Why a Mix of Assets Beats Betting on One Winner

A mix of assets beats a bet on one holding chosen in advance, and it wins mainly on the spread of outcomes and the size of losses, not on return in every case. In hindsight the best single asset always beats a mix. A mechanical rule that bought last year's best asset class also earned more than an equal-weight mix from 2000 to 2025, though its worst year was about twice as deep. The evidence is starkest for single stocks.

Key takeaways

In Bessembinder's study of US stocks, 4.3% of companies created all the wealth above Treasury bills from 1926 to 2016, and most stocks trailed bills over their lifetimes.
From 2000 to 2025, US stocks led five broad asset classes in 12 years and gold in 8; the previous year's leader repeated 9 times and finished last 5 times.
Buying last year's leader beat an equal-weight mix on return, 10.13% a year against 6.71%, but its worst year was −18.0% against −9.7%. Holding one asset throughout turned $10,000 into anywhere from $16,426 to $149,514, depending on which one.
The Nasdaq Composite fell 77.9% in 2000–2002 and bitcoin 76.7% in 2021–2022; a 77.9% loss needs a 353% gain to recover.
Plans holding publicly traded employer stock, apart from standalone ESOPs and one-participant plans, must let participants sell shares from their own contributions with no service requirement, at least quarterly, and employer-contributed shares after three years of service.

What the single-stock record shows

The strongest evidence against a single pick is how individual stock returns are spread. Hendrik Bessembinder's study "Do stocks outperform Treasury bills?" (Journal of Financial Economics, 2018) tracked every US common stock in the CRSP database since 1926. Its abstract reports that most had lifetime buy-and-hold returns below one-month Treasury bills, and that the best-performing 4% of companies explain the US market's entire net gain.

Arizona State University's W. P. Carey magazine summary (Autumn 2018) gives the counts: 1,092 companies, 4.3% of nearly 26,000 traded, account for all $35 trillion created above T-bills from 1926 to 2016, and 58% of stocks failed to beat T-bills over their lives. The cause is skewness: a few huge winners pull the average up, so the typical stock does worse than the average stock.

A broad index fund owns the few huge winners by construction. A single stock is most likely to be one of the many that did not deliver, which the abstract says helps explain why poorly diversified active strategies most often underperform.

Leaders rotate across asset classes

The same pattern holds one level up. The table ranks five asset classes by calendar-year return from 2000 to 2025, using Aswath Damodaran's historical returns data set (data through 2025): the S&P 500 with dividends, 3-month Treasury bills, 10-year Treasury bonds, Moody's Baa corporate bonds and gold (year-end LBMA prices, price change only). Bond returns are computed from yields. Real estate (home prices without rent) and international stocks are left out.

Asset, 2000–2025	Years first	Years last	Best year	Worst year	Losing years
US stocks (S&P 500)	12	6	+32.1% (2013)	−36.6% (2008)	6
Gold	8	2	+66.2% (2025)	−27.6% (2013)	5
10-year Treasury bonds	3	7	+20.1% (2008)	−17.8% (2022)	6
3-month T-bills	2	11	+6.0% (2000)	+0.0% (2014)	0
Baa corporate bonds	1	0	+20.0% (2009)	−15.2% (2022)	5

No asset led for long. Stocks had the longest run, first in every year from 2012 to 2017, and gold led from 2005 to 2007. Otherwise the top spot changed hands in most years. The previous year's leader came first again 9 times, more than the roughly 5 that chance alone would give with five assets, but it also came last 5 times: stocks in 2000, 2018 and 2022, Treasury bonds in 2009 and T-bills in 2019. The gap between the top and bottom asset averaged 26.6 percentage points a year, and a single holding is exposed to all of it.

Chasing last year's leader, tested

The test runs two mechanical rules over 2000–2025, with no fees, taxes or trading costs. "Chase" puts everything each January into the previous calendar year's top asset (stocks for 2000). "Mix" holds 20% in each asset, reset every January. Largest decline uses year-end values, so declines within a year were deeper.

Holding, 2000–2025	Annual return	$10,000 became	Worst year	Largest decline
US stocks (S&P 500)	7.99%	$73,856	−36.6% (2008)	−37.4% (2000–2002)
10-year Treasury bonds	3.82%	$26,495	−17.8% (2022)	−21.5% (2021–2022)
Baa corporate bonds	6.30%	$48,953	−15.2% (2022)	−15.2% (2022)
3-month T-bills	1.93%	$16,426	No losing year	None
Gold	10.96%	$149,514	−27.6% (2013)	−36.3% (2013–2015)
Equal-weight mix	6.71%	$54,158	−9.7% (2022)	−9.7% (2022)
Chase last year's leader	10.13%	$122,964	−18.0% (2022)	−21.1% (2021–2022)
Perfect hindsight (not possible)	20.77%	$1,352,422	+1.4% (2015)	None

Historical results, not a forecast. Computed from the Damodaran data set; annual return is the compound annual growth rate.

Where the chase rule won, and why

In this data the chase rule beat the mix on return: 10.13% against 6.71% a year, ahead in 17 of 26 years. Related research points the same way, though it tests a different rule. Moskowitz, Ooi and Pedersen's "Time series momentum" (Journal of Financial Economics, 2012) found that each of 58 futures markets in stock indexes, bonds, currencies and commodities tended to keep moving in the direction of its own past returns for 1 to 12 months, then partly reverse. Their signal compares a market with its own past, not with other markets. The edge here was uneven. Ending in 2024, before gold's 66.2% year, cuts the gap to 8.3% against 6.2%, and the rule trailed the mix in 10 of the 44 ten-year windows starting each year from 1973 to 2016.

The rule also carried about twice the mix's worst-year loss and largest decline, and switched holdings 17 times in 25 years. In a taxable account, each of those sales can trigger capital gains tax. The rule also rotated only among five broad asset classes, not single stocks. The mix also held 20% in T-bills; without them it returned 7.85% a year, with a −12.6% worst year (2022).

What the mix does win

The mix's case is the range of outcomes. Committing to one of the five assets in January 2000 turned $10,000 into anywhere from $16,426 to $149,514. The mix ended at $54,158, above three of the five, and its worst year, −9.7% in 2022, was the smallest of any holding that could lose money. In 2008, when stocks fell 36.6%, the mix lost 2.8%.

Real chasing rarely looks like a disciplined annual rule. Sirri and Tufano's study of equity mutual fund flows, "Costly Search and Mutual Fund Flows" (Journal of Finance, 1998), found that buyers put disproportionately more money into funds that did very well in the prior period, so new money tends to arrive after the strong results have already happened.

How deep concentrated losses go

A narrow holding can fall much further than a broad asset class, as two series from Federal Reserve Bank of St. Louis (FRED) daily data show:

Nasdaq Composite. The index closed at 5,048.62 on March 10, 2000 and at 1,114.11 on October 9, 2002, a 77.9% fall. It did not close above its 2000 peak again until April 23, 2015.
Bitcoin. The Coinbase bitcoin price went from $67,510.06 on November 8, 2021 to $15,755.75 on November 21, 2022, a 76.7% fall. It first passed the 2021 peak on March 4, 2024.

Recovery starts from the smaller base, so losses need larger gains to undo:

Gain needed to recover = 1 ÷ (1 − loss) − 1

loss = peak-to-trough decline as a decimal (0.50 for a 50% fall)

50% loss   → 1 ÷ 0.50 − 1 = 100% gain needed
77.93% loss → 1 ÷ 0.2207 − 1 = about 353% gain needed

A holding that is 10% of a portfolio and falls 77.9% costs the portfolio 7.8%. The same fall in a holding that is 60% of the portfolio costs 46.7%.

Employer stock, the closest bet

For many workers the largest single-stock position is their employer's, and Enron is the documented case. In February 2002 testimony, GAO (GAO-02-480T) cited Labor Department figures showing that 63% of Enron's 401(k) assets were in company stock at the end of 2000. The company match went only into its stock fund and could be moved only from age 50, and GAO said the concentration came from both employee choices and that match. Enron filed for Chapter 11 on December 2, 2001, as reported in a Dynegy Form 8-K filed with the SEC, so many employees lost savings and paychecks to the same company.

Why employees load up after gains

Shlomo Benartzi's "Excessive Extrapolation and the Allocation of 401(k) Accounts to Company Stock" (Journal of Finance, 2001) ties employer-stock concentration directly to chasing. Employees of firms with the worst 10-year stock performance put 10.37% of their discretionary contributions into company stock, those with the best put in 39.70%, and those allocations did not predict future performance.

The right to sell employer stock

The Pension Protection Act of 2006 added a diversification right, 29 U.S.C. § 1054(j), for plan years beginning after December 31, 2006. Its conditions:

Who it covers: defined contribution (individual account) plans that hold publicly traded employer securities. A plan holding private shares is generally covered if a company in the employer's controlled group has publicly traded stock.
Employee money: shares bought with employee contributions and elective deferrals can be sold and reinvested with no service requirement.
Employer money: shares from employer contributions can be diversified once the participant has at least 3 years of service (beneficiaries of such participants, or of deceased participants, have the same right).
Alternatives: the plan must offer at least 3 other investment options, each diversified with materially different risk and return. It may limit switches to periodic, reasonable opportunities at least quarterly, and may not impose restrictions on employer stock that it does not impose on other plan assets, except those required by securities laws.

The right lets a participant sell; nothing in it makes the plan sell for them.

Frequently asked questions

How much of a portfolio in one stock is too much?

There is no official limit. FINRA's guide to owning company stock (August 2023) says no single formula suits all investors. It notes that some experts suggest no more than 10% of total investment assets in one stock, including the employer's, and that even 10% could be too high depending on goals and circumstances.

Does the diversification right cover employee stock ownership plans?

Not always. Under 29 U.S.C. § 1054(j), a standalone ESOP is excluded if it holds no contributions, or earnings on them, subject to sections 401(k) or 401(m) of the tax code and is a separate plan from the employer's other retirement plans. One-participant retirement plans are also excluded.

Can owning only funds still leave a portfolio concentrated?

Yes. FINRA's concentration risk guide (June 2022) gives the example of owning individual technology stocks, a technology fund and an index fund that also holds technology stocks. It also notes that a holding can grow into a concentration through gains alone, and suggests reading each fund's holdings for overlap.

How are employer shares taxed when taken out of a 401(k)?

IRS Topic 412 explains that if a lump-sum distribution includes employer securities and the payer reports net unrealized appreciation (NUA) in box 6 of Form 1099-R, that appreciation is generally not taxed until the shares are sold. The recipient may instead elect to include it in income in the year of distribution. Money rolled over to an IRA follows the regular IRA rules on later withdrawals.

How long does a typical stock stay listed?

Not long. The W. P. Carey magazine summary of Bessembinder's research puts the median time a stock appears in the CRSP database at 7.5 years. Stocks also leave through mergers, so a short listing is not always a loss.

A concentration check in five steps

List every holding as a share of all investable assets, combining brokerage, retirement and employee stock purchase accounts, and look through funds to their largest positions.
Mark any single stock, sector or asset above a limit written down in advance, such as the 10% figure FINRA mentions.
For each marked holding, multiply its share by 0.78, roughly the Nasdaq and bitcoin falls above, and write the result as a dollar loss on today's balance. For a single company's stock, also run 1.0, a total loss, since a company can fail outright, as Enron did.
For employer stock in a 401(k), check the plan's rules for when shares from employee and employer contributions can be sold and how often switches are allowed.
Decide what to do with a holding that has grown past its limit, for example trimming on a set date each year. The site's Rebalancing calculator shows the trades needed to return to a target; in a taxable account, count each sale's tax cost.