Shiller CAPE 39 and Forward P/E 23: The Honest Math on the Next Decade of Stock Returns

The Shiller CAPE ratio is at 39. The forward P/E is 23. The equity risk premium just touched 0.02%, essentially zero. The bears point at these numbers and predict a lost decade for stocks. I take the numbers seriously, and I still think the lost-decade call is built on a flawed foundation. This post is the honest version of the math, including the parts that should make bulls uncomfortable too.

The disagreement in one line: valuation matters, but the historical model assumes mean reversion in earnings growth, and AI is the largest productivity unlock since electrification, which means the earnings side of the equation may not revert the way the model expects.

First, respect the data. CAPE above 35 has historically led to roughly 0% real returns over the following decade. That is not a bear talking point. That is the record. Anyone fully ignoring it is gambling, not investing.

If you have never met the metric: CAPE compares today's price to the average of the last ten years of inflation-adjusted earnings. It smooths out booms and recessions to ask one question. How much are you paying for a dollar of normalized earning power? At 39, the answer is: a lot, by any historical standard.

Here is my issue with stopping there. The model treats every era as comparable. The decades it learned from include the Great Depression, World War 2, the oil shocks, and the dot-com bust. None of them, not one, had a productivity-enhancing technology being adopted across every white collar job within 24 months. The model is not wrong. It is trained on a world that did not include the variable now doing the heavy lifting.

When the inputs to a model change in a way the model has never seen, the honest move is not to throw the model out or to obey it blindly. It is to widen your range of outcomes. That is what I have done.

Let me show why the earnings side matters so much, slowly. If AI raises corporate operating margins by even 200 basis points across the S&P 500 over the next 5 years, that single change adds roughly $80 to S&P earnings power. Put even a 20x multiple on that incremental $80 and you get about 1,600 index points of justified valuation that the static models cannot see, because they assume margins revert to historical averages.

To be clear about what this is: a scenario, not a forecast. The 200 basis points might not materialize. But notice the asymmetry of the debate. The bears need margins to mean-revert downward against the largest productivity rollout of our lifetimes. The bulls only need a modest fraction of the AI promise to land. I know which side of that bet I find easier to defend.

Now the cold water. Yes, AI is real. No, that does not mean every stock at 30x earnings is a buy. The productivity gains will not be distributed evenly. The AI infrastructure plays and the businesses with genuine productivity leverage will earn their multiples. A long tail of expensive companies riding the theme will not. Dispersion between winners and losers inside the same index is going to be enormous, which means pricing discipline matters more in 2026 than at any time since 1999.

My working ranges. US large caps: 6 to 9% annual returns. International developed: 7 to 10%. Emerging markets: 8 to 12%. Cash and short Treasuries: 4 to 5%. The bears are modeling 3 to 4% for US large caps. The bulls are modeling 13%+. I think both are wrong, the truth lands in the middle, and the dispersion ACROSS asset classes is the actual opportunity hiding in this debate. When everything is priced for the same future, you get paid for owning the things priced for a worse one.

Why fight over a few percentage points? Hypothetical math. Say you invest $300 a month for the next 10 years, $36,000 of contributions. At 6% annual returns that grows to roughly $49,000. At 9%, roughly $58,000. Real money, but notice what did the heavy lifting: the contributions. In every scenario, including the bear one, the person who kept investing finishes miles ahead of the person who went to cash waiting for a crash that may never be timed right. The valuation debate changes your expectations. It should not change your habit.

Should I sell because CAPE is 39? A single ratio has never been a complete plan. High CAPE argues for moderating expectations and diversifying, not for exiting. The historical record on timing exits with valuation alone is brutal.

Does the near-zero equity risk premium scare you? It tells me stocks offer little cushion over Treasuries right now, which is exactly why my model holds cash and short Treasuries yielding 4 to 5% as a real allocation, not leftovers.

Why not just wait for cheaper prices? Because expensive markets can stay expensive for years while compounding passes you by. New money can go to the cheaper corners, international, EM, select names, without abandoning the core.

Holding core S&P 500 exposure. Adding fresh capital to AI infrastructure names with lower starting valuations than the Mag 7 (see MRVL and ARM). Holding 20% cash and short Treasuries yielding 4.5%. And ignoring anyone, bull or bear, who predicts the next decade from a single ratio with total confidence.

Read next: The 4.8 Trillion Mag 7 Month | Position Sizing in High-Valuation World

*Disclaimer: MentorSurge is not a financial advisor and this is not financial advice. This post is for educational and entertainment purposes only. Nothing here is a recommendation to buy or sell any security. Long-run return projections are scenarios, not promises. Investing involves substantial risk of loss. Numbers cited were accurate when written and change constantly. Always do your own research and consult a licensed professional before making decisions with real money.*