Valuation is not a timing tool. It is a return tool. Most investors keep skipping that part.
When U.S. stocks are priced this high, history does not promise another decade of easy double-digit gains. It usually promises something closer to low-single-digit returns, and sometimes nothing at all. From today's starting point, the valuation-based range is roughly -2% to +2% a year over a full cycle. That is not a crash call. It is the math of paying too much for future earnings.
The problem is not that stocks can go higher. They can. The problem is that too many portfolios are built as if they must.
"The data hasn't mattered until it is the only thing that will matter. Then it's too late, and investors have never been more complacent," says Revolution Group's Founder and CIO.
The Market Is Expensive by Almost Every Long-Term Measure
As of late August 2026:
The Buffett Indicator, total U.S. stock-market value divided by GDP, sits near 237% to 239%. Official readings earlier this year were even higher, around 250%.
In 1929, that ratio was about 65%. In 1987, about 85-90%. In 2000, about 170%.
Warren Buffett called this "probably the best single measure of where valuations stand at any given moment." His warning was simple: if the ratio approaches 200%, "you are playing with fire." We are well past that line.
The Shiller CAPE is around 41-42. The long-run average is about 17. Only the late-1990s peak was higher.
The S&P 500's trailing P/E is near 29.5.
Households now have about 46% of financial assets in equities, an all-time high and in the 99th percentile of history. That same measure peaked near 39% in 2000 and 42% in late 2021. Both periods were followed by large drawdowns.
None of this says the market has to fall next month. It says the price of ownership is already high. High prices do not create wealth. They consume future returns in advance.
What "Normal Returns From Here" Actually Look Like
This is where the -2% to +2% range comes from. It is not a guess. It is what valuation models have historically delivered after similar starting prices.
Shiller-based estimates recently implied about 1.2% to 1.4% a year from here.
Research Affiliates has projected about 3% a year for U.S. large caps over the next decade, far below what most 401(k) plans assume.
A blend of nine historically useful valuation indicators recently pointed to about -3.2% real per year. In plain English: stocks lag inflation.
John Hussman has shown the same relationship for years. The more you pay, the less you should expect over the next 10-12 years. At extremes, that expected return collapses toward zero or worse.
Other firms are more optimistic. Building-block forecasts from shops like J.P. Morgan still see something closer to 6-7% if earnings stay strong and multiples never fully mean-revert. That is possible. It is also the generous case, not the case that starts from price alone.
The Honest Summary
If valuations stay permanently high, you may earn a muted but positive return. If valuations normalize the way they have in every prior cycle, the next decade can look like -2% to 2%, or a long trip to nowhere after inflation.
That is why "stocks always make 10% a year" is a dangerous sentence at this price.
The Index Looks Calm. The Stocks Inside It Are Not.
S&P 500 1-month implied correlation is still in the single digits. Cboe's COR1M recently printed around 9, after readings as low as 7.5. Three-month implied correlation is only a little higher, near 10. Realized pairwise correlation among S&P names has been near 0.07.
Correlations have rarely, if ever, been this low coming out of a Q2 earnings season. In the AI age, they have kept moving lower.
That sounds like good news. It is not good news for people who own the index and expect the last decade to repeat.
Index variance is roughly average single-stock variance times average correlation. When the VIX is quiet and correlation is this low, the index looks asleep because winners and losers cancel. Constituents are not asleep. The average is.
That is why this tape can feel fine and still argue for muted index returns. You are not being paid for the excitement underneath. Single names can have huge years. The cap-weighted index is what is left after they offset. The AI age broke the old "one market" story. AI stocks and the rest of the S&P have at times moved in opposite directions. The index is two markets glued together by a ticker. Concentration plus low correlation is a fragile mix. A handful of expensive leaders can keep the average up. When that leadership stumbles, correlations usually spike. The diversification the options market is pricing today tends to disappear on the way down. Earnings season used to reset the tape. Coming out of Q2, correlations normally firm as the market digests a common story. This time, they did the opposite. That is not proof of health. It is proof the market is being driven by idiosyncratic AI outcomes, not a broad, cheap opportunity set.
Low correlation is useful for stock pickers. It is not a reason to raise expected returns on the index. If anything, it is a reason to lower them.
Paul Tudor Jones Connected the Pieces
In his conversation with Patrick O'Shaughnessy, Paul Tudor Jones made the valuation argument concrete. The United States is not just expensive. It is over-equitized.
His numbers: stock-market cap at 252% of GDP at the time of the interview. Significant mean-reversion episodes have shown up roughly every decade since 1970. A move back toward a more normal 25- or 30-year P/E would imply a 30-35% decline. A 35% drop on a market that is 250% of GDP is not a private portfolio event. It is 80-90% of GDP in vanished paper wealth. Capital-gains taxes are a meaningful slice of federal revenue. If those receipts collapse, the deficit widens, bonds can sell off, and the wealth effect feeds on itself. Households have never had more of their money in stocks. Private equity was about 7% of institutional portfolios in 2007-2008. It is about 16% now. The system is less liquid than it was going into the last crisis.
Then the line that matters for anyone buying the index today: history shows that buying the S&P at a P/E around 22 has produced negative 10-year forward returns.
The S&P is still a spectacular long-term asset if your horizon is a century. That century average includes years when the P/E was 6, 7, or 8, roughly one-third of today's multiple. The average hides the starting point. The starting point is what you actually own.
Risk Mitigation Is the Missing Piece
Most investors are not positioned for every version of what may be coming. They are positioned for one version: the last 15 years, again.
That version assumes U.S. large-cap growth keeps leading, multiples stay elevated, private assets stay marked smoothly, a 20-35% drawdown is only a buying opportunity, and bonds will hedge stocks when it counts.
"It's all about correct diversification and preparing for every version of what may be coming. Investors have lost their sense of risk mitigation. There is a strong likelihood returns will be from other hard-to-find asset classes away from equities over the next 8-10 years."
— Ryan Fleischer, Founder & CIO of Revolution Group
Jones's point is that a real mean reversion would not stay inside the stock market. It would hit household wealth, tax receipts, the deficit, and liquidity at the same time. Private equity does not become more liquid in that tape. It becomes harder to sell.
Diversification that only owns different flavors of the same expensive U.S. equity bet is not diversification. It is concentration with extra steps.
Preparing for every version means owning assets that can still work if U.S. multiples compress: cash and short-duration quality when yields compensate you, cheaper international and value exposures, real assets that are not priced like a growth stock, and enough liquidity to buy the discount if it arrives.
The goal is not to predict the crash. The goal is to survive the range of outcomes that current prices already imply.
Valuation matters. Correlation this low in the AI age does not cancel that. It shows why the surface looks safe while the expected return on owning the whole market should stay muted.
Fleischer went on to say, "Our clients are not average, and they don't want to lose money. We are taking everything into consideration from a tax and risk mitigation standpoint. Hedging and using a modern version of Portfolio Theory is a big part of what we do. When the market cycles, our clients are glad to have their custom portfolio."
The data has not mattered. Until it is the only thing that will matter. By then, it is too late.