
A new data product launched recently by CNBC and wealth technology platform Addepar offers something genuinely useful: a quarterly window into how family offices (some of the most sophisticated pools of private capital in the world) are actually allocating their money.
The inaugural reading is instructive, and not entirely for the reasons its authors intended.
According to the tracker, equities now account for 34% of family office portfolios, up from 32% a year ago. This makes equities the largest and fastest-growing asset class in the cohort. Public stocks, the report notes, were one of the only categories to grow as a share of portfolios over the past year. Eighty per cent of those equity holdings sit in domestic US stocks.
That is a lot of conviction in a single asset class. And when an asset class commands that kind of weight in the portfolios of the most well-resourced investors in the world, it is worth pausing to ask a simple question: what does the price of that asset actually reflect?
What the Numbers Say
The Shiller CAPE ratio is the cyclically adjusted price-to-earnings ratio that smooths earnings across a full business cycle to filter out short-term distortions.
- It currently sits at approximately 40.7.
- The long-run median is 16.
- The long-run average is closer to 17.
To put 40.7 in context: this is only the second time in the 155-year recorded history of the metric that US equities have traded above the 40 threshold. The first was the peak of the dot-com bubble in late 1999 and early 2000. History does not repeat, but the company the current reading keeps is worth noting.
The Buffett Indicator is the total US stock market capitalisation expressed as a percentage of GDP, which Warren Buffett himself described as "probably the best single measure of where valuations stand at any given moment".
- It currently sits at approximately 220-240%, depending on the precise measure used.
- That places it approximately two standard deviations above its long-run trend, firmly in territory that Buffett described in 2001 as "playing with fire."
Based on current levels, some models project the US equity market to deliver negative real returns over the next eight years. These are not fringe indicators. They are among the most widely respected tools in long-run valuation analysis. And right now, they are aligned in pointing in the same direction.
The "This Time is Different" Argument Deserves a Hearing
The most intelligent counter-argument to any valuation-based concern is the structural one: that the composition of the market has changed so fundamentally that historical averages are no longer the right benchmark.
It goes something like this. The S&P 500 today is dominated by a handful of companies including Microsoft, Nvidia, Apple, Alphabet and Meta that are not merely large but potentially transformative. Artificial intelligence, the argument runs, represents a genuine productivity step-change: the kind that compresses costs, expands margins, and accelerates earnings growth across the economy in ways that prior cycles simply did not.
If AI delivers on even a fraction of its projected economic impact, the earnings denominator in every valuation ratio is going to look very different in five years. On that view, a CAPE of 40 may not be irrational but simply forward-looking in a way that backward-averaging cannot capture.
This argument is not frivolous. There are serious economists and investors who make it carefully, and it deserves to be engaged rather than dismissed. But there are three problems with anchoring a 34% equity allocation to it:
- The Track Record of Structural Arguments: The first is that "this time is different" has a reliable track record of being wrong at precisely the moments it feels most convincing. In 1999, the internet was also real, also transformative, and also going to change everything. It did. It just didn't justify the valuations attached to it at the time. Structural change and stretched valuations are not mutually exclusive, and the latter has historically resolved itself regardless of the former.
- Timing and Distribution: The second problem is one of timing and distribution. Even if AI does deliver a step-change in corporate earnings, the market has to some extent already priced that expectation in. The question is whether current equity prices already reflect the optimistic scenario, leaving little margin for error if delivery is slower, more uneven, or more competitively distributed than the consensus assumes.
- Concentration Risk: The third is concentration risk. The AI thesis is not a broad equity thesis, but a thesis about a small number of companies. A portfolio with 34% in public equities, 80% of that in US stocks, is substantially a bet on a narrow slice of the index that carries the AI premium. That is a very specific wager dressed up as asset class diversification.
The honest position is this: AI may well be structurally significant enough to justify higher-than-historical valuations.
But "higher than historical" is not the same as "as high as they currently are," and it is not a framework that removes the need to ask what happens if the thesis takes longer, costs more, or distributes differently than expected.
The Problem with Consensus
There is an important distinction between a decision that looks correct and a decision that is correct. When the world's most sophisticated investors are adding to equities at exactly the moment those equities are trading at historically extreme valuations, the two things can diverge significantly.
This is not a criticism of family offices. The past several years have rewarded equity concentration generously, and the behavioural pull of recent performance is well-documented. Professor Ulrike Malmendier's research on what she calls the "experience effect" demonstrates that investors systematically overweight the market conditions they have personally lived through.
A decade of strong equity returns leaves a mark. It shapes expectations, calibrates assumptions, and makes elevated allocations to equities feel entirely reasonable. Until it doesn't.
The CNBC/Addepar data shows equities growing as a share of family office portfolios at the same time that every major long-run valuation metric is flashing caution. That is not a coincidence. It is precisely what the experience effect predicts.
What Diversification is Actually Supposed to Do
The case for trend following as a complement to equity-heavy portfolios is rarely more relevant than it is in a high-valuation environment. Trend following does not require a view on whether equities will correct, or when. What it does is participate in sustained price moves - in any direction, across any asset class - without requiring the market to go up.
In environments where equities are priced for near-perfection and deliver something less than that, trend following has historically provided what researchers call crisis alpha: genuine uncorrelated returns precisely when conventional portfolios need them most.
There is a further point worth making, and it is one that often surprises people. Trend following does not require equities to fall in order to perform. If the AI thesis is correct and US equities continue their ascent, a trend-following strategy will participate in that move. It is, by design, long whatever is going up.
The irony is that trend following can trade a bubble just as effectively as it can trade a correction. It does not need to predict which one is coming. What it needs is for prices to move in a sustained direction (and markets at extreme valuations tend to be nothing if not directional) on the way up and, eventually, on the way down. The exit from a trend is governed by the same rules as the entry: no heroics, no forecasting, no requirement to be right about the macro.
The ECCM Systematic Trend Fund trades across more than 90 global futures markets - commodities, fixed income, currencies, equity indices - allowing it to capture trends wherever they emerge. In Q1 2026, that meant energy markets, where sustained directional moves provided meaningful returns while equity-heavy portfolios struggled.
The point is not that this will always happen. The point is that the opportunity set is genuinely broader than a portfolio concentrated in US public equities. The efficient frontier - the concept that the best risk-adjusted portfolio is not necessarily the highest-returning one, but the one that combines assets most efficiently - is often invoked in theory and ignored in practice. The Addepar data suggests that family offices, for all their sophistication, are no exception.
A Note on Expectations
None of this is a prediction. Elevated valuations can persist for longer than any rational model suggests they should. The CAPE ratio was above 30 for years before the dot-com correction. Markets can stay expensive.
But there is a difference between accepting that markets can remain expensive and deciding that 34% of a portfolio in expensive equities (and concentrated 80% in a single country) is the appropriate response to the current opportunity set.
At current valuations, US equities are being asked to do a lot of heavy lifting in portfolios that may have limited room to absorb disappointment. The question sophisticated investors should be asking is not whether equities belong in a portfolio. They do.
The question is what work those equities are being expected to do, and what happens to the portfolio if that work doesn't get done. Genuine diversification - across return types, not just asset labels - exists precisely for that scenario.
Wholesale clients can find more information on ECCM and the ECCM Systematic Trend Fund at Australian Fund Monitors and ECCM's website.