The Charts Behind the Costume

JGB 10yr through 3% for the first time since 1996. It's the end of thirty years of monetary history in one print. Government assumed 3% for FY2026 debt-servicing costs, so this isn't academic, it's already blowing through the MOF budget math.

PM Sanae Takaichi has doubled down on Abenomics reflation, not abandoned it, an investment-led growth platform, aggressive fiscal expansion, ambitions in semiconductors and AI, just with the BoJ finally now having to raise rates into it rather than suppressing them.

Here's what actually matters globally. Japan was the marginal buyer of duration everywhere for a decade, pension funds, insurers, even the BoJ itself, all running the same carry trade, yen funding, buying UST, Bunds, anything with yield. That buyer is gone. Yen isn't a funding currency in this anymore, it's the thing being defended.

JGB 2yr at a 31-year high near 1.8%, 5yr at a record. This is the BoJ playing catch-up after being called "behind the curve" for years, with markets now pricing a hike this month as close to certain. The short end is telling you the wait is over.

And watch the independence question closely. It was unnerving, if I was a Japanese citizen, watching the photo opportunity posted on US Treasury Bessent's X feed. The day before, Bessent was on record publicly praising Ueda as an "underrated, savvy market operator" and insisting Washington doesn't dictate BoJ policy. Then, per NHK, behind closed doors he told Tokyo directly its next move should be a hike, so its currency strengthens and Japan doesn't have to intervene in yen/USD by selling its US Treasuries. Though that's better than a forced seller, it still guarantees no marginal buyer. Japan needs to be seen raising rates because its own inflation and fiscal math demand it, not because Washington asked.

Gold futures weekly, at $4,700+ bulls were too exuberant. COT showing expanding speculators, call buying notional close to records seen at the January 26 highs. Monthly chart showing a morning doji star, love my candlestick patterns, with the green up candle overshooting the downside candle from two months prior.

Market kept failing in the overnight session near highs, London and New York never reclaiming them, the first tell-tale of exhaustion going into Core PCE and Jackson Hole. Once headline PCE came in higher than expectations, it seemed hard to see how Warsh couldn't acknowledge that reality, though I was braced for the productivity spin, which his mentor Greenspan was so masterful at, and ultimately so wrong on.

Targets $4,350 to cover some put risk on. I covered some shorts at $4,450 to 90. Premature for sure, even in my spec book I find it hard to remain short for too long, and I'd been fighting most of the week to keep a sizeable short within my risk management tolerance. For once it paid off. A rare victory on the short side.

The weekly market profile hints at a target of $3,750 to 80. See the right-hand side histograms, short histograms are areas the market hasn't traded through, and they can often act as a magnet, pulling price back to fill them.

I'm not in that camp as even an initial sell-off in US equities. Like 2007 to 08, the first 20% downdraft in equities saw gold rally first, before getting caught up in the wholesale deleveraging later in the subsequent 53% correction in the SP500.

And to my mind, gold and silver led the way ahead of bonds and equities, to the downside.

As will see later, the 4hr showed a low-volume thrust break into the JH event, with the GSR having lagged horribly, gold outperforming silver into the lead-up to the PCE release, and JH itself seeing gold sticky bid while silver constantly lost a bid, down to$67.50 many times over, with gold rebounding simultaneously up at $4,640 to 70 on spot.

A crude ABC correction off the highs of 72 takes us to the $61.80/62 zone, though my initial target was $63.25 to 75, which, ahead of NFP, is clearly a decent location off the highs if wanting to be long again.

For me, the GSR chart is showing risk of over 69 on the spread, and that silver will very much get caught up in any SPX-NDX correction is the weight of probability. Though holding long silver vs equities on spread, while volatile, may be a way to play it for some. See later tactical charts at the end.

Gold stocks vs gold, breaking out of a consolidatory period of almost fifteen years, spanning two gold cycles. That's the real story here, not the price of gold itself but the market finally re-rating the businesses that produce it. Gold stocks lead. When the miners break out first, the metal usually follows.

The fundamentals back it up. Median profit margins for gold miners now sit at 31%, the highest of any sector in the S&P 500, tech included. Free cash flow is running at records across the majors, Newmont posting $7.3 billion for full-year 2025 and already $5.3 billion through the first half of 2026 alone, both Q1 and Q2 individually its best quarters on record, Agnico Eagle posting its best quarter ever, Kinross its fourth straight record. Debt-to-equity has been cut roughly in half. This is a completely different animal from the sector generalists learned to avoid for thirty years, miners plowing every extra dollar into overpriced acquisitions at cycle tops. This cycle, it's buybacks and balance sheet repair instead.

SILJ vs silver, similar story to the gold miners, though silver itself is still a long way off its highs, down roughly 45% from the January 2026 peak near $118 versus current levels in the $60s. That gap is exactly the point.

Silver producers' all-in sustaining costs are running at $12.21/oz on the Silver Institute's own global average for 2025, the lowest since 2022. At current silver prices that's a margin of $45-plus an ounce, the widest in the industry's recorded history. This is interesting as costs aren’t necessarily catching up to price, it's costs falling while price rises, the best of both directions at once.

Yet the miners haven't caught up to that reality. Silver miners have historically run at roughly three times the metal's move through a full cycle. In 2025, silver rose about 300%, SILJ only around 200%. That's an inversion of the normal relationship, and it's the gap that closes once the market excepts that earnings are remaining high relative to what the cost structure is already showing.

Same conclusion as gold. The miners lead, and right now they haven't even started catching up.

Bloomberg Commodity Index, I believe a secular supercycle is underway. This isn't one sector's narrative, it's geopolitical risk touching every corner of the complex at once.

Start with gold. S&P Global confirms zero major discoveries, 2 million ounces or more, in both 2023 and 2024, the first time that's happened in two consecutive years since records began. Only six major discoveries total since 2020, just 27 million ounces combined, average discovery size down from 7.7 million ounces a decade ago to 4.4 million now. The Aurelians and Redbacks of the world aren't being replaced, a couple of the 10-million-ounce, high-grade discoveries. Exploration spending fell 15% in 2023 and another 7% in 2024, even as prices hit records, the exact opposite of what a functioning market is supposed to do. Of course this will change, and is changing, due to renewal in the sector and FCF, but it takes time to ignite exploration success.

Copper tells the same story through a different mechanism, falling grades compounded by an input crisis nobody was pricing. Chile produces roughly a quarter of the world's mined copper and depends on imported sulphuric acid to leach it, more than a third of that from China. China banned acid exports from May 2026 to protect its own fertiliser industry, on top of a Middle East conflict that choked global sulphur supply through the Strait of Hormuz. Chinese acid shipments to Chile hit zero in March, the first total stoppage since 2023. Morgan Stanley puts over half of Chile's refined copper output at risk. That's not even a grade issue or a demand one, it's a single feedstock chemical nobody thought about until it vanishes, unless you are a producer of said copper. (Robert Friedland).

The substitutes aren't clean either. Aluminium, pushed as the EV-transition alternative to copper, runs into the same Middle East supply chain, the region accounts for roughly 8% of global production and took its own hit from the same conflict. There is no simple escape hatch here, the inputs are shared across the whole complex.

Layer on critical minerals and rare earths, China restricting exports as a matter of explicit policy, and potash, still concentrated in Belarus and Russia and still tangled in sanctions years after the fact. Add agriculture, cattle and livestock herds rebuilding slowly after years of drought-driven culling, grains exposed to the same El Niño and climate volatility that's disrupted planting windows repeatedly this decade, and fertiliser inputs squeezed by the same sulphuric acid crunch hitting copper. Softs carry their own version of this, cocoa and coffee both still working through supply shocks that price alone hasn't fixed.

None of this is being met with policy that matches the scale of it. A decade of ESG-driven capital starvation hit exploration and greenfield development across the entire extractive complex at once, right as demand didn't slow down. Add an increasingly isolationist, protectionist turn from both Washington and Beijing over the metals that matter most, and you have a policy environment actively worsening a supply problem it barely acknowledges.

A significant deleveraging event, if equities crack the way this whole page suggests they might, would pull commodities down with it, no sector is immune to a genuine liquidity event. But that pullback wouldn't fix any of the underlying supply constraints, it would just remove capital from a sector that already can't fund enough new supply. It sows the seeds for the next leg being larger and more violent than this one, not smaller.

DXY, and here's the tension. Fiscal incontinence is a recipe for dollar decline over the medium term, that's the structural case and I hold it. Protectionism and tariffs upend the trade and capital account balance the dollar's reserve status has always depended on, the US can no longer assume strangers will keep funding the twin deficits, especially while pulling back from being the world's policeman, Japan being the one clear exception, the Pacific bases there are too strategically important to walk away from.

But structural and immediate aren't the same trade. A correction and a genuine tightening in financial conditions, which I think we're about to get a whiff of, means dollar shortages, not dollar abundance. That's what a real deleveraging event does, it forces a scramble for the currency the world's debts are denominated in, regardless of the long-term fundamental case against it. I'd watch for DXY over 101, and if this gets properly nasty, 105 is not an unreasonable target.

Positioning supports this. CFTC data showed speculative dollar shorts at their most extreme level in over a decade, only recently starting to unwind. Too many people are already in the debasement trade. That's exactly the kind of crowding that gets squeezed hardest when the tightening actually arrives.

ULSD futures, the sharp end of everything just discussed. This is the actual price the crack spread translates into at the pump and in home heating oil.

National average retail diesel was $4.796 a gallon back in July, and that was with crude well off its highs. This is the chart that shows whether relief at the crude level is actually reaching consumers, and right now it isn't. This is the US, but across Europe a similar theme on pricing in the home for natural gas and generated electricity exists.

SPX, up close to 19% over the last year and still near highs, even after this week's wobble. That alone isn't the point. What matters is what credit's been doing underneath it, priced completely differently. Oracle's five-year CDS has widened 70 basis points this year to around 215, a multi-year high. Broadcom's up 48, Nvidia's up 32 to its widest level since the contract started trading. Hyperscaler bond spreads have widened across the whole curve, short paper, belly, long end, all moving the same direction. Equities have kept rewarding AI capex. The market that actually insures the debt behind it has been charging progressively more to do so. That's the disconnect, and credit is usually right.

iBoxx, the same disconnect from the other side. This is the broad investment-grade tape, not just the AI-adjacent names, and it's still where you'd expect a market this late in a cycle to be, spreads historically tight, priced like nothing's wrong.

That's exactly what makes the individual widening in Oracle, Broadcom and Nvidia worth watching closely. It isn't showing up in the index yet. Credit stress nearly always starts as a single-name move before it becomes market-wide, and right now that's precisely the stage this looks like it's at.

Note on Dow Theory

The following equity charts are also worth viewing through the lens of Dow Theory. At its heart, Dow Theory holds that the market discounts the future and that a genuine primary trend should reveal itself through confirmation across economically related parts of the market. Charles Dow originally looked for confirmation between the Industrials and the Railroads. If industrial companies were prospering, the companies transporting their goods should ultimately be telling the same story. A failure to confirm was not itself a sell signal, but it was an important warning that the headline index might not be telling the whole story.

The particularly clever aspect of Dow Theory is that this principle can be extended far beyond the original Dow averages. The question is simply whether the market underneath is confirming the headline price. Today we might look for confirmation—or divergence—between equities and credit, the S&P and transports, semiconductors and the broader market, AI equities and the bonds financing the AI build-out, or gold and the mining shares. In other words, apparent strength in one market becomes more credible when the economically connected markets confirm it; persistent non-confirmation tells us to look more closely.

This is especially relevant to the charts that follow Next month I will return to Dow Theory in more detail, including the legendary Richard Russell and his Dow Theory Letters, one of the longest-running investment letters in market history. Russell's great contribution was to treat market action not merely as technical analysis, but as a means of diagnosing the underlying economic, monetary and credit cycle.

DJ Automobiles & Parts Titans 30, my stand-in for the transports leg of Dow Theory, and the non-confirmation here is textbook. This index topped near 969 in March 2026. The second attempt in April/May couldn't even get back there, stalling out near 919, a lower high. From there it broke down hard into the 750s by mid-summer, and even the recent recovery back to 843.86 leaves it well below both peaks.

DJIA did not confirm any of this. It kept making new highs through the same stretch. That's precisely the warning Dow himself would have flagged, not a sell signal on its own, but a genuine divergence between the headline average and the economically connected market underneath it.

Worth remembering while looking at this pair specifically, the DJIA is price-weighted, not cap-weighted and not equal-weighted, a quirk unique to the Dow. A handful of expensive names can carry the average to new highs while the broader market, and clearly this one, isn't agreeing at all.

Dow Jones Utility Average, the classic third leg of extended Dow Theory alongside Transports, and it's telling the same story as the Autos & Parts Titans chart. DJU hit a high of 1,202.79 in early 2026, failed to hold it, made one more attempt at a similar level months later, and has since broken down to 1,068.09, roughly 11% off that high, with its 50-day average now crossing below the 100-day.

DJIA confirmed neither of these. Not the transports proxy, not the utilities. Two of the classic confirming legs both topped out and rolled over months ago while the headline average kept making new records through the same period. This is not a single index behaving erratically. It is the breadth of non-confirmation that Dow Theory was formulated to identify.

Taken together with the Autos & Parts chart, the argument carries considerably more weight. One non-confirming index is an anomaly. Two, moving in the same direction and independently signalling the same thing, is precisely the divergence Richard Russell spent a career instructing investors not to disregard.

SMH Semiconductors sector. I think of this as an earnings bubble as much as a price one. The July rout was just that, SMH fell nearly 18%, SOXX 21%, and it wasn't random. It coincided almost exactly with the collapse of Situational Awareness, the AI hedge fund run by 24-year-old former OpenAI researcher Leopold Aschenbrenner, ironically named after his own 2024 essay about seeing the future clearly. Running roughly 4x leverage on a concentrated semiconductor book, the fund went from $45 billion to $10 billion in weeks as Sandisk, Micron, Bloom Energy and Nebius all cratered together. Days before the forced sale, Aschenbrenner wrote to investors calling it "some of the most attractive opportunities since early 2025" and asked for more capital. None came. Citadel picked up the $16 billion public portfolio at roughly a 10% discount, an opportunistic bailout dressed up as a rescue.

NVIDIA. Low P/E here is not the same thing as cheap. Earnings have been growing so fast that the multiple compresses even as the underlying risk in the business is rising, that's the reflexive trap, a falling P/E looks like value right up until the earnings themselves are the thing in question. Heading into the recent print the stock still jumped nearly 9% on a beat and a 70% FY28 growth guide against 44% expected, which tells you sentiment is still doing a lot of the work here regardless of the multiple.

And the 10-Q underneath that beat is genuinely revealing. Three customers account for 44% of first-half revenue, five customers for 70% of receivables, and Nvidia has been extending payment terms to some of its largest customers, receivables jumped from $38.5 billion to $63.1 billion in six months. None of that shows up in the P/E. It shows up in the footnotes, and it's exactly the kind of concentration risk a cheap-looking multiple is currently masking rather than reflecting.

Worth noting Nvidia's own presence in the US equity indices has grown right alongside the semis, replacing Intel in the Dow in 2024, with Alphabet joining behind it this June. Mega-cap tech now effectively sets the tone for the "blue chip" average too, not just the Nasdaq. Hence the disconnect of DJIA to DJT and DJU.