The lost decade thesis — the idea that U.S. stocks will deliver near-zero returns into the early 2030s — is the most important market debate happening right now. It’s also the most misunderstood.

“Buy and hold is dead” has been declared dead so many times it’s almost a joke. The phrase gets trotted out every time the market drops 10%. But this time the argument comes with receipts. Goldman Sachs, Morgan Stanley, Bank of America, and Apollo have all published research projecting U.S. equity returns well below the historical 10% average for the next ten years. The question is whether they’re right.

What Goldman actually said

In October 2024, Goldman’s then-chief U.S. equity strategist David Kostin forecast the S&P 500 would return just 3% annualized over the next decade — a number that would place it in the 7th percentile of all 10-year returns since 1930. A striking call from the bank that writes much of Wall Street’s playbook.

Thirteen months later, in November 2025, Goldman’s chief global equity strategist Peter Oppenheimer published the firm’s official 10-year forecast. The number: 6.5% annualized total returns for the S&P 500. Better than 3%, but still well below the 10% historical average. Oppenheimer’s base case breaks down to 6% annualized earnings growth, a 1% annual drag from valuation compression, and a 1.4% dividend yield.

Goldman also ranked U.S. equities dead last among all major global markets — behind emerging markets (10.9%), Asia ex-Japan (10.3%), Japan (8.2%), and Europe (7.1%).

By June 2025, Goldman’s new chief U.S. equity strategist Ben Snider had bumped the number to 7%. Even so, that’s still three percentage points below the long-term average. The range of Goldman forecasts — 3% to 7% — tells you something: the bank itself isn’t sure how bad it gets, just that it’s below average.

Apollo’s chief economist Torsten Slok also called for a flat decade. These are outlier forecasts, but the math behind them is defensible.

The CAPE ratio and why it matters

The core of every lost decade argument is the cyclically adjusted price-to-earnings ratio, or CAPE — also called the Shiller P/E. It takes the S&P 500’s inflation-adjusted price and divides it by the average of its real earnings over the prior ten years. By smoothing the business cycle, it gives a steadier read on valuation than a one-year P/E.

The S&P 500’s CAPE ratio sits at 41.37 as of July 2026. That’s in the 99th percentile of the full dataset going back to 1881. It is 138% above the historical average of 17.4. The long-term median is 16.1.

Every time the CAPE has been this high, subsequent 10-year returns have been meaningfully below average. Not always negative. But always lower. The mechanism is simple: when you pay 41 times a decade of smoothed earnings for a dollar of earning power, a lot of future return has already been borrowed.

There are counterarguments. Ben Snider at Goldman makes the best one: profit margins at 13% (versus 5.5% in 1980) and structurally lower interest rates mean valuations deserve a structural premium. The CAPE’s historical average was set in a different interest rate regime. Maybe this time really is different.

Maybe. But “this time is different” is the most expensive four-word phrase in investing history.

The historical precedent

The lost decade claim sounds apocalyptic. It has happened before.

The Nifty Fifty period of the late 1960s and early 1970s created the same setup. Fifty high-quality growth stocks — Xerox, Polaroid, Avon — were deemed “one-decision” stocks: buy and hold forever. The top cohort traded at 42 times forward earnings. The broader market peaked in 1972. Then inflation, oil shocks, and Nixon-era policy chaos triggered a 50% drawdown. The Dow crossed 1,000 in 1972 and didn’t reclaim that level for a decade. From December 1964 to December 1974, the largest stocks returned -0.44% annualized. A decade of dead money for the cap-weighted index.

The dot-com bubble produced the same outcome. From March 2000 to March 2010, the S&P 500 returned -0.99% annualized. The top ten tech stocks lost an average of 74% peak to trough. The cap-weighted index needed until 2013 to reclaim its 2000 high. That’s 13 years.

Both lost decades followed periods of extreme market concentration. In 1972, the five largest stocks made up about 25% of the S&P 500. In 2000, the top seven tech names represented 22-24%. Today, the Magnificent Seven make up roughly 34% of the index — the highest concentration in modern history. The CFA Institute published research in 2025 showing that after periods of extreme concentration, the largest-cap stocks systematically underperform over 1-year, 3-year, 5-year, and 10-year horizons.

The pattern is a recurring structural feature of equity markets, observed each time concentration has reached historical extremes.

Is the comparison valid?

This is where the bull case and bear case diverge, and both sides have real arguments.

The 2000 tech bubble had top stocks trading at 56 times forward earnings. Today’s Mag 7 trade at roughly 28 times — expensive, but not lunatic. Earnings growth has been phenomenal. Balance sheets are stronger. The underlying businesses are actual cash-generating machines, not speculative stories.

But those are arguments about the downside rather than the upside. The bear case doesn’t require a crash. It only requires mean reversion — a slow, grinding compression of multiples from today’s 99th-percentile CAPE back toward something more normal. That alone would produce sub-7% returns for a decade. That’s the baseline forecast from Goldman, which is hardly a perma-bear shop.

If the optimists are right and we get 7% annualized for ten years, that’s $1,000 turning into $1,967 — versus $2,594 at the historical 10% rate. The difference is $627 per thousand dollars. Not catastrophic, but meaningful.

If the bears are right and returns are flat to 3%, that same $1,000 becomes $1,000 to $1,344. That’s nearly a decade of dead money.

What buy and hold defenders miss

The reflexive defense of buy and hold — “time in the market” — is true on average. But averages conceal the fat tails. The 1970s and the 2000s each produced negative real returns for a full decade. Anyone who retired in 2000 with a portfolio 60% in the S&P 500 saw their nest egg shrink for 13 years.

Nobody has a 20-year time horizon for all their money. People retire, buy houses, and send kids to college. A decade of dead money in your peak accumulation years is a real problem, not a theoretical one.

The lost decade thesis isn’t saying the market is about to crash. It’s saying the next ten years’ returns may look more like 1965-1975 or 2000-2010 than 2010-2020. That’s a regime shift, not a disaster. But it makes a passive buy-and-hold strategy substantially less attractive than the backtest everyone cites.

The question Keith Kaplan and the marketwise community have been asking — if the next decade delivers 3% to 7% annualized, what does a portfolio plan look like — is the question the lost-decade thesis actually raises. It reframes the buy-and-hold debate from “will the market go up” to “what return environment the next ten years sit in, and what that return environment means for a portfolio built on the assumption that the next decade looks like the last one.”

For the full index, see Promo Watch.