Q4 Outlook for Traders: Three tests for the wall of worry.

Chart: Global 10-year government bond yields, 2000-2026. US Treasury, UK Gilt, German Bund, French OAT and Japanese JGB yields. For the entire period through the Covid pandemic of 2020-21, ever-lower sovereign yields in times of market turmoil always provided a powerful tailwind for financial-asset valuations. Q4 begins with the opposite question: how high can the global cost of capital rise before it forces meaningful adjustments in growth, fiscal policy and asset prices? Note the Japanese yield conundrum is no more as yields have now risen to levels nearly on par with Germany.

This isn't the post-GFC policy regime

The policy response to any instability is likely to differ from the post-GFC playbook. Back then, weak growth and financial stress were generally disinflationary, allowing central banks to cut rates, launch Quantitative Easing (QE) and push investors out along the risk curve while fiscal policy remained comparatively restrained.

Today, governments face structurally large borrowing needs driven by defence, energy security, industrial policy and ageing populations, while inflation limits how aggressively central banks can suppress sovereign yields. Simply launching another enormous QE programme into an inflationary supply shock could create more problems than it solves.

If sovereign debt markets eventually suffer a bout of indigestion because markets are unable to absorb the growing supply, policy is unlikely in the first instance to pivot to fiscal austerity or to QE, but instead towards determining who is encouraged or required to hold government debt. We are not there yet, and ordinary Treasury buybacks or slower central-bank balance-sheet runoff are not financial repression. But over time we may see more favourable treatment of sovereign debt on bank balance sheets, incentives for domestic pension funds to hold more government bonds or other ways of mobilising domestic savings. Japan bears watching closely, including any eventual shift in the domestic-asset allocation of its enormous GPIF pension fund, which market participants have inferred may be forthcoming following an unusually timed meeting in August with the fund’s asset allocation on the agenda less than six months after its March decision to not change domestic bond allocations.

In short, the next response to an unstable sovereign-bond market may look less like QE for everybody and more like increasingly directed demand for sovereign debt.

US equities: when reflexivity goes the other way

The US economy has another unusual vulnerability: the stock market itself. The wealthiest households account for a disproportionate share of consumption, making activity unusually sensitive to the wealth effect from equities. Rising portfolios support confidence and spending, which in turn support profits and markets. The wealth effect aside, the vast capital expenditures of the AI data centre buildout also flatter the pace of S&P 500 earnings growth as they are quickly recognized on income statements while the negative cash flows are obscured in various debt instruments and externalized into off balance sheet special purpose vehicles (SPVs) and third parties.

But the reflexivity works in both directions. A sufficiently large equity correction could weaken spending and earnings and reinforce the initial decline, with elevated margin debt adding another potential source of forced deleveraging. At the same time, the most aggressively leveraged balance sheets built on assumptions of ever increasing spending could find themselves in a liquidity crunch.

None of this tells us when or if a correction is looming. Markets have repeatedly humbled those who assume that stretched valuations and leverage or an obvious macroeconomic risk must immediately translate into lower prices. There is also a stabilising mechanism: if equities fall far enough, weaker expected growth might eventually pull longer-term yields somewhat lower and help rebuild support for risk assets or even trigger another surge in risk appetite. At the same time, most major economies, especially the USA, continue to run large deficits despite this continued expansion, and any economic downturn would likely lead markets to anticipate renewed fiscal stimulus and even larger deficits. So any rapid return to significantly lower longer yields looks like a stretch.

AI: from a race to build intelligence to a race to deploy it

Anthropic CEO Dario Amodei's call to "pace the frontier" of AI development deserves to be taken seriously on its stated terms. Anthropic has long emphasised AI safety, and increasingly autonomous agents create genuine concerns around alignment and cybersecurity. But the market significance goes beyond safety. For the first time, the safety case for slowing the frontier may also increasingly align with the economics of doing so.

Today's frontier models are already remarkably capable and are generating enormous revenue growth. The problem is what happens at the margin. Each new generation requires extraordinary spending to push capabilities another step forward, while competitors can close much of the gap surprisingly quickly. That creates the risk of a Red Queen race in which every frontier lab spends heavily simply to avoid losing its relative position.

China makes this dynamic even more important. Chinese developers have operated under tighter capital and chip constraints and have responded by focusing aggressively on efficiency and open-weight models. They also make it difficult for US labs simply to stop advancing, which is why safety proposals increasingly overlap with chip controls and other measures intended to preserve a Western technological lead.

There is also a more basic question: how much work actually requires the most expensive model available? Model routers can send routine tasks to much cheaper small or open-weight models and reserve frontier intelligence for the hardest problems. For large companies, meanwhile, the valuable asset is often their own data, intellectual property and business processes. Much enterprise AI may therefore involve inference over proprietary data using locally hosted or tightly controlled models, retrieval systems and specialist applications, with frontier models being used only when genuinely needed and with strict guarantees that corporate data is not used to train the next public model.

That could mean more AI rather than less AI, while changing who captures the economics. A shift from frontier training toward deployment should favour inference-optimised and custom silicon, networking, storage, private-cloud and on-premise infrastructure, cybersecurity and data-management software. It does not mean demand for top-end GPUs or HBM disappears: large reasoning models can make inference extremely compute- and memory-intensive. But a slower frontier race would challenge the assumption that the most training-intensive parts of the hardware stack must keep accelerating at today's rate.

Regulation adds another twist. Anthropic's safety argument could look like an attempt to raise barriers to entry for competition as it eyes an IPO. While the company has explicitly rejected a blanket ban on open-weight models, the imposition of common safety standards, external evaluations and compliance infrastructure would require fixed costs that only the largest labs can readily absorb. Regulation can therefore address genuine externalities while also erecting barriers to entry and entrenching incumbents. Whether intended or not, that is where the safety argument begins to overlap with the risk of regulatory capture.

For markets, the second derivative of growth is what matters for the outlook for AI-related names. For a further significant market adjustment ahead, AI demand and hyperscaler capex wouldn’t necessarily have to fall. Rather, if spending that was expected to keep accelerating instead merely grows more slowly, the adjustment could be violent in hardware names whose earnings expectations already reflect extraordinary growth.

The cost of capital is an additional speed limiter. The AI buildout is itself becoming an enormous borrower as governments continue issuing extraordinary amounts of debt. We would not claim that hyperscaler issuance is causing the sovereign-bond selloff, but at the margin the AI boom is competing for the capital required to sustain itself. Higher yields therefore matter twice: they compress the valuations of long-duration growth assets while raising the cost of financing the infrastructure on which those growth expectations depend.

What would change the picture?

There is a perfectly plausible bullish outcome for Q4. A Middle East breakthrough could send oil and gas sharply lower, easing the inflation outlook and allowing long yields to retreat, while hyperscalers could reaffirm spending plans and show stronger evidence that AI investment is generating revenues and productivity. That combination would remove all three immediate constraints and could see markets climb the wall of worry yet again.

The more negative outcome does not require disaster in any single area. It is the three pressures reinforcing one another: energy staying expensive enough to damage real growth, long yields refusing to decline much even as risk sentiment weakens (or worse still rising further) and concrete evidence emerging that AI capex growth is being revised lower. That would hit household purchasing power, equity valuations and one of the most important investment engines of the US economy at the same time.

FX: capital flows matter as much as yield spreads

For the US dollar, the immediate response to a sharp risk-off episode can still be positive. But there is an important qualification: foreign investors have lately directed unusually large amounts of capital into US equities relative to Treasuries (See chart below). A sustained US equity drawdown could therefore remove one of the dollar's key sources of structural inflows after any initial scramble for dollar liquidity.

Moreover, the world’s largest net saver, Japan, may look to encourage more of its savings to remain at home and even return home should pressure on Japan’s government bonds and currency increase. We remain biased, therefore, toward an eventually stronger Japanese yen, although the surge in global yields complicates the timing. Japanese policymakers appear more interested in a stable and moderately firmer JPY than in engineering a dramatic reset. Any much larger yen rally than the base case for a solid move below 150 and even toward 140 in the USDJPY exchange rate may require the kind of global deleveraging event that forces more aggressive repatriation.

For the euro, France is increasingly the point to watch. Another significant widening of French spreads would test Europe's willingness to support sovereign debt markets, given the assumption that France is too big to fail. Sterling faces a related test around the Burnham government's budget, with high gilt yields on the UK’s twin deficits sharply limiting fiscal room for manoeuvre. Will Burnham dare to seriously challenge the bond vigilantes?

Bottom line for the equity market outlook.

Our Q3 outlook argued that the growth mathematics behind the AI hardware trade were becoming increasingly difficult. So far, the market has again proved more resilient than those concerns suggested, and that is worth remembering, even if we have seen a major correction in many AI hardware names after the parabolic advance of late Q2.

Chart: The chart above is of foreign purchases of US equities and US treasuries since 2000, expressed as % of GDP on a trailing twelve-month basis. The chart above shows that the inflow into US equity markets over the last year has risen to levels far greater than during the previous most significant episode during the tech and telecom bubble that peaked in 2000. It would be interesting to see how the US dollar behaves in a weak equity market if foreign inflows into equities reverse, especially if US treasuries are no longer seen as a safe haven asset.

Q4 nevertheless brings an unusually interesting collision. Energy is testing consumers and margins, sovereign yields are testing valuations and public finances, while the AI industry itself is beginning to ask whether continually accelerating the race at the frontier makes economic or security sense. None of these has to break the market. Together, however, they may tell us how solid the foundations underneath the wall of worry really are. Nor should we expect austerity from any major government to address any bond market instability – nominal economies must continue to power ahead to run out from under the shadow of staggering debt loads.

Q4 2026 commodities outlook: Scarcity continues to support hard assets

Commodity markets enter the final quarter of 2026 with physical scarcity, geopolitical uncertainty and mounting fiscal concerns continuing to support hard assets. While the Middle East energy shock has dominated recent months, the broader theme remains one of demand repeatedly running into supply constraints, whether in energy, mined metals or selected agricultural commodities.

Precious metals remain a key expression of this environment. Gold continues to face traditional headwinds from elevated bond yields and rising funding costs, but its resilience highlights an important shift in the composition of demand. Interest-rate-sensitive investors may hesitate at current yield levels, but central banks and other strategic buyers are less focused on the opportunity cost of holding a non-interest-bearing asset.

At the same time, rising government bond yields are themselves becoming part of gold's longer-term support story. Higher yields increase the cost of servicing already elevated sovereign debt, adding to concerns about fiscal sustainability, currency debasement and ultimately the credibility of fiat assets. Against this backdrop, demand for gold as a scarce asset that sits outside the financial system is unlikely to disappear. Silver and platinum may benefit from the same investment theme while retaining additional exposure to industrial demand drivers.

Energy faces a potentially dramatic transition during Q4. We see limited prospects for a durable political solution that materially increases flows of crude oil, refined products and LNG from the Gulf before the US midterm elections in November. Until then, tight physical markets and exceptionally strong backwardation are likely to keep crude and, especially, fuel products supported and volatile.

Thereafter, the picture could change rapidly. A political breakthrough and the reopening of disrupted supply routes could unleash what might resemble a mini tsunami of pent-up crude, fuel and gas supplies, triggering a potentially sharp correction in prices and flattening the currently elevated backwardation (in which the prompt price is far higher than the forward price).

However, we do not expect energy prices simply to return to pre-war levels, with several opposing forces likely to shape the eventual equilibrium. Months of extreme prices have already triggered demand destruction, some of which may prove permanent, while also accelerating efficiency gains, substitution and electrification. Offsetting this, the need to rebuild depleted strategic and commercial stockpiles could create a sizeable source of demand once supplies normalise. The eventual post-war price level will therefore depend on the balance between how much lost demand returns and the scale and speed of inventory rebuilding.

Industrial metals remain supported by a more straightforward scarcity story. Copper sits firmly in the crosshairs of accelerating demand from electrification, grid expansion, renewable energy and data centres, while miners continue to struggle to deliver sufficient new supply. Declining ore grades, higher costs, permitting delays and long development timelines leave the supply response relatively inelastic, increasing the risk of renewed upside pressure on prices whenever inventories tighten.

Finally, agriculture enters Q4 with weather and geopolitics taking centre stage. A rising El Niño threat will affect crops differently across regions, benefiting some while reducing yields and disrupting supply elsewhere. At the same time, the war in Ukraine continues to create uncertainty around Black Sea production and exports.

Rather than expecting a broad-based agricultural rally, we see opportunities increasingly driven by individual supply stories. Sugar, cotton, rice and wheat are among the markets we will be watching particularly closely as weather patterns, energy costs and geopolitical disruption evolve.

The common thread across commodities remains scarcity. Where supply cannot respond quickly, volatility stays elevated, backwardation can enhance investor returns, and hard assets retain their attraction in an increasingly uncertain financial and geopolitical environment.