Last December I rebalanced around five theses for 2026. Five months in, the rotation out of US large-cap and into everything cheaper has mostly paid, with one exception. As always, this is a personal-portfolio review and in no way a recommendation.
Portfolio performance so far
Through May 29, the portfolio is up +5.2% in CHF on a time-weighted basis. A fairly-constructed 60/40 benchmark (60% MSCI ACWI in CHF, 40% global aggregate bonds CHF-hedged) is up +5.4%. The S&P 500 in CHF, total return, is up +9.0%. So the portfolio roughly matched the diversified benchmark and trailed pure US exposure by almost four points.
I was up modestly through January, gave it all back and then some by late March, bottomed near -4%, and climbed steadily after. A Middle East conflict in late February and a brief threat to the Strait of Hormuz sent oil sharply higher, and a year that was supposed to be about disinflation turned, for a few weeks, into one about sticky prices. I guess that single event reset the macro backdrop for everything that followed.
The 2025 outperformance, when this same approach beat the S&P 500 by a wide margin in CHF terms, was driven by the dollar’s collapse against the CHF. USDCHF fell about 11.5% last year. So far in 2026 it has fallen about 1.2%. The FX tailwind that made the strategy look brilliant is gone, and the portfolio is doing what it should in a year when the diversifier isn’t paying much: keeping pace with a balanced benchmark and trailing pure US.
What worked, what didn’t
Emerging Markets returned about +20.7% in CHF, the single biggest contributor relative to weight. US Small Cap +13.2%, Japan +13.1%. The rotation thesis, that the valuation gap between US large-cap and almost everything else was too wide to ignore, paid in three of the four sleeves where I added exposure.
The part I only half-anticipated is what drove the EM win. It wasn’t Chinese stimulus, which is what I leaned on in December. It was the Korean and Taiwanese chipmakers at the heart of the AI supply chain. My broad EM holding caught that; my dedicated Chinese-tech slice did not. I was right to own EM and right about Asia, just for the opposite reason I wrote down - fair enough.
The fourth rotation was Europe, and it didn’t deliver. Up +4.8% in CHF, it trailed US large-cap (+9.9%) by five points and trailed every other rotation by eight to sixteen. Germany’s fiscal pivot is real (€500bn infrastructure fund, €400bn defense, €600bn in private commitments). Citi’s European equity strategy team kept its overweight call and projects German EPS growth at a 13% CAGR through 2029. None of that has shown up in prices yet. A cheap market can stay cheap until a catalyst forces the re-rating, and the catalyst I was counting on hasn’t done its job (so far).
Bitcoin and Listed PE were the worst calls, down about 19% and 13% in CHF. Crypto I’ll come back to, because what it did in March changed how I size it. Listed PE I’m watching more carefully, since the discount-to-NAV widening that drove the drawdown isn’t a thesis call gone wrong so much as the trade getting more crowded into a year of higher dispersion.
Valuations got more extreme, not less
The CAPE was the spine of the December argument. The S&P 500 traded at 40.5x cyclically-adjusted earnings, more than twice the long-run mean of 17.3, and the gap had to close one way or another, through multiple contraction or earnings growth.
It closed in neither direction. The CAPE is now about 42.7, up from 39.8 in December. That puts US large-cap closer to the dot-com peak (44.2 in December 1999) than at any point since. The compression I expected didn’t happen; the multiple expanded while earnings kept up, and the rest of the world stayed cheaper.
Concentration is messier. NVIDIA, grew from 7.2% to 8.17%. Apple sits at 6.70%, Microsoft at 4.96%, down from 5.9%. The concentration risk has become a single-name risk, not a Magnificent-Seven risk.
The Europe trade is alive but not paying
The European valuation discount was 22% on a forward-earnings basis (US 23x, Europe 14x, per UBS’s Year Ahead 2026). On a trailing basis today, US trades around 28x and Europe around 18x, a discount near 35%. The gap is now wider: the cheap market got cheaper relative to the expensive one, and being early on a value gap that won’t close is indistinguishable from being wrong for as long as it lasts.
Two quarters of strong US earnings (Q1 2026 net margin hit 13.4%, a record per FactSet) made the multiple expensive without making the index look fragile. JPM’s Lakos-Bujas raised the S&P 500 year-end target to 7,600 in April, lifting 2026 EPS to $330 (+22% YoY). Morgan Stanley’s Mike Wilson is at 7,800 with explicit “early cycle” framing. The bear is BofA’s Subramanian at 7,100. Consensus has converged at the higher end of the range. For the Europe overweight, the lesson is that a wide discount can persist for years. Cheap-versus-expensive was never going to be enough on its own. It needed a catalyst, and Germany’s fiscal pivot, hasn’t yet moved the multiple.
Dollar, AI, bonds
Dollar. December’s call was for 4% to 10% USD depreciation through 2026. So far it’s down about 1.2% against the franc, a stall rather than a fail. Goldman, UBS, Pictet, ABN AMRO and MUFG all kept their dollar-weakness calls but pushed the bulk of the move into H2. The fundamentals (twin deficits, a narrowing rate differential, fiscal-sustainability concerns flagged by the IMF in April) haven’t gone anywhere. What changed is timing: the war revived safe-haven demand for the dollar and a hawkish Fed propped up the carry. I assume the thesis is delayed, not dead.
AI capex. Hyperscaler capex was forecast at $571bn for 2026 in December. Q1 actuals plus guidance now point to roughly $660-725bn at the conservative end, with one sell-side estimate as high as $805bn. No major house trimmed. Cloud revenue is accelerating alongside the spend. My own view, that AI may end up a competitive commodity where the value flows to the companies using it rather than a winner-take-all windfall for the model providers, is unchanged. But the capex cycle has clearly not topped, and betting against it on valuation alone would have cost me.
Fixed income. This is the clean miss. I built the sleeve around two to three Fed cuts in 2026, and there have been none. Inflation reaccelerated toward 4% on the energy shock, the Fed held through every meeting, and by spring the market expected the ECB, not the Fed, to move next, and to move by hiking. CHF Corporate Bonds returned about +0.2% in CHF, Euro Govt CHF-Hedged about -0.1%, US Treasuries about -1.3% in CHF, the FX drag wiping out a roughly flat dollar return. Carry has been fine. The capital-gains case I was implicitly making, that falling policy rates would lift bond prices, simply hasn’t arrived.
The two diversifiers split
The clearest lesson of the first five months is that my two “diversifiers” are not the same animal, and the March shock was the test that separated them.
Gold behaved like a diversifier should when it mattered, holding up through the energy-shock drawdown while equities fell, supported by continued central-bank buying (244 tonnes in Q1, per the World Gold Council). Over the full five months the CHF-hedged sleeve is up only about +1%, because the hedge gave back what unhedged gold earned as the dollar held firm. That’s a cost I knowingly took. But the metal did its job in the only window that tested it, and it remains the book’s one genuine diversifier.
Bitcoin did not. In the March drawdown it fell with stocks, harder than stocks, and it’s down about 19% in CHF on the year. It behaves like a leveraged bet on risk appetite and liquidity, not like ballast.
Three changes
Three changes, all small, none of them a reversal of the strategy. A midyear review mostly earns the right to leave things alone, and that’s most of what this one did.
Gold from 5% to 6%. It’s the one genuine diversifier in the book, it held up in the only stress window of the year, and I funded the increase with new cash rather than selling anything. The hedge cost me the headline gains this year, and I’m keeping it anyway: I own gold to dampen equity drawdowns in francs, not to make a dollar bet on the side.
Crypto from 4.5% to 3.5%. Not a loss of faith, I keep a meaningful position for the asymmetric upside. But its job in the portfolio was diversification, and it failed that job in March. I’m sizing it as what it actually is, and I did it by letting it drift down and redirecting new contributions, not by selling into weakness, so it costs nothing to implement.
Japan, switched into a CHF-hedged share class. Japanese equities were among the best markets in local terms, up well into the teens, but a falling yen handed most of that back to me as a franc investor. The fix is to keep the equity exposure and hedge the currency, which at current rates costs almost nothing: the SNB sits at 0%, Japanese rates are also low, so the differential I pay to hedge is a rounding error against the spot move it removes. Same 3% weight, different currency. I did not extend the same logic to the dollar, because hedging dollars at a roughly 4% rate differential isn’t worth it to neutralize a one-point spot move on a 2% Treasury position. There I’d rather keep the carry.
When I measured the actual AI exposure hiding inside my supposedly de-concentrated book, the Nvidias and TSMCs embedded in cap-weighted index funds, it came to roughly 13% of the portfolio, with no single name above about 2%. The de-concentration worked. I don’t have a concealed AI bet masquerading as diversification, which was the thing I most wanted to rule out.
What would change my mind
The thesis rests on a few load-bearing assumptions, and the point of a review is to name what would break each one.
On AI and US concentration, the trigger isn’t price, because expensive can stay expensive. It’s the hyperscalers cutting capital spending while cloud revenue stalls, or an efficiency shock that makes today’s infrastructure look overbuilt. So far capex is rising and getting paid for. If that flips, I cut the AI look-through properly.
On rates, if inflation rolls back over and the Fed finally starts cutting, the duration case I got wrong this spring comes back and I’d extend. If inflation stays sticky and the energy shock turns into a wage-price problem, gold does the work and the bonds stay short.
On Europe, the overweight stays as long as the fiscal story is intact, because the cause of the underperformance is a missing catalyst, not a broken thesis. If German spending stalls or the earnings upgrades start getting cut, that’s different, and I’d close the gap to neutral.