As I suggested after Wednesday’s bond bloodbath, this market likely needed 24 to 48 hours in the recovery room before traders decided whether the surge in yields was a cardiac event or simply the price of doing business in an economy that refuses to cool. By Friday, roughly 36 hours after the mini VaR shock, the diagnosis looked considerably less terminal. The casino doors swung back open, the oxygen masks came off, and almost on cue traders started feeding their risk-on nickels back into the slot machines.

• The bond shock was digested, not reversed. Stocks were willing to step back into the casino once yields stopped accelerating, but a 10-year Treasury north of 5% means the house is still charging a much higher cover.

• Oil remains the short-term steering wheel, but diesel is the real inflation transmission belt. A diplomatic drop in Brent helps sentiment; it does not automatically remove the energy pressure already embedded in the physical economy.

• AI still owns the equity narrative, but credit is beginning to read the fine print. Mega-cap earnings remain the engine, while hyperscaler financing costs, narrow breadth and ugly Treasury supply are becoming the cracks worth watching.

Risk Gets Its Chips Back

That does not mean the bond market suddenly became friendly. Far from it. It means investors looked at a 10-year yield above 5%, a hawkish Fed pivot, a run of punchy economic data, and an economy still motoring at something close to a 3.6% annualized clip, and concluded that higher yields were not necessarily a recession telegram. They may simply be the admission price for stronger nominal growth.

The Dow jumped 479 points on Friday, the S&P 500 and Nasdaq both gained 0.5%, and all three major indices closed an otherwise bruising week in positive territory. The remarkable part was not that stocks bounced. Markets bounce after bond shocks all the time. The remarkable part was that investors were willing to step back into risk while the 10-year Treasury finished at 5.18%, only a few basis points below the week’s 5.228% high, while the 30-year sat around 5.50% and the two-year remained close to 4.86%.

In other words, bonds stopped punching for a few hours, but nobody should confuse that with them leaving the ring.

Oil gave equities the opening they needed. Brent fell back toward $104/bbl as another round of diplomatic headlines raised hopes that some form of agreement could eventually restore more normal traffic through the Strait of Hormuz. That took some pressure off the inflation trade, calmed Treasury volatility and allowed the equity market to do what it has repeatedly done whenever the macro smoke clears for more than a session: go hunting for earnings.

But oil is still holding the short-term steering wheel.

The problem is that crude prices on a Bloomberg terminal and energy costs moving through the physical economy are not the same thing. Brent can drop three dollars on a diplomatic headline while soaring diesel prices are still crawling through the bloodstream of the economy like cholesterol. Freight, trucking, agriculture, construction, manufacturing, and virtually every shelf in every supermarket eventually get a piece of that bill.

That is why I continue to think of additional Fed tightening less as an attempt to mug the economy in a dark alley and more as an insurance premium against the energy shock leaking into wages, services and expectations. If the economy really is running anywhere close to the pace suggested by the Atlanta Fed and this week’s survey data, the Fed has room to buy that insurance.

And this week certainly gave the hawks ammunition.

The flash composite PMI jumped to 58.4 from 56.0, the strongest reading since 2021. Services accelerated, manufacturing pushed to its strongest level since 2022, orders were firing, hiring strengthened, backlogs increased and input costs rose at their fastest pace in years. Add jobless claims hovering around historically low levels and there was not much here for anyone still clinging to the idea that September’s hike was a ceremonial one and done.

October hike odds moved toward 70%, and the curve began pricing a path that increasingly resembles one plus maybe another rather than one and thank you for coming. Fed officials emerging from blackout had no incentive to throw cold water on that repricing. If anything, the language reinforced the idea that persistent supply shocks cannot simply be waved through the inflation checkpoint because they happen to originate in energy.

This is where Wednesday’s bond massacre starts to make more sense.

Higher oil, stronger growth, hawkish Fed communication, and a wall of hyperscaler issuance all hit the Treasury market at roughly the same time. Duration did not so much stumble as get thrown down the stairs. The five-year crossed 5% intraday for the first time since 2007. The seven-year pushed above 5.05%. The 20-year marched toward 5.45%. The 10-year broke above 5.20%, while the 30-year moved beyond 5.50%.

Over four weeks, the 10-year has risen roughly 50 basis points.

That is not a rounding error. That is the bond market resetting the financing charge on the global economy.

And what makes this episode more interesting than the earlier oil shocks is where the adjustment has taken place. Inflation compensation has not exploded. Real yields have done most of the heavy lifting. That suggests investors are not simply pricing another barrel shock. They are asking whether the US economy can live with structurally higher real interest rates for much longer than the market had previously assumed.

That question has gone global. Gilts, Bunds and JGBs have all been marching to variations of the same drummer. Global sovereign yields are pressing levels not seen in decades. The old 5% Maginot Line in long duration has been crossed, and the next battle is not really the absolute level of yields but the velocity of the move.

That is the chart I would keep front and centre. Equities can digest high yields. What they have historically struggled with is high yields arriving through the front door carrying a baseball bat. The uploaded research shows the market pushing directly into Goldman’s yield-velocity trigger zone, where the speed of the bond selloff matters more for risk appetite than the headline level itself. a

Another yellow light is blinking on the dashboard. Higher yields are not yet generating the sort of demand response you would normally expect. Three poor Treasury auctions this week suggested buyers are demanding a fatter concession to warehouse all this duration, while 30-year mortgage rates have pushed back above 7%.

That is where the bond market stops being a quotation on a screen and starts walking around the economy in work boots.

A 5.2% 10-year is one thing. A mortgage above 7%, more expensive corporate refinancing, more expensive project debt and a higher hurdle rate for every capital-intensive investment decision are another. Eventually the Treasury market rations credit somewhere else. The only question is which borrower hears the bartender call last drinks first.

So far, it has not been mega-cap equities.

That is probably the most important message from the week. Wall Street spent five sessions simultaneously worshipping the AI future and raising the financing cost of building it.

Meta’s Muse announcement put the agentic AI trade back under stadium lights, and the market added roughly $220 billion to Meta’s capitalization as investors moved almost overnight from AI extinction anxiety back toward the idea that agents may become a genuine economic multiplier. The Nasdaq led the weekly gains, technology dominated sector performance, and momentum kept grinding higher.

But under the hood, the limousine was being followed by a hearse.

The S&P, excluding the AI complex, struggled. Small caps lagged. Market breadth deteriorated badly. There were more NYSE new lows than new highs for a ninth consecutive session. Mega-cap stocks did most of the lifting while rate-sensitive and balance sheet-sensitive companies were asked to carry progressively heavier backpacks.

That is not necessarily bearish. Narrow markets can stay narrow for much longer than people imagine when the companies leading them are also printing the earnings. But it means the index is telling a cleaner story than the average stock.

And the credit market has begun telling an even messier one.

Oracle’s force majeure notice around its New Mexico data centre project was not a default notice and did not mean the project had been abandoned. But markets are not paid to read only the headline. They read the escape clauses too. For a financing ecosystem that has pushed hundreds of billions of dollars into AI infrastructure, seeing a hyperscaler reach for contractual protection was enough to remind investors that electrons still have to meet economics.

Project debt weakened. Oracle credit risk rose. Hyperscaler spreads widened even as technology equities remained buoyant. a

That divergence is one of the more interesting fault lines on the board. Equity investors are pricing the AI earnings engine. Credit investors are beginning to price the bill for building it.

The first phase of the AI cycle was about chips. The second was about data centres and electricity. The next one may be about who can finance all this capacity at a cost of capital that no longer looks anything like the free-money era. The bond market is slowly becoming the meter running outside the AI party.

Elsewhere, the cross-asset message was remarkably clean. Strong US data plus a Fed willing to lean against inflation sent the dollar higher for a second consecutive week. Gold struggled beneath the weight of the stronger dollar and the surge in real yields. That is about as toxic a cocktail as one can pour for a non-yielding asset in the short run, even if the structural central bank demand story remains firmly underneath the market.

Bitcoin, curiously, refused to read the same script. It pushed above $87,000 intraday and held much of the move as ETF flows and broader technology risk appetite helped offset the real-yield headwind. For the moment, crypto appears to have been invited to the AI table rather than being seated beside gold at the macro hedging table.

Global equity funds attracted about $44.1 billion, including roughly $37.6 billion into US equities and $5.3 billion into technology. Yet almost all of the enthusiasm was concentrated near the top of the food chain. Large-cap US funds pulled in more than $36 billion, while small-cap funds saw outflows. Investors were buying the generals, not the whole army.

Bond funds also attracted money overall, but government bond funds lost capital while shorter duration and loan products drew interest. Gold and precious metals funds added another $886 million, their tenth inflow in eleven weeks. a

So perhaps the cleanest description of the week is that investors selectively bought almost everything except the piece of the market actually setting the price of money.

The equity rebound made sense. Once yields stopped accelerating and crude slipped, investors had no obvious reason to keep throwing perfectly good earnings overboard. An economy running hot can support corporate revenues for quite some time before a higher discount rate catches up. The market spent Friday deciding that Wednesday’s bond move was painful, but not yet fatal.

I would not fade risk simply because the 10-year is north of 5%. If growth is strong enough, equities can coexist with surprisingly high yields. History is littered with investors who sold stocks because a bond yield crossed some supposedly magical number, only to discover that earnings kept moving the goalposts.

But at some point, the bond market stops being background music and becomes the bouncer.

One reason I am not there yet is that this still does not look like the final stages of a classic productivity bubble. The research in the uploaded note shows baskets of companies expected to benefit from AI productivity have been roughly sideways relative to the market this year, while implied long-term growth expectations remain only modestly above historical norms. That is a long way from saying investors have already capitalized some fantasy productivity miracle out to infinity. a

That leaves us with the market's central contradiction.

The same economic strength supporting earnings is keeping inflation sticky. The same AI investment boom supporting equities is increasing the demand for capital. The same energy shock lifting nominal growth is giving the Fed cover to tighten. And the same Treasury yields that tell us the economy can probably absorb higher rates are steadily raising the price everyone else has to pay to find out.

This is why oil remains the short-term steering wheel while bonds remain the axle.

If Hormuz genuinely reopens and physical energy costs start moving lower, not just front-month crude futures, the bond market gets a pressure release valve. If incoming data cools without collapsing, the Fed can step away from the hammer and risk assets suddenly have room to breathe. If auctions begin clearing cleanly again, the market can convince itself that 5% yields are a level rather than a launch pad.

But reverse that combination and the next act becomes more difficult. Hot payrolls, hot ISM, sticky PCE, and another turn higher in oil would leave the Fed with little reason to soften its language and the Treasury market with little reason to stop charging more rent.

Payrolls, ISM, and PCE will tell us whether this economy is really running as hot as the surveys suggest. Micron will give us another window into AI demand. Nike will say something about the consumer. Accenture should offer a useful read on whether enterprise AI spending is moving from PowerPoint slides into actual budgets.

For now, Friday told us that Wednesday’s bond bloodbath did not kill the bull market. Traders digested the shock, looked at strong growth, resilient earnings and slightly cheaper oil, then walked back into the casino and put another nickel in the machine.

Just remember, it’s the Fed that owns the building now.

The equity market may still be playing the slots, but the bond market stands behind the cage, setting the exchange rate on every chip.

And in the increasingly long-running series of better to be lucky than good, the yen trade has decided to cooperate almost on cue. Only a short while after I put out the call to buy JPY (FX Daily: The Yen Tide Is Turning, but the Beach Is Already Crowded) on the view that USD/JPY had wandered too far from fair value, the market started digesting what the cast of characters with their hands on the machinery had been singing from roughly the same hymn sheet: Japan does not want an endlessly weaker currency, and at the very least would prefer to see the yen dragged back toward something resembling fair value.

Then came the headline: Yen Rises as Japanese Officials Say Weak Currency Is a Problem.

Sometimes the macro thesis works because you have mapped every gear in the machine. Sometimes the people operating the machine simply decide to turn the same wheel you are already leaning into.

As I always say, and as FX has reminded me repeatedly throughout my career, you are often better off being lucky than good. Ask anyone who has spent two decades in the hot seat, and you will probably get the same answer: skill keeps you in the game, but timing and a little bit of fortune are usually sitting somewhere in the P&L too.

Running Update

We’re now into roughly 48 hours of torrential, almost non-stop rain, with another 24 to 48 hours potentially still to come. Up my way in Nonthaburi, parts of my usual running route are close to 40 cm underwater. The drainage simply cannot clear it quickly enough, especially this close to the Chao Phraya.

Fingers crossed it eases enough for me to sneak in at least an hour, but I may just have to accept that the next 48 hours are a write-off. With the Luang Prabang Half Marathon only four weeks away, I would obviously rather keep the kilometres ticking over.

The good news is that I banked some decent runs earlier this week, so missing a day or two now is hardly going to undo the work. Sometimes the smartest training decision is simply not to turn a flooded Bangkok pavement into an obstacle course.