---
title: "Foundations: U.S. Market Structure & Value"
source: "https://michaeljburry.substack.com/p/foundations-us-market-structure-and?r=2xl6z&utm_medium=ios&triedRedirect=true"
author:
  - "[[Michael Burry]]"
published: 2026-03-03
created: 2026-03-12
description: "Powerful Trends, Increasing Fragility & Coiled Tension"
tags:
  - "clippings"
---
### Powerful Trends, Increasing Fragility & Coiled Tension

As you know, many on CNBC and across the social media spectrum are very excited about AI and the potential for very high stock valuations continuing and even expanding.

$6 trillion. I see $6 trillion! Can we have a $7 trillion? $7 trillion! Can we have a $10 trillion? $10 trillion oh my!

I have a different view rooted in the fundamental arguments I have made, so far, including **[Cardinal Sign of A Bubble: Supply Side Gluttony](https://open.substack.com/pub/michaeljburry/p/the-cardinal-sign-of-a-bubble-supply?utm_campaign=post-expanded-share&utm_medium=post%20viewer)**, **[Unicorns & Cockroaches: Blessed Fraud](https://open.substack.com/pub/michaeljburry/p/unicorns-and-cockroaches-blessed?utm_campaign=post-expanded-share&utm_medium=post%20viewer)**, as well as **[The Supply Side Gluttony Recurrence](https://open.substack.com/pub/michaeljburry/p/the-supply-side-gluttony-recurrence?utm_campaign=post-expanded-share&utm_medium=post%20viewer)** and the **[Blessed Fraud Recurrence](https://open.substack.com/pub/michaeljburry/p/the-blessed-fraud-recurrence?r=4repfn&utm_campaign=post&utm_medium=web)**.

I have also penned my bearish take on Palantir with **[Palantir’s New Clothes: Foundry, AIP & the Failure of Reason](https://michaeljburry.substack.com/p/palantirs-new-clothes-foundry-aip?r=4repfn)**, and **[Palantir, An Accounting](https://michaeljburry.substack.com/p/palantir-an-accounting?r=4repfn)**.

I have been attacking the narrative, warning of failed expectations.

I have yet to tackle directly the overvaluation of the market itself, which is generally a good thing - for a very specific reason. Along these lines, however, I do track something that I want to share with you.

Most of us are familiar with the long-term logarithmic chart of the U.S. stock market. It is a glorious thing, and looks like a remarkably steady march over the last century.

![](https://substackcdn.com/image/fetch/$s_!ZhAC!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F05dd11f6-7f9d-42c7-93e5-884cd4a807a7_1460x977.jpeg)

A baby born today, put $1,000 in an index fund and be done with it, right? That’s the idea behind 530A accounts, a.k.a. Trump Accounts, delivered by the One Beautiful Bill, perhaps at an auspicious time.

Nevertheless, a long chart like that glosses over some extremely painful periods that would require almost inhuman fortitude of an individual investor or portfolio manager. Few can hold through a decade or more of significant losses – that is fighting against human nature in a dozen different ways.

Below, a chart of the **Dow Jones Industrial Average from 1900 to the present day**.

I added some perspective. The blue line is the Dow Jones prices arranged logarithmically. The red line is a more sobering view, which is the Dow Jones adjusted for inflation.

![](https://substackcdn.com/image/fetch/$s_!cMfu!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8e9ddea0-705e-4450-94ab-efa3417af83d_1890x1218.jpeg)

Additionally, I used the average **Shiller cyclically adjusted PE ratio (CAPE**) from 1900 (purple dotted line) and from 1990 (orange dotted line) to chart the Dow Jones as if it were at one average constant PE ratio the entire time. This takes out variance in PE multiple over time. Those two dotted lines look less violent, less dramatic.

Multiple compression and expansion comprise a large portion of the volatility in stocks over the years.

Also, note the vertical light red shaded areas are secular **inflation-adjusted bear markets**. That is a 100-year chart, so right off the bat you notice, those last a long time and are as common as inflation-adjusted bull markets.

**For the Dow Jones, the inflation-adjusted drawdowns were, from past to present, 65%, 74%, 65%, and 39%.**

The pattern in the first three bears is devastatingly consistent. The real trough didn’t stop at fair value — it overshot dramatically below either PE-adjusted line before reversing. In 1921, in 1932, and at the 1974 + 1982 bottom, the market was trading at Shiller PEs of 5–8x — roughly half the century average, let alone the modern era average.

The 2000–2013 bear is the outlier – its trough in 2008–2009 landed roughly at the 1990+ average PE line and modestly above the century average — meaning the market found support at “fair value” rather than overshooting into deep undervaluation, on this metric.

I gave the S&P 500 Index the same exact treatment, starting at 1928, instead of 1900, as you can see in the chart below.

**For the S&P 500, the inflation-adjusted drawdowns were 58%, 55% and 43%. The S&P 500 Index does not capture the 1906-1921 drawdown.**

![](https://substackcdn.com/image/fetch/$s_!5kbN!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa286dee1-f734-42d3-8bf7-dc2c9584a028_1605x1084.jpeg)

Those are significant, painfully deep and painfully long drawdowns. Stocks protect investors from inflation, but not without serious patience, a good diet and regular exercise.

Here too, with the S&P 500, we see the 2000-2013 bear is, not surprisingly, an outlier. Of course, never had there been such scale of policy intervention: zero interest rates, quantitative easing, interest on excess reserves held at the Fed, TARP, nationalizations of three large financial concerns. The government’s backstop was powerful and effective in its delivery of moral hazard for future generations.

**I have maintained for some time that there was no natural conclusion to the bear market of 2007-2009.** Zero interest rates, IOER, and other crisis-era policies remained in place for a decade or more. AIG’s rescue saved Goldman Sachs, and the whole narrative shifted, ushering in our current era of populist politics and nihilistic youth.

But I digress.

### Interest Rates, Who Needs Them?

Truth is, avoiding those drawdowns is hard. In fact, people lose a lot of money trying to do just that. Many still try, and interest rates and inflation are big foci for investors. For this very reason, few government events are more highly anticipated or analyzed by markets than ones involving the Federal Reserve Chairman.

Recent rate and inflation history put the interest rates/stocks theory to a serious test, however.

Heading into the late 2010s, rates had been falling, short term rates had been near zero for almost a decade.

Many thought that higher interest rates, when they came, would hurt stock market valuations. Would bring price-earnings multiples down, and could even hurt earnings. Certainly private equity’s exceedingly long honeymoon would come to an end, as would commercial real estate. So we thought. So I thought.

In the wake of COVID, nearly $7 trillion in government stimulus dovetailed with supply chain disruption. Higher rates arrived, riding that pale white elephant, inflation.

![](https://substackcdn.com/image/fetch/$s_!78UU!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9ca2914c-84a6-4f8d-abfd-92fb27a50f0d_624x401.png)

10-year yields spiked and broke a four decade downtrend, but the 450 basis point increase in rates saw the Shiller PE **increase** from 28.4x to 40.1x. Multiples expanded, stocks rose.

In fact, the S&P 500 more than **tripled** off its COVID low of 2192, even as both short-term and 10-year rates rose from near zero to near 5%.

The conventional **Fed Model** – the idea that earnings yields (inverse of the PE ratio) should trade at a positive spread to risk-free Treasury yields, simply did not work this cycle. Neither did it work in the post-World War II period (1940s-50s) nor did not work in the 1990s.

Ed Yardeni named the Fed Model in 1999, just before the 2000 technology bubble peak. But it has been of little use, it seems, and few discuss it anymore. Which is curious, as it was an attempt to link stock prices to interest rates, a common conceit of many.

As it is, we have quite a bit of proof, historical and recent, that interest rates have little to do with the earnings yield on the market, and therefore little to do with valuation ratios like price/earnings.

If one takes interest rates from valuation, then multiple expansion and multiple compression amounts to a beauty contest. Discounted cash flows mean little if interest rates mean little. Outside of extreme rate scenarios that forces rationing of credit and shrinking of the economy, the fight becomes one between **narrative capture** and **expectations disappointment**.

This is in fact why I have spent so much time on the AI narrative and related investor expectations that I expect to disappoint- and why I have generally not spoken of market multiples, or market overvaluation.

That ends today.

### There Comes A Time in the Life of a Man...

Perhaps, there are times when the extremes are so extreme, it might make sense to sit back and do some of that simple price and earnings math. Maybe that is what Buffett is doing with his $373 billion of cash, maybe that is what Ryan Cohen is doing at GameStop, and it is indeed what I am doing now.

Using some math, we can calculate potential drawdowns from current levels based on historical data sets.

As a warning, the potential drawdowns in the market look shocking, unreal, not possible.

To prepare for such a sight, there are some concepts we all should understand. These are long-held theories of mine, some of which you may have seen in the news now and again. You are in for an explanation or three.

Now, I said above, multiple compression and expansion is a large portion of the volatility in stocks over the years. This is obvious from those squiggles above that hold PE constant.

This quality can be measured, of course. Below I plot the results of an **Annual Log Return Decomposition Analysis** run over nearly the last century, from 1928 to 2026. I use the Shiller cyclically adjusted PE ratio, Shiller earnings data, and price and rate data from Bloomberg.

The decomposition is pretty straightforward. Since **Price = Earnings × PE**, any change in price can be split (decomposed) into precisely what came from earnings growth and what came from multiple expansion or compression.

The top chart below breaks down each year’s return into earnings growth and multiple expansion/compression. The bottom chart is a **10-year cumulative rolling logarithmic attribution**.

![](https://substackcdn.com/image/fetch/$s_!njdu!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F276bb65f-8f14-455c-bf87-a8ff076d9925_1200x1097.jpeg)

The chart takes some study and thought, I know. Visually, the rolling attribution chart has a long cyclical pattern much like the inflation-adjusted logarithmic stock-market chart. Today, in both, we have been well up for some time.

The math shows that Shiller PE multiple and total stock market price have roughly identical annual volatility (19.0% vs 19.1%). Earnings are more volatile (26.1%) but are very strongly negatively correlated with PE changes (r = −0.94) — when earnings collapse, the PE increases. Makes sense.

The 1966–82 secular bear saw PE compression drive the entire loss while earnings actually **grew**. During the “lost decade” of 2000–09, PE compression of −98% **overwhelmed** earnings growth of +46%.

In the 1982–2000 bull market, on the other hand, PE was the major factor, but earnings helped as well, especially early. As that bull aged, in the later 1990s, PE expansion really took over.

**Multiple compression is the primary mechanism of bear markets.** For bull markets, multiple expansion shares the stage with earnings growth.

### Reversion to the Mean

Now let’s look at some PE multiple on an absolute basis. The Shiller PE ratio to start 2026 was 40.2x, still short of April 2000’s 43.5x. The 1929 crash happened from September 1929’s 32.6x, and the 1970s bear market started with December 1968’s 22.2x. The GFC fell from May 2007’s 27.5x. **The abbreviated bear in 2022 started at 34.5x and did not complete a full multiple correction cycle like the others.**

Apart from the current run above the mean, the Dow Jones and S&P 500 have **always** reverted to their 1900-present and 1990-present mean Schiller cyclically adjusted PEs. Always.

Below we look at these mean reversions.

![](https://substackcdn.com/image/fetch/$s_!jjk3!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F426ef667-03c3-486b-8319-08b31b7919da_705x513.jpeg)

**For each index, and for each average PE, the current streak is a record, and it is not close.**

34 years since touching the 1900-2026 Average Shiller’s Cyclically Adjusted PE for the Dow Jones.

34 years since touching the 1928-2026 Average Shiller’s Cyclically Adjusted PE for the S&P 500.

The prior record was 10 years, from 1960-1970, and preceded a devastating period for stocks. The current streak 3.4 times that prior record, and over 10 times the historical average cycle of 3.2 years.

If we look at the modern era 1990-present data, the average Shiller Cyclically Adjusted PE is about 27.4x for both the S&P 500 and the Dow Jones, almost 50% higher than the average for the century-plus periods.

Relative to this higher modern era mean, it has been 9 years for each index since mean reversion. 1990 is almost too recent for validity and generalizability. The prior mean reversions were 6 years and 1 year. The data set is not large enough to be interesting.

The longer-term records, however, are interesting, and each index is instructive in this regard.

Below, the S&P 500 Index mean reversion history over about a century.

![](https://substackcdn.com/image/fetch/$s_!jhSy!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8b152aaa-64ed-4ea4-9bb5-0f2990c9d356_1015x669.jpeg)

Note the reversion mean PE varies over the years – earlier being lower, more recent being higher. Not only is the 34 years a record, the Shiller CAPE of 19.4x is **relatively high** and easier to reach than prior periods. One might expect this with a very long period above the mean.

Same is true with the Dow Jones Industrials, to a lesser extent.

![](https://substackcdn.com/image/fetch/$s_!k2EN!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F8b42909e-e3f0-4f01-9f38-ab76a5508bf0_964x775.jpeg)

Before the current 34 year streak of non-reversion, the 10 year streak really stood out. In fact, the rest were all just a few years apart.

The modern investor experience is an exceptional period of expanded multiples on stocks, and with the Shiller PE at 40.1x, as mentioned, it is the second highest peak (to be announced) on record. Of course, from the first place peak, the NASDAQ fell nearly 80% from 2000, and the S&P 500 fell about 50%.

So now, with that context, I present the downside scenarios based on valuation and mean reversion for the Dow Jones. PE27 references the modern era average Shiller PE, and PE18 is the 126 year data set.

![](https://substackcdn.com/image/fetch/$s_!IIHG!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F45c4e82a-d8c8-4f41-8459-4347f627c247_678x468.jpeg)

Below, the same for the S&P 500 Index. PE27 is again the modern era data set, and PE19 the 100-year one.

![](https://substackcdn.com/image/fetch/$s_!D_IF!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F82a5461f-b629-4197-8363-641db6b53383_670x452.jpeg)

Those declines look like the declines after 2000, after 1929. To me they also look like the kind of declines that might be expected from an extreme disappointment of exuberant speculation on a capital asset buildout, with both margin compression and unit volume compression.

Maybe something has changed, however. With respect to time of above the mean valuation, 34 years is very different. So, **what is different this time?**

### Market Structure

Since 2000, there has been an explosion in passive investing through index funds. Index funds buy when cash flows in, and they buy the whole index, regardless of relative or absolute value variety within the index. This is not traditional smart money, or intelligently driven money. It just might be idiot savant money. In any event, zero price discovery happens at the individual stock level.

Indexed investment now dominates trading in stocks (>60% in passive equity strategies), and this is seen as an automatic, permanent and growing bid. Sticky money, diamond hands.

This did not happen by accident.

401(k) plans were created in 1978 by Congress. It was not until the automatic payroll deduction for investment in 401(k) plans became available in the mid-1990s that 401(k) assets became big driver of growth in passive investment in indexed funds.

In 1998, **IRS Ruling 98-30** blessed automatic 401(k) contributions, and in 2006 the **Pension Protection Act** created the **Qualified Default Investment Alternative** framework, making index funds the legal default for these automatic contributions.

Somewhere around this time President Bush was making noise that to save Social Security it needed to be invested in stocks too. He was not wrong. I doubt he did this analysis, but maybe one of his aides did.

The impact was immediate. In 1996, passive investments were 6% of equity fund assets. By 2010, 19%. Then passive index investment really took off. **30%** by 2015, **48%** by 2023, and **60%** of the market in 2024. Accelerating rapidly.

These charts below – top is the **passive share of US equity funds** over time; bottom is the **total Defined Contribution (DC: 401K and IRA) asset** s- show the growth in passive investments is highly correlated and driven by the growth.

![](https://substackcdn.com/image/fetch/$s_!b03O!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F4161d983-6a65-4141-a2bd-47718c50bae3_744x666.jpeg)

Defined contribution plans (401K + IRA) surpassed Defined Benefit (Pension) assets in **2012**. This is represented by the **vertical red dashed line**, above.

Especially notable is how the growth seems to be accelerating lately - 48% in 2023, 60% in 2024. 2025’s numbers are not available at this writing, but the totals are certainly much higher.

This is a powerful **structural bid** and driver of the stock market. Below I plot the **S&P 500 from 1990 to present** alongside **total defined contributions (401(k) and IRA)** and the **percentage of assets tied to passive investing.**

![](https://substackcdn.com/image/fetch/$s_!A8SA!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffbb3b26f-709e-427b-9646-0637be222922_957x722.jpeg)

As passive investment grows, active managers are by definition squeezed out. Active managers are tasked with beating an index that does no analysis and gets a new shot in the arm each and every business day. **Each dollar that moves from active to passive is another dollar removed from price discovery - long the primary social reason for stock markets to exist.**

Market-based price discovery facilitates capital raising, which traditionally went to the better businesses with better securities prices, and helped sort wheat from chaff in business and industry. Not so much anymore.

There are other forces at work too. **Demographics** is one. 401(k) growth is to a very large degree a function of the demographic whale that was the baby boomer generation. During their peak earnings years, they replaced the defined benefit plan – the pension, actively managed, mostly invested in bonds – with the self-directed 401(k) system and overwhelming self-directed into equity index funds. **This was a titanic, decades-long inexorable shift from bonds to equities without historical precedent.**

Blackrock, Vanguard and State Street are three of the biggest index players, and they control over $20 trillion assets alone.

Another force at work is mechanical, machine-like reflexivity. The largest indices are market capitalization weighted. The larger the company’s market capitalization, the greater percentage of the index, and hence the greater percentage of 401(k) money that mandatorily flows to that larger market capitalization stock. Which helps increase the market capitalization of that stock more than a smaller stock in the same index. And around again.

It is no wonder interest rates seem to matter less and less to stock valuations. A growing majority of assets are invested never having to create a valuation spreadsheet or run a dividend discount model.

Certainly, a new era of higher stock valuations has paralleled this development. We have been well above the mean ever since.

This structural bid and mechanical reflexivity did not exist near to this extent in 2000. Nor in any prior historical period. Those periods were ruled, overwhelming, by the human psychology running mutual funds, hedge funds, and institutional separate accounts.

Another major factor pushing the biggest stocks higher reflexively are corporate buybacks, which are concentrated in the largest, most profitable companies. By propping up the market capitalization, buybacks assist passively (market cap-weighted) indexed capital flowing into the same shares.

### Corporate Stock Buybacks

In 2000, S&P 500 Index stock buybacks were about $140 billion annually - roughly the same as dividends.

In 2015, buybacks were $572 billion, relative to $382 billion in dividends.

By 2022, buybacks surpassed $1 trillion. 2025 broke that record.

Like everything else, buybacks are concentrated in the largest technology companies. The top 20 S&P 500 companies are about half of total buybacks.

It is worse than that. Apple alone is 11% of all buybacks. The top 6 were 30%.

As an aside, these were at high valuations too. Buying above intrinsic value per share dilutes intrinsic value per share.

Times are changing quickly, however.

Goldman reported for the 2Q 2025 that the Mag 7 showed zero percent growth in buybacks.

For the **4 <sup>th</sup> Q of 2025**, I calculate combined buybacks by Amazon, Alphabet, Microsoft, Meta and Oracle at just $12.6 billion total, **a 74% year over year decline.**

Pivotal Research projects for 2026 that Alphabet’s free cash flow will **fall 90%**. Morgan Stanley says Amazon will have **negative** **$17 billion** in free cash flow. Bank of America sees **negative $28 billion** at Amazon.

They are borrowing now – Oracle borrowed $25 billion, Meta borrowed $30 billion. Alphabet is lining up $15 billion or more. Borrowing to fund capital expenditures, not buybacks.

Buyback support for the market is set to fall hard.

### Consequence

I claimed publicly in 2017 and 2019 that this was going to end poorly. I saw passively invested assets take the crown from actively invested assets, and I considered this is to be the same as filing people into a crowded expandable room through a single small door. The room may grow to gargantuan size, but there is only one door to get out, and the rate of outflow is limited. A fire or panic one day could be something to see, as all move in one direction at the same time, I mused.

If these structural reasons for elevated multiples and valuation actually exist, and my logic for the consequences holds, then one would expect more recent market shocks to be something special in breadth and depth. **And they were.**

**COVID was in fact the most acute, broad equity selloff in history**. Think about that. There have been some doozies. Black Mondays and the rest. But COVID created the **fastest** bear market on record, correlation near unity (1) within equities, across three dozen countries simultaneously. Fortunately, flows to bonds/Treasuries and the dollar provided an outlet, a safety valve. Cross-asset contagion did not happen.

**Liberation Day was actually worse**. Liberation Day was the rare simultaneous crash in stocks, bonds, and currency. **U.S. equities, Treasuries, and the U.S. dollar all fell**. Oil fell 7%, Gold crashed. Liberation Day ranks fifth by percentage decline among major acute crashes — but first by real dollar value destruction, and it is not close.

Liberation Day was scarier for another reason – the weakness in the dollar meant that foreign investors in the U.S. got a double whammy. They lost in their stocks and bonds, and they lost additionally due to the falling currency. This is why a cross-asset breakdown is dangerous – it could trigger a **structural rotation** away from the United States.

Which would be truly catastrophic.

### The Shock Index

For more insight, I compared the loss in each major two-day crash to GDP per capita, and call that the Shock Index.

![](https://substackcdn.com/image/fetch/$s_!W7Wo!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F05cbb5f2-367c-4b8b-b1af-ec1e5cf5ab79_800x553.jpeg)

When adjusted for individual economic output, the “Liberation Day” crash of 2025 wiped out the equivalent of **~22%** of the average American’s annual income in two days, short of the **25.6%** in 1929, but a little more than the **20.6%** COVID crash and a lot more than the 1987 crash, which was effectively one day. Extending 1987 to two days, with the second being a recovery day, creates a shock index of 12.7%.

The COVID crash and the Liberation Day Crash are, by different measures, as bad, nearly as bad, or even worse than 1929, but the rest of 1929 did not happen. Plunge Protection Team, with some help.

The passive investment/401(k)-driven expansion to 60% or more that sustains increasingly above-mean valuations has also ensured that every 1% sold now carries the force of an entire prior-era crash. As the passive money simply sits there, **active money is dealing with a market that is getting less liquid as it gets larger.**

**As the table shows, stock ownership is so much more widespread than it was in 1929, yet the loss per stock-owning household was much greater during COVID, and much greater still for Liberation Day.** This validates my thesis on the danger in the market, and it has become **more** dangerous from COVID to Liberation Day.

I believe stock market crashes will continue this trend: becoming more severe, more correlated, and, ultimately, more consequential and of greater length.

### Where Were You When We Needed You

In 2026, the IMF studied this and came to its own conclusions, validating mine, but not making the connections I did in 2017 and 2019.

Looking back, our analysis shows that the turning point for correlations came around the end of 2019. With the onset of the pandemic the following year, the historical relationship changed significantly, resulting in sharp selloffs of both stocks and bonds to occur more frequently together.

The IMF also noted that bonds are less effective equity hedges.

![](https://substackcdn.com/image/fetch/$s_!XdAH!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F9940ccca-7c05-4745-83c2-d8a56380ca44_750x750.jpeg)

This is one consequence of interest rates being less important to stock valuation.

**When rates do not anchor valuation, traditional negative correlation between stocks and bonds breaks down. The IMF data shows that, but what is important is that this feeds cross-asset contagion.**

As I mentioned, without a discounted cash flow valuation as an anchor, stock price becomes a competition between narrative and expectations, a fundamentally subjective endeavor.

### High Frequency Trading & Pod Shops

The market becoming more **subjective** as the market also becomes more **skittish** is a problem. In the midst of a panic-driven information storm, an anchor to numbers is more reliable than an anchor to a narrative.

Since the GFC, high frequency trading (HFT) has grown rapidly, hitting critical mass just before COVID. **Today, 60% of trades today are HFT trades**.

These traders are typically market makers (~75%), and they have largely replaced the old analog specialist system. Specialists were obliged to sit there and buy when everyone sells or sell when everyone’s buying. That was the specialist’s job. Now, **HFT market makers may simply step away when there is a shock or high volatility event**. Therefore, it is more fragile than the specialist system and tends to mechanically amplify sudden selloffs and short squeezes.

About 15% of all trading is high frequency trading that is not market making.

The bigger issue is Pod Shops. These are multi-strategy funds, where typically a mathematical model allocates and withdraws from “pods” – independent teams managing money for the shop. Each pod works on very tight stop-loss rules because if it loses even 5%, it will get capital yanked back to the shop immediately. The vast majority of these pods are long momentum, short vol, and engaged in statistical arbitrage. Weighted correlations between pods are managed by the shop to be low, but during stress correlation converges to unity (1).

These are very hot alternative investments now, and have attracted nearly $1 trillion in assets, having grown ten times faster than hedge fund assets from 2017 to 2023. They exhibit positive returns and low volatility most of the time, as markets are rising most of the time.

So, passive investing sits there holding stock with diamond hands, **reducing** liquidity in stocks.

High-frequency trading provides **phantom** liquidity that turns to its gaseous form during stress.

Pod shops all hit stop losses on just about everything at the same time during market shocks.

Thank goodness, mostly, for those diamond hands.

As it is, the defined contributions that helped to create those inert diamond hands may be softening too.

### Defined Contributions, Revisited

The Baby Boomers were very important to the growth of 401(k) plans, but are now facing required minimum distributions (RMDs), which will be the first headwind the passive investing trend has **ever** had. This reverse process, too, will be **completely lacking in fundamental analysis or price discovery on the underlying stocks**.

RMDs will be withdrawn from a pool of IRA and 401(k) asset that exceeds $32 trillion currently. People over 70 own over 30% of all stocks, both directly and indirectly.

The RMD trigger age is 73. This is the legally mandated unwinding of retirement wealth in defined benefit plans. The percentage to be withdrawn grows with age.

The mean boomer is 71 years old today. The oldest are turning 80, and the youngest are 62. The maximum RMD withdrawal pressure should arrive from 2030 on.

**At $250 billion a year now and peaking over $1 trillion in the early-mid 2030s, redemptions will increasingly offset 401(k) contributions until 2028, when the redemptions first exceed contributions.**

![](https://substackcdn.com/image/fetch/$s_!XdFn!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F0c6109af-6b2b-4ff0-8110-ccdeda3f1b48_927x1155.jpeg)

**In 2028, for the first time in its over 3 decade existence, the defined contribution juggernaut driving much of the growth of passive investing turns negative**

### A Coiled Spring

The market is an amazing compounding machine, but there have been major, very long periods of poor returns after adjusting for inflation. Any one of them would have required good health, patience, and some luck to hold onto stocks through the entire bear.

Today, markets stand well above mean valuation for over 3 decades. This has never happened before.

The passive investing structural bid, the stock buyback structural bid, the demographic engine, the federal government’s plunge protection team.

These four forces have set up a market structure that has put valuations near all-time highs and recruited individual investors like no prior market ever has.

Newer market staples such as pod shops and high frequency trading are much more jittery than the active investors and specialists they replaced. In shocks, they help correlation go to unity, which is dangerous, especially if it crosses assets and international boundaries, putting the primacy of the U.S. as an investment destination in jeopardy.

The passive bid’s defined contributions engine is demographically challenged. In 2028, the defined contributions **bid** will become an **ask** for the first time in history.

The baby boomers who engineered a seemingly endless tide of contributions to index funds will see the tide go out in a few years. Slowly at first, then not slowly at all.

The buyback bid is already in retreat. This is not a pause, but a structural shift. The thesis behind the AI buildout has no end, and neither does the reason behind the data center buildout spending.

As Meta’s CFO Susan Li said on the earnings call February 6, 2026, “The highest priority is investing our resources to position ourselves as a leader in AI.”

It is not clear when the buyback bid may return in similar size. Perhaps not for decades.

Or, perhaps in a few years at much lower valuations.

There is a tension, like a coiled spring, in a market so unmoored from historic valuation measures, techniques, and levels.

That it became unmoored due to definable, predictable forces that now seem to be on the verge of reversing is one thing.

But for that to happen in a market that is structurally becoming more fragile, more prone to correlation and cross-asset contagion is another.

In a world, no less, full of line-of-sight geopolitical risk, not to mention a potentially reckless, budgetless endeavor to build a future, through AI data centers, that is ruining the purity of our best monopolies.

The catalyst may not be much of anything. The stock market might realize that $5 trillion of VC unicorns do not make lasting customers, or that the free cash flow is disappearing from our largest companies in a blink of an eye. The market might just roll over because it is time. Like in March of 2000.

My point is that the next one is likely to be even more violent than Liberation Day. And one day, the gloom might stick, backed by disappointed expectations and ruined narratives, yet still a long ways from anyone examining rate-sensitive discounted cash flows.

Long Live the Plunge Protection Team!

Until Next Time!