returns multiply, not add. Doubling and halving aren’t mirror opposites: a stock that goes up 100% and then down 50% is back where it started, but a stock that goes down 50% has to go up 100% to recover. Over decades of compounding, this asymmetric arithmetic creates a long, thin right tail and a fat left mass concentrated near zero. Farago and Hjalmarsson showed formally that for holding periods beyond about five years, multiplicative compounding alone tends to dominate the distribution’s shape, generating large positive skewness even when the underlying log returns are roughly symmetric.⁵

That’s why the mean stock return can be respectably positive while the median stock return is negative. The mean gets pulled up by a few enormous winners; the median sits in the crowded middle where most stocks live.

The stock picker’s problem

Reading these numbers, the natural question is: “OK, but I just need to pick the few winners.” That’s the whole game. To beat broad indexing through stock picking you need to:

1. Identify the next Apple before it’s obvious. Apple in 1980 was a niche personal-computer company. The eventual mega-winner often isn’t the consensus glamour name at the time of investment, and may not even exist as a publicly-traded firm when you’d need to buy in.
2. Buy at a price that leaves room for excess return. A great company with a 100x price-to-earnings multiple can be a terrible stock. Most of the wealth creation in winners came from buying before the market priced in the eventual outcome, not from buying after.
3. Avoid the losers in the same theme. The internet era produced Amazon and Google. It also produced Pets.com and Webvan and hundreds of other zeroes. “Bet on the theme” isn’t a strategy if you can’t pick the winning firm within the theme.
4. Hold through 50%+ drawdowns. Amazon dropped 95% from its 1999 peak before becoming Amazon. Apple dropped over 75% in the early 2000s. Even the eventual winners punish their early holders with brutal volatility. Most pickers sell at the bottom.

Each of these steps has a low base rate of success on its own. Their joint probability of getting all four right on the same stock is what makes the “just pick the next Apple” argument much harder than it sounds. The Bessembinder data shows that even with the benefit of hindsight, the winners were a tiny minority of the universe and unevenly distributed across decades.

See it for yourself: the simulator

Set the number of stocks you’d hold and the holding period. The simulator runs 5,000 random portfolios drawn from the Bessembinder-calibrated distribution and shows you the result. Notice what happens to your probability of beating T-bills as you go from 1 stock to 500. Notice how far the median portfolio sits below the market index baseline at low stock counts. Notice the long, thin right tail in the histogram.

The diversification lesson

Broad index ownership is not “settling for the average.” It’s the systematic way to guarantee you own the rare winners without having to identify them in advance. Three flavors:

• An S&P 500 index fund captured every U.S. mega-winner of the past 50 years by construction. You didn’t have to predict that Apple would beat Atari, Microsoft would beat Lotus, or Google would beat Yahoo. The index just held them mechanically as they grew.
• A total U.S. market fund (VTI, ITOT) extends that to 4,000+ U.S. stocks, covering small-cap winners that the S&P 500 missed early.
• A total world fund (VT) extends globally to ~9,000+ stocks across developed and emerging markets, capturing the fact that tomorrow’s extreme winners can emerge anywhere, not just in the country that won the last cycle.

For more on the global extension, see the Case for Global Equity Diversification guide. The Bessembinder result reinforces it: if the next generation of extreme winners can emerge in Korea, India, the Netherlands, or Brazil, a U.S.-only portfolio is a bet that the country with winners over the past 30 years will keep winning.

Three things this clarifies that often get conflated

Active mutual funds vs index funds

The Bessembinder result is the structural reason most active mutual funds underperform their benchmarks. Active managers run relatively concentrated portfolios, typically 30 to 100 names. With only a few percent of stocks producing the bulk of market wealth, a narrow portfolio is almost certain to miss most of them. SPIVA reports consistently show that 80%+ of active U.S. equity funds underperform their benchmark over 15-year horizons. Bessembinder explains the mechanism: the math is against narrow portfolios. See also the Passive Investing Is a Label guide for the related point that “passive” doesn’t mean “no decisions”; it means delegated decisions baked into the index rules.

Concentrated single-stock portfolios

If you hold three to five tickers, your retirement is now riding on identifying the rare winner. The simulator above quantifies this: at 5 stocks held for 30 years in the U.S. universe, your probability of beating the market is dramatically lower than your probability of underperforming T-bills. The Concentration Risk guide covers how to measure single-stock exposure with HHI and build a multi-year diversification plan when you’ve accumulated a concentrated position.

Employer stock and RSUs

Tech workers with significant RSU income face the Bessembinder problem in compounded form: not only is your portfolio concentrated in one stock, your salary, future grants, and career outcome are all correlated with the same company. See the Human Capital Risk for Tech Workers and RSU Tax Strategy guides for the diversification framework. The Bessembinder math adds an additional argument: even setting aside concentration risk, the base-rate odds that any single company will be one of the rare Bessembinder winners are tiny. Employer stock gets concentrated for practical reasons (vesting schedules, lockups), but the long-run case for systematically diversifying out is strong.

When concentrated bets are defensible

The Bessembinder result is not an argument that nobody should ever own individual stocks. It’s an argument that concentrated stock picking is a high-risk strategy with bad base rates, which means it deserves a small allocation, a clear thesis, and explicit kill criteria. Reasonable cases:

• Position sizes under 5% of net worth. If your single-stock portfolio is genuinely play-money (money you could lose entirely without changing your retirement plan), the Bessembinder math doesn’t apply because you’ve already accepted the lottery-ticket framing.
• High-conviction theses with kill criteria. If you have a specific, falsifiable thesis (“I’m long this name because of X; if Y happens I’m out”), and you enforce the exit, you’ve at least cut left-tail risk. Most retail picking lacks the discipline.
• Genuinely employer-restricted stock you can’t immediately sell. Lockups, blackouts, or trading windows impose concentration whether you want it or not. The right move is to diversify aggressively as soon as you can; see the Concentration Risk guide for the multi-year framework.

The case the Bessembinder data argues against is the more common one: a 30%, 50%, or 80% retirement allocation to three to ten individual stocks because they’re “obvious winners” or “the future.” That’s a bet that you’ll identify the next decade’s extreme winners with enough confidence to size them as if they were already certain. The base rates say no.

The Summitward punchline

The goal of equity investing is not to avoid the asset class because most stocks disappoint. The goal is to own equities broadly enough that your portfolio doesn’t require you to identify the tiny minority of stocks that make the whole asset class work.

Broad, low-cost, market-cap-weighted, globally-diversified index funds aren’t a compromise position for people who can’t pick stocks. They’re the structurally correct response to a return distribution that rewards owning everything and punishes narrow concentration.

Frequently asked questions

Doesn’t the 4% statistic mean 96% of stocks went down?

No. About 38% of U.S. stocks beat T-bills modestly; another 4% beat T-bills enormously. So 42% of stocks did outperform the risk-free benchmark in absolute terms. The 4% figure is specifically: “these stocks accounted for ALL the net wealth above T-bills after netting out the losses of the worst stocks.” Some of the modest winners gained value too, but the gains roughly canceled the losses of the underperformers, leaving the extreme top to drive net creation.

If most stocks lose to T-bills, why would anyone invest in stocks at all?

Because the cap-weighted market index isn’t the median stock; it’s a weighted average that captures the right tail. The S&P 500 returned roughly 10% annualized over the long run because Apple, Microsoft, Amazon, and the other mega-winners contributed disproportionately. Owning the index gives you the market’s return, not the median stock’s return. Buying random individual stocks gives you the median outcome (mostly).

How is this different from “active funds underperform”?

It’s the structural cause. Active funds underperform partly because of fees and partly because their concentrated portfolios are unlikely to capture the few extreme winners that drive the market. Bessembinder explains why the diversification math is so unforgiving to narrow strategies, beyond the usual fee argument.

Does owning all 500 stocks in the S&P 500 mean I’ll match the market?

Roughly yes, if you cap-weight them like the index does. If you equal-weight 500 random stocks, you’ll typically underperform the cap-weighted market because the market index gets dragged up by its mega-cap winners with disproportionate weight. The simulator above models random equal-weighting, which is why even 500-stock portfolios fall short of the market line.

Has the 2010s mega-cap concentration broken this?

The opposite. It has reinforced the result. The 2010s and early 2020s saw Apple, Microsoft, Amazon, Alphabet, NVIDIA, and Meta become increasingly large fractions of total U.S. market capitalization. That’s exactly what positive skewness produces over time: winners compound faster than losers, so the market gets more concentrated in the right tail. Bessembinder’s 2026 century study quantifies how far it has gone: Apple and Nvidia together produced more than 10% of all U.S. net wealth creation since 1926, and the top five firms account for about 21%. The lesson isn’t that the era is over; it’s that this is how equity markets always work.

What if I disagree with the historical evidence?

Even if you believe the future skew distribution will be different from the past, the burden of proof is on the picker. To rationally pick stocks instead of broadly indexing, you need to claim either (a) that you can predict which firms will be in the rare top 2-4%, or (b) that future skewness will be much lower than past skewness so that diversification matters less. Neither claim is well-supported by evidence at this point.

Related guides

This guide sits next to several other Summitward arguments for broad diversification:

• The Case for Global Equity Diversification extends the within-market diversification lesson across borders.
• Concentration Risk covers how to quantify single-stock exposure with HHI and build a multi-year diversification plan.
• Passive Investing Is a Label unpacks what “passive” really means and where the discretion in indexing actually lives.
• The Tech Bro Portfolio shows how three-ticker portfolios (VOO + QQQ + NVDA) are the inverse of the broad-ownership lesson.
• Stocks Are Always Risky covers Bernstein’s four deep risks and why long horizons don’t repeal equity risk.
• When to Sell a Winning Stock is the follow-up question: if you happen to own one of the 4% of stocks that drives all the market wealth, when should you sell? Tax-cost-of-selling calculator included.
• Should You Buy IPO Stocks? applies the same skewness logic to a specific decision: most newly public stocks underperform broad markets in the aftermarket; the structure rarely favors retail buyers.

Sources

1. Bessembinder, H. (2018). Do Stocks Outperform Treasury Bills? Journal of Financial Economics, 129(3), 440-457. SSRN: 2900447.
2. ASU W.P. Carey School of Business research hub. Do Stocks Outperform Treasury Bills. Includes the headline U.S. statistics through 2017 and links to the underlying research.
3. Bessembinder, H. (2026). One Hundred Years in the U.S. Stock Markets. Financial Review. SSRN: 6438198; doi:10.1111/fire.70045. The 1926-2025 century update: 29,754 stocks, median lifetime return −6.9% vs mean over 30,000%, $91 trillion in net wealth, 46 firms for half and 1,082 firms for all of it, and the Apple/Nvidia concentration figures.
4. Bessembinder, H., Chen, T., Choi, G., & Wei, K.C.J. (2023). Long-Term Shareholder Returns: Evidence from 64,000 Global Stocks. Financial Analysts Journal. Global 1990-2020 study; 55.2%/57.4% underperformance figures and the 2.39% extreme-winner stat come from this paper.
5. Farago, A., & Hjalmarsson, E. (2024). Long-Horizon Stock Returns Are Positively Skewed. Review of Finance, 28(5). Formal treatment of how multiplicative compounding generates positive skewness in long-horizon returns.
6. Rational Reminder Podcast Episode 346 with Hendrik Bessembinder (primary source on the stock-return research) and Episode 397 (sidebar on long-horizon investor outcomes and constant leverage).