The Limits of Bullishness

A Market Update on The Missing Billionaires

Note: with some summer busyness coming up, and my next pieces likely to be delayed due to being longer and higher effort, I’m sending this somewhat more time-sensitive, shorter post to get it “on the record” only a week after my last dispatch.

The Missing Billionaires is a book I keep returning to as a foundational work of personal finance. Strangely, it does a poor job of explaining its thesis (a friend of mine thinks it fails to account for confiscatory taxation of fortunes) while presenting a rational, market-agnostic method of investing. Technical readers in particular will appreciate its deep dive into the mathematics of probability and favorable gambles, the largest of which is the public equity market.

They Really Will Drop Money from Helicopters

For better or worse, I’ve spent much of my life being overly pessimistic about the general economy and investments. Only after the 2020 Covid crash did we become fully invested when it became apparent that a) Covid was not the end of civilization, not even as bad as the Spanish flu, which preceded the Roaring 20s, b) certain mass delusions about green energy (perhaps long-term directionally correct, but way too early) gave old economy natural resource stocks insane low-end valuations that were too incredible to pass up, and c) the 2008 crisis was confirmed not to be a one time exception in demonstrating that a massive deflation like The Great Depression can never happen again because fiat money gives central banks the means to prevent it. The pandemic showed their repeated resolve to drop infinite money from helicopters, just as Ben Bernanke promised.

The latter implies that cash and dollar-denominated bond investments are always a long-term sucker play, only useful to those whose liabilities consist of nominal dollars, such as life insurance companies or pensions. Normal investors have lifestyle liabilities in real dollars, and holding fake ones without a specific warrant is to bleed out wealth slowly. If the real risk is holding fiat-denominated assets, some smart commentators question whether there is such a thing as a durable risk premium in equities.

Is the Efficient Market Hypothesis Correct? Sort Of.

All of this would seem to validate the Efficient Market Hypothesis, and its philosophy of buy-and-hold through dollar-cost-averaging into minimal-fee mutual funds. One danger, however, of this is that it has no limiting principle. If the price of stocks is always fair, why would we expect them not to get bid up to the point where they have no advantage over alternatives?

The key insight of The Missing Billionaires was to accept EMH at a broad level but use different market conceptions of it, critically comparing the earnings yield of the S&P 500 to that of the single dollar-denominated exception to inflation risk, TIPS securities. My original review details that methodology. It was brilliant in unifying theories of positive-expectation gambling, governed by the Kelly equation, with allocations in a model portfolio of stocks and inflation-protected bonds. This approach outperforms, on a risk-adjusted basis, buy-and-hold by avoiding extreme market valuations and providing a reliable trigger for going all-in when the market is at historic values. It is not necessary to follow the model exactly to get an objective read on market risk.

Based on first principles thinking, it demonstrates why Shannon’s Demon works in the markets, and why approaches like Buffett’s 90-10 portfolio (his recommendation for average investors, 90% broad stock indexes, 10% t-bills, constantly rebalanced, which juices returns by keeping a cash buffer to capture value as the market falls) and The Permanent Portfolio (25% stocks, 25% cash / money market, 25% gold, and 25% long-term bonds) can often produce superior returns. By systematically selling appreciating assets and buying depreciating assets, investors end up owning better values over time and taking profits off the table in times of extreme valuation.

The key mechanism of the SPX/TIPS portfolio is comparing the S&P’s CAPE yield to the real yield of TIPS, and adjusting for the volatility of the former. CAPE stands for cyclically adjusted price-to-earnings ratio, which smooths current earnings by incorporating a ten-year window, normalizing those for inflation, and dividing today’s price by the result. Since earnings are cyclical, often unsustainably high or low at market tops and bottoms, respectively, this calculation is empirically proven to predict 10-year forward market returns better than the current earnings yield or long-term average returns. To predict where the S&P will be in ten years, simply divide 1 by the CAPE. Currently it’s 41.71, implying an annual return of 2.4% over the next ten years. By comparison, the current risk-free TIPS yield (i.e., the return exceeding inflation) is 2.2%. The S&P currently provides very little compensation for risk.

The model works on a very simple principle: capital seeks the highest risk-adjusted returns it can receive. When risk-free real rates are low or negative, it disciplines bears to buy stocks even if they seem overvalued historically. When real rates are high relative to equity valuations, it forces bulls to limit exposure.

Post-TMB Refinements

The authors, who manage money at Elm Wealth, later refined their measures to be more bullish, as while CAPE is superior to alternatives, it still tends to under-predict stock returns somewhat. By incorporating an iterative calculation that compounds retained earnings at the CAPE yield, they derive a new ratio called P-CAEY, and show that it better predicts both forward returns and earnings than CAPE. That modified earnings yield, the expected annual return over 10 years, now sits at ~2.86% for the S&P, using March earnings and June prices (since earnings are calculated quarterly, the numerator remains fixed for three-month windows).

Another modification mentioned but not precisely quantified in the book was further enhancing returns with the use of momentum. While the S&P has average volatility of 20% — the math is complicated but this means that the average deviation will be ~16%, so if expected return is say 5%, then investors would see an average range of returns anywhere from -11% to 21% in any given year — history shows that when it is rising, as indicated by being above its one-year moving average, volatility is lower at ~15%, and when falling (below the one year average) volatility is higher at ~27%. Bull markets are safer than bear markets.

By switching out the volatility measure for the current market state, higher stock allocations can be justified at higher valuations that exceed the risk-free rate, riding out a top, and faster exits when momentum deteriorates. This can result in frequent trades in and out at market transitions, but this is less important because trading costs for highly liquid instruments like S&P funds are low generally, and in practice they sample momentum states weekly and limit any week’s change to only 25% of the portfolio. They also critique concerns about taxes as being likewise often mathematically minimal in consequence (and non-existent for most investors who utilize tax-deferred retirement accounts).

Application to June 2026 Valuations

Thus, I have been convinced, against my natural pessimism, to trust strangers with my money across a wider range of conditions: that my default ought to be owning equities and other assets that produce income. Sitting around in t-bills whining that we can’t achieve the charmed Boomer-tier returns of Warren Buffett by buying companies for less than cash in the bank would have been ruinous over the last 30 years. Something has definitely changed in the market, and there does seem to be a general glut of capital. We should not expect bear markets to be as long and deep as they were historically. Robust 5-10% real returns are treasures when found.

That said, it does not do to throw in the towel and become a perma-bull. Within living memory, the S&P underperformed t-bills from the 2000 peak to around 2013. This risk is more pronounced for those with high wealth relative to income (e.g., retirees or family offices with legacy wealth), who will not benefit as much from future dollar-cost averaging. And it’s most perilous when valuations reach extremes.

The challenge then is to discipline a bias toward bullishness, subject to a limiting principle and an EMH-compatible, emotion-agnostic framework. I find that in the modified Merton models from the book, in my assenting to the authors’ latter-day bullish modifications for retained earnings and momentum, and rest on the empirical evidence, rather than staying skeptical about management’s ability to reinvest capital or the inherent irrationality of trend-following. I must submit myself to the gods of the straight lines and sacrifice my congenital bearishness on their altars.

Where does that take us in June 2026? With Codex’s help, I built a Google Sheet implementing this most maximally bullish rational model that I can in good conscience endorse. Featuring live Google Finance and FRED data, both SPX and TIPS yields are refreshed daily at close, with only the P-CAEY value needing manual revision quarterly (the price component of the proportion updates daily, with the derived numerator for earnings, which moves slowly anyway as a ten-year average, holding steady for the quarter).

The current recommended S&P allocation as I write this in mid-June 2026? Only 13.9%! Most of this is due to the momentum signal. If the S&P were to go below its one-year moving average, the target would shrink to 4.3%.

One interesting twist is that Elm does not use their own models! Why? I speculate that clients are resistant to following them, as it would be extremely psychologically difficult to go to a 14% exposure in today’s market. It’s very hard to sit out when everyone else is making easy money. They instead feature a “target allocation” of stocks and use the Merton models to modify it, along with adding in some other sectors such as emerging markets and real estate. For most clients who will only tolerate 50% of their “dynamic scaling,” this means a minimum exposure of 42.5%, which is admittedly worlds better than 100% for perma-bulls or 0% for perma-bears. Most people want to have some chips in the casino, regardless of valuations. But strictly following the model, that’s ~14% right now.

And as much as I fight my skeptical priors, I can’t help but notice a recent study by Visa showing that much of the recent economy is driven by wealth effects:

There is a wide body of research on the wealth effect, with most studies concluding that consumers spend between 4 percent and 15 percent of newfound wealth. According to our estimates, the wealth effect between 2002 and 2017 was 9 percent. Said another way, for every $1 increase in household wealth, consumer spending increased by 9 cents. Over the last few years, however, something changed dramatically. Using data through the third quarter of 2022, we find that the wealth effect has increased to 34 cents, almost quadruple the pre-pandemic average. For wealth held in stocks, bonds, and pension entitlements, the spending responsiveness was 24 cents. For wealth held in owner-occupied housing, it was 20 cents.

For comparison, the Federal Reserve found that the long-term average for this measure from 1964 to 2011 was 3.5%, so we are looking at an almost 10x increase in people spending paper gains in the economy. That is, for every dollar of marginal wealth in stocks or housing, people would spend about 3.5 cents annually, which seems reasonable. Visa says that number is now 34 cents, or fully a third of unrealized gains. Paper net worths as a percentage of income are at historic highs.

This strikes me as unsustainable, and accounts for why there appear to be so many “rich” people around. When we vacationed earlier this year in Hawaii at a ridiculously expensive resort, but apparently not expensive enough, I was shocked by how many fat people with garish tattoos were waddling around eating $26 personal pizzas. Where do these people, who clearly have a high time preference, get the money to afford this? The answer, I think, is that they cannot, but spend anyway because that is their expectation as 401k balances balloon.

It’s Never Felt Dumber to Be Smart

People with longer-term outlooks, who are trying to spend wisely and save, have never felt poorer. In my little corner of Texas, median incomes are about $45,000 per worker, but you wouldn’t know it with most people’s lifestyles. By the numbers, nothing makes sense in today’s economic environment.

However, aggregate debt and debt service levels are reasonable. The economy is indeed “K-shaped” at the moment, with a larger-than-ever prosperous class and a shrinking middle class. The top 10% of earners account for about half of spending. The wealth effect may be accelerating an eventual rise to unsustainable borrowing, but it hasn’t happened yet.

2000 Redux?

The closest analogy seems like 2000, not 2008. The standard weighting of indexes is by market capitalization, which means that of the 500 S&P companies, investors end up with less diversification than they might expect, because they own more of the top stocks. The higher certain stocks go, the more standard indexes force passive investors to hold them. Currently, the top 10 SPX stocks account for ~38% of the index, and the top 50, ~63%. This is a historically unprecedented concentration:

Alternative funds feature equal weighting, sidestepping this issue, though with the liability of theoretically owning more of lower-quality companies than investors might prefer.

Currently, the low earnings yield of the SPX is almost entirely an artifact of tech companies, because extreme valuations are notably not present on an equal weight basis, nor in mid or small caps. Using a linear proportion to estimate the forward P-CAEY (messy but good enough):

Using the standard model and assuming comparable volatility, the rational allocation to the equal-weight S&P is 25%, and up to 45% under the momentum-informed model:

This is somewhat overpriced, but hardly severe crash territory, for the equal-weight S&P. Meanwhile, the standard S&P has a worse concentration overall than during the 2000 Internet bubble, though with a somewhat more favorable earnings yield relative to the risk-free weight. ChatGPT summarizes the numbers:

And in that era, the equal-weight SPX did much better, declining only 20-25% by the best estimates compared to 78%, 49%, and 38% for the cap-weighted Nasdaq, SPX, and Dow 30, respectively. Note that things don’t have to be as extreme as 2000 to get a crash; for example, 2007:

And comparing both to today (numbers compiled 6/23/2026) with the simple, less invested Merton model for relative comparison:

So right now we’re somewhere between the 2007 and 2000 peaks in terms of valuations. Some caution, or at least reallocation away from the high-fliers through an equal-weight index, is cheap insurance. I hate to be forced into a semi-bearish posture, especially given my previous errors, but what cannot go on forever will eventually stop.