The year concentration paid and the diversifiers quietly won

A 2025 year-end review of a market that rose at the top of the page while its real work happened everywhere else.

For three years the market seemed to teach a single lesson: own American mega-cap technology, and treat everything else as a lapse to be forgiven. Every other position was a tax on performance, an apology to a benchmark that kept pulling away. 2025 did not overturn that lesson so much as expose its price. The S&P 500 finished the year up roughly 17.4 percent on a total-return basis, a third consecutive double-digit gain and, on its face, another vindication of the concentrated American bet. Look inside the print, however, and the story inverts. Two sectors, information technology and communication services, produced the bulk of the advance; strip them out and the index rose a pedestrian six percent or so. Of the Magnificent Seven, only Alphabet and Nvidia decisively beat the index. The green number at the top of the page was real, and it was hollow.

Meanwhile the assets that had been dead weight for a decade did the actual work. Gold rose about 65 percent, its best year since 1979; developed international equities and emerging markets both returned better than 30 percent for a dollar holder; and the dollar itself fell close to 10 percent, turning from threat into tailwind. The year opened in the shadow of American exceptionalism and closed as a study in broadening: out of the dollar, out of a handful of names, into hard assets and the rest of the world.

The throughline is uncomfortable and clarifying at once: the cost of home-country, single-factor concentration is invisible until the year it is not, and this was the year the bill came quietly, inside a rising market rather than a falling one.

Diversification stopped being an apology.

The debasement trade went mainstream

Gold led every major asset class in 2025. It closed near 4,368 dollars an ounce after an all-time high of 4,449 on 23 December, up roughly 65 percent on the year, its strongest calendar year since 1979 and nearly four times the S&P's total return. It was not alone: silver gained about 144 percent, its sharpest rise in decades; platinum rose roughly 127 percent to multi-year highs; and the gold miners, through the VanEck GDX, returned about 153 percent as bullion outran their costs.

What made the move notable was not its size but its respectability: a position that began the decade as a contrarian eccentricity became, by year-end, a consensus institutional trade with a name. Large and conservative houses, JPMorgan and Deutsche Bank among them, were grouping gold, silver and bitcoin together under the debasement label, and a record 43 percent of central banks surveyed by the World Gold Council said they intended to add to their own gold, up from 29 percent a year earlier. The stated rationale was fiscal dominance: a federal deficit near 5.9 percent of GDP, debt approaching 120 percent, net interest crossing a trillion dollars for the first time, and a Fed cutting into still-above-target inflation.

Two honest qualifications belong in the file. First, 2025's official-sector buying cooled from the record 2022-to-2024 run, roughly 634 tonnes through the third quarter against 724 a year earlier; the strongest-in-decades framing describes the structural shift more than this single year's tonnage. Second, the trade is now crowded and richly priced, and the very fact that it stopped being contrarian is what raises the odds of a sharp, sentiment-driven reversal.

The durable lesson is that a hedge against debasement belongs as a strategic allocation held before it is obvious, not a tactical bet chased after a 65 percent year; assets that cannot be printed re-rate when confidence in fiscal restraint erodes. The higher-beta legs, silver and the miners, carry operating and squeeze-like risk that amplifies losses as fast as gains.

Concentration set records and kept paying

In 2025 US equity concentration set fresh records and, for a third straight year, was rewarded for it. The top ten names in the S&P 500 swelled to about 40.7 percent of index weight by year-end, well beyond the roughly 27 percent intraday peak of the dot-com era. Nvidia became the first company in history to close above a five-trillion-dollar market capitalization, on 29 October, and ended the year near 7 percent of the entire index; Apple, second, sat near 6.5 percent, and Alphabet, third, near 6.1 percent, so the three largest names together approached a fifth of the market.

The concentration paid. The cap-weighted index returned about 17.4 percent while its equal-weighted twin managed only around 11, a six-point gap in a single year and, across 2023 to 2025, the widest three-year margin since 1971. Underweighting the leaders was therefore the most expensive active decision available: roughly 54 percent of active large-cap managers trailed the index in the first half, and the average large-blend fund held about 22 percent in the Magnificent Seven against a first-half benchmark weight near 30. To diversify away from the giants was itself an active bet, and it lost.

Yet even here the payoff was narrower than the weight. Of the seven, only Alphabet, up about 66 percent, and Nvidia, up about 39, clearly beat the index; Apple, Amazon, Meta and Microsoft matched or lagged it. Owning the few rewarded owning the right few, and the cohort's share of the index's return, roughly two-fifths, was lower than in the prior two years even as its weight hit a record.

The lesson we take is that a market-capitalization index is not a diversified holding; it is a momentum bet that grew more concentrated as it won. Record concentration is at once the reason the index won and the reason it is riskier to own than at any point since 2000; the discipline is to size that exposure with open eyes.

The everything rally and the cost of holding cash

For most of 2025 nearly everything rose, and often together. US equities returned about 17.4 percent, gold about 65, investment-grade bonds about 7.3 (their best year since 2020) and high-yield credit about 8.6. The dominant decision was simply presence: being invested at all mattered more than any refinement of timing or hedging.

Against that backdrop cash was the conspicuous laggard. The Fed held through the first half and then cut 75 basis points across September, October and December, pulling money-market yields toward the high threes, so cash returned only about 4 percent. That is a positive nominal number and a large opportunity cost: thirteen points behind equities, four and a half behind high yield, and roughly sixty behind gold. A heavy cash allocation was not a neutral, defensive stance; it was an active and expensive call on the market.

The honest exception was bitcoin, which led the rally for three quarters, then gave it back to finish modestly below where it began. Everything did not literally rise, and the same togetherness that flattered the year is a warning: when gold, equities and credit climb in unison, diversification is quietly offering less protection, and an everything rally can invert into an everything correction on the same correlations.

The durable point is that presence compounds while cleverness leaks, and cash is never truly costless: its 4 percent did its defensive job in April but over a full year of broad gains was the drag it usually is.

The pivot arrived, the long end refused

The long-awaited pivot came, and duration did not pay. After holding at 4.25 to 4.50 percent since December 2024, the Fed cut a quarter point at each of its final three meetings of 2025, landing at 3.50 to 3.75, with a December dot plot signaling only a shallow path beyond. The front end obliged: the two-year yield fell roughly 80 basis points to around 3.48. The long end did not. The ten-year declined only about 41 basis points to 4.16, and the thirty-year finished essentially unchanged near 4.82.

The long end refused for fiscal rather than cyclical reasons. The ten-year term premium topped 0.8 percent in January, its highest since 2011; Moody's stripped the United States of its last triple-A rating in May, the final agency to do so; and later that month the thirty-year pushed back above 5 percent, its highest since 2023 as a weak auction met deficit anxiety over the tax bill. Behind it all stood the same 1.8-trillion-dollar deficit and interest costs above a trillion dollars.

Bonds still made money. The Bloomberg US Aggregate returned about 7.3 percent, its best since 2020, but the return came from the belly of the curve and from spread tightening in corporates and mortgages, not from the long bond. The iShares 20-plus-year Treasury fund returned just 4.25 percent, roughly its coupon and almost none of the capital gain the pivot was supposed to deliver.

The lesson is regime-conditional: the reflex that Fed cuts rally duration is a relic of the quantitative-easing era. When deficits, heavy issuance and a rebuilt term premium anchor the long end, the long bond pays carry, not appreciation, and can sell off even as policy eases. Sizing duration now means watching auctions, issuance and the term premium as closely as the policy rate.

Bitcoin finished becoming an institutional asset

By the measures that define an institutional asset, 2025 was the year bitcoin completed its transition, even as its price finished lower. It reached an all-time high of 126,198 dollars on 6 October, briefly a 2.5-trillion-dollar asset, then slid roughly 30 percent to close near 87,500, down about 6 percent for the year. The price was a down year. The plumbing was not.

Underneath, the institutional story deepened in every dimension. The US spot ETFs matured into large asset-gathering vehicles: BlackRock's IBIT drew some 25 billion dollars of net inflows, sixth-most of any US ETF, while posting a negative return, the only fund in the top 25 by inflows to do so: structural demand, not performance-chasing. The count of public companies holding more than a thousand coins doubled to 49; ETFs and treasuries together absorbed far more supply than post-halving issuance created; and realized volatility fell to the lowest levels in the asset's history, with 30-day readings dropping from roughly 85 percent in early 2024 toward the high twenties.

The qualification matters as much as the claim. Institutionalization changed who owns bitcoin and how, not whether it can fall. The wrapper that channels inflows transmits outflows just as efficiently, as the late-year slide showed. The holder base, while broader, remains concentrated: a single treasury company held more than 670,000 coins, and three issuers held more than 85 percent of ETF assets.

The lesson we file is to judge an asset's maturation by its structure, separately from its price: adoption, access and a durable fall in volatility all advanced in a year of negative returns. A maturing market lowers the friction of owning an asset; it does not underwrite performance or abolish drawdowns, and lower volatility is emphatically not lower risk.

The dollar's quiet turn

The dollar's decline was the hinge of the year. The DXY opened 2025 above 109 and fell steadily to close near 98, a loss of roughly 9 to 10 percent and its worst calendar year since 2017. The first half alone was down about 10.8 percent, the steepest first-half fall since 1973 and the end of the Bretton Woods era. The move was orderly, not a crisis, but broad, and it repriced returns across the whole book.

That single repricing turbocharged everything held outside the United States. The euro rose about 13 percent from near 1.04 to above 1.17. For a dollar-based holder, MSCI EAFE returned roughly 32 percent and emerging markets about 34, the widest emerging-market outperformance in some seventeen years. Europe's Stoxx 600 reached record highs and beat the S&P in dollar terms, led by a 65 percent surge in bank shares. Gold, priced in a softening dollar, did the rest. Early on the drivers were tariff shocks and worries over Fed independence; later, a narrowing yield advantage as the Fed cut three times into year-end.

The caveat is that correlation is not causation, and a quiet turn is not yet a secular one. The gains abroad also owed something to cheaper valuations, the semiconductor cycle and strong European bank earnings, some of which would have helped even with a flat dollar; the index stabilized near 98 in the second half, and the trend could reverse.

Still, the durable lesson is that currency is a portfolio-level exposure, not a footnote. A single macro variable quietly drove returns across gold, international and emerging equities in one year, and for an unhedged dollar holder those assets captured both the local return and the currency gain, a double tailwind that would run just as hard in reverse. Dollar regimes tend to be gradual and persistent, which favors diversifying before the turn is obvious.

Private markets met their first real cracks

Private credit both scaled and cracked in 2025. It grew to roughly 1.7 trillion dollars and became the lead financier of the AI build-out; the Meta and Blue Owl Hyperion venture, announced 21 October, committed some 27 billion dollars to a single Louisiana data center, the largest financing of its kind. Beneath the growth, direct-lending spreads compressed to multi-year lows, around 544 basis points against 716 in early 2023, as record dry powder and evergreen-fund deployment pressure chased limited deal flow. The illiquidity premium was being bid away.

Then, in the autumn, the first genuine stress. Tricolor filed for bankruptcy on 10 September amid allegations of double-pledged collateral, costing JPMorgan a 170-million-dollar write-off; First Brands followed on 28 September with more than 10 billion in liabilities built on off-balance-sheet factoring. Jamie Dimon warned in mid-October that where there is one cockroach there are likely more, and the IMF's Global Financial Stability Report the same week flagged that more than 40 percent of private-credit borrowers had negative operating cash flow and that deteriorating quality had yet to show up in appraisal-based marks. Non-traded BDC redemptions pressed toward their quarterly caps; non-accruals roughly doubled at some large vehicles as payment-in-kind income masked the gap between reported yield and cash collected.

These were cracks, not a crisis: the autumn failures were idiosyncratic and fraud-driven, direct BDC exposure to First Brands was tiny, and most portfolios performed. But the episode revealed what the premium had been paying for.

The lesson is that the illiquidity premium compensates two things, illiquidity and opacity, and 2025 showed what happens when record capital compresses it at exactly the moment both features assert themselves. First cracks are a signal to check the foundations, not yet proof the building is unsound.

Breadth began, quietly, to matter

After a decade of the American mega-cap trade, 2025 was the year breadth reasserted itself, decisive abroad and only nascent at home. Non-US equities and a softer dollar closed a large slice of a fifteen-year gap in a single year, driven less by US weakness than by mean reversion: the S&P traded near 22 to 23 times forward earnings against roughly 15 for developed international and 13 for emerging markets.

Domestically the shift was only beginning. Concentration stayed extreme for most of the year: the top ten names contributed more than half of the index's return through October, and the cap-weighted index carried a valuation premium near 29 percent over its equal-weight sibling. Only at the very end did the first crack appear. By mid-December the equal-weight index's rolling one-month outperformance reached about 2.1 percent, a 90th-percentile move since 1990, and its relative-strength signal favored equal weight for the first time since the first quarter.

The nuance is essential and cuts against over-reading the turn. Much of the international outperformance for a dollar holder was currency, not earnings: in local terms the Stoxx 600 roughly matched the S&P's price gain, and the extra points came from the falling dollar. The equal-weight signal is one quarter old and statistically stretched; over three years equal weight still trailed cap weight by roughly 40 percent.

The lesson is that concentration is a loan against future returns, and the bill can arrive without a bear market. Breadth reasserted itself in 2025 not through a crash but through relative performance, and the earliest evidence of such a rotation looks easy to dismiss precisely because the long-term trend still appears intact. Rebalancing toward the neglected side of the ledger is cheapest before a regime is obvious.

Volatility was cheap until it clustered

Volatility in 2025 was low and cheap for long stretches, then clustered violently around a handful of policy events. The VIX spent January to mid-February in a calm 14-to-22 range as the S&P set records; from May to year-end it never closed above 29, finishing in the mid-teens. The exception was singular and severe. After the Liberation Day tariffs of 2 April, the VIX ran from under 17 to above 60 in eight sessions, reaching 60 intraday on 7 April; the S&P fell about 10 percent over two days, and one-month realized volatility hit roughly 43 percent, the most since 2020.

What made the cluster dangerous was a correlation flip. US stocks, Treasuries and the dollar fell together, so the conventional bond hedge failed exactly when it was needed; the MOVE index of rate volatility spiked toward 172. The reversal came as fast as the shock: on 9 April a 90-day tariff pause lifted the S&P 9.5 percent, one of its best sessions on record. Insurance, in other words, was at its cheapest in late March, days before it became indispensable, and by 7 April it cost several times its pre-shock price. The hedge that actually worked across the whole year was not duration but gold.

The lesson is that volatility clusters and mean-reverts: it is cheapest when complacency is highest, it arrives far faster than it departs, and it drags cross-asset correlations up with it, so the standard stock-and-bond hedge can fail in the very tail it is meant to cover. Convexity is bought cheaply in quiet regimes, not after the spike; underwriting tail risk simply because realized volatility has been low for months is how the cheapest insurance comes to be sold at exactly the wrong moment.

Temperament beat cleverness

If the year had a single verdict, it was that temperament beat cleverness. Almost the entire equity gain of 2025 was compressed into a violent V around an unforecastable policy sequence. The S&P set a record on 19 February, fell about 19 percent to its 8 April low on the tariff shock, then recovered all of it by late June and closed near 6,932 on 24 December. The best single day, the 9.5 percent surge on 9 April, arrived the session after the low. An investor who de-risked into the panic, over-hedged, or waited for clarity locked in losses and forfeited a roughly 20 percent rebound that landed in weeks.

Meanwhile a plain, diversified, fully invested portfolio compounded across several engines: US equities recovered to records, international markets and gold delivered outsized gains, and the Aggregate added a mid-single-digit return as the Fed eased. The one prominent momentum trade that punished its late chasers was bitcoin, which ended roughly 30 percent below its October high. Chasing the hottest single asset late was the losing move; durable, diversified exposure was not.

We hold this lesson with humility. The 2025 drawdown rebounded within weeks; 2000 and 2008 did not, and staying invested is a long-horizon principle, not a promise that every dip snaps back. The 9 April pause was a discretionary political decision that could have gone the other way; standing on 8 April, the recovery was neither knowable nor assured. Presence beat prediction this year only because the resolution was favorable, and a family office that mistakes a good outcome for a small risk has learned the wrong thing. The durable edge is time in the market, breadth of exposure, and the discipline to sit still through a frightening drawdown, sized so that sitting still is actually possible.

What we carry into the new year

Ten conclusions, one throughline. Gold and the hard assets re-rated on fiscal dominance; concentration set records and paid, while quietly becoming the largest single risk in every diversified book; an everything rally rewarded presence and made cash expensive; the pivot arrived but the long end refused, punishing reflexive duration; bitcoin finished becoming an institutional asset without becoming a rising one; the dollar turned and repriced the world; private credit met its first real cracks; breadth began, unevenly, to matter; volatility proved cheap until it clustered; and temperament beat cleverness. Read together they say one thing. For three years the market taught that diversification was a tax on conviction. In 2025 the tax reversed: the concentrated American bet merely kept pace while the diversifiers a family office is meant to hold, gold, international equities, real assets, did the heavy lifting, and they did it in a year whose headline print still looked benign. The cost of concentration was invisible until it was not, and it came due inside a green number.

What we carry forward is not a forecast, which the year taught us to distrust, but a posture. It holds the hedges before they are consensus and sizes them to survive their own drawdowns. It treats a market-capitalization index as the momentum bet it has become, reads yield net of the liquidity and credit risk it compensates, and decides currency exposure on purpose. Above all it keeps enough temperament in reserve that the next 8 April finds the portfolio able to do nothing, deliberately. The diversifiers earned their keep this year. The harder discipline is to keep holding them into a year when, once again, they may look like an apology.

Fanaura is the private family office of Maxim Levoshin, investing principal capital worldwide. This note reflects the office's own view and is not investment advice.