Quant Capital

Energy, memory, liquidity, the long end. Every figure sourced to Fed, EIA, BoK and FINRA data. Positioning over narrative. No forecasts.

[Quant Capital @TheQuantt]

On July 20, with Brent near $88, the consensus said the oil rally was short-covering with an Iran headline attached, and nearly exhausted.

This desk said do not fade it. A squeeze exhausts. A blockade persists.

Two days later we named the four impaired fronts, Hormuz, the Red Sea, the Caspian terminal, Iranian production, and put $100 as the line that would force an inflation-expectations repricing.

Thursday, Brent closed at $100.69. Highest in eight weeks.

And the second half fired. Per CME FedWatch, odds of a Fed hike on July 29 went from 10.7% on July 15 to 34.7% on July 22.

Respect the premium, don't fade it. Watch the long end, not the crude screen.

The data was visible before the move. It always is.

[Quant Capital @TheQuantt]

Nvidia's cash line is $22.4bn. The same balance sheet holds $34.1bn of marketable debt and $42.8bn of marketable equity — $99.4bn liquid. And the $366bn of commitments is a schedule: $120/$100/$98/$16/$10/$22bn by year. Six months' net income: $118bn. 10-Q, filed 26 Aug.

[Kashyap Sriram @kashyap286]

AI bubble co Nvidia has added an additional section to its off-balance sheet obligations - $108.5 billion in guarantees.

Like parents co-signing a lease, $NVDA has to guarantee its customers obligations in order to book sales.

Total obligations now stand at a record $530.5 BILLION.
Total cash on the balance sheet: $22.4 billion.

It is simply impossible for Nvidia to pay its bills when they come due.

And no one is talking about this.

That's the opportunity.

[Quant Capital @TheQuantt]

@StealthQE4 July broke the streak. Real disposable income rose 0.5% in July while real consumption was flat (BEA monthly through July, FRED DSPI and PCEC96), and the saving rate rebounded to 3.0. Claims are 203k. The 1970s comparison was priced off a print that ended this morning.

[Quant Capital @TheQuantt]

Four nominal Treasury auctions in August 2026 left dealers holding 8.6% to 12.5% of the award. A day after the 20-year, the 30-year TIPS reopened: dealers took 2.1%, indirect bidders 84.4%. Same duration, same issuer. The refusal is the nominal coupon, not the maturity.

[Quant Capital @TheQuantt]

@KobeissiLetter Census attaches a confidence interval to both figures: -10.5% MoM (±14.0%) and -6.3% YoY (±19.6%). Both ranges contain zero, and the release's own note says that means it is uncertain whether there was an increase or decrease. It also says it takes 4 months to establish a trend.

[Quant Capital @TheQuantt]

Only the long end went back to 2007. On 21 August the 30-year was 5.27% against 5.22% on 12 July 2007, the 20-year 5.25 against 5.29. Inside ten years nothing did: the 2-year is 70bp lower, the 3-month 109bp. In 2007 that was the price of money. Now it is the price of duration.

[Quant Capital @TheQuantt]

@WallStreetMav $1.3trn is gross, and a fifth of it never leaves the government. Treasury's own interest expense series, ten months to 31 July: $900.1bn on public issues, $270.0bn on Government Account Series, the trust funds. Bondholders are paid the first number. The curve reprices that one.

[Quant Capital @TheQuantt]

@charliebilello Two numbers wear that name. $40trn is gross debt outstanding, and $7.76trn of it is intragovernmental, money the government owes itself. What the market must absorb is debt held by the public: $32.28trn on 21 August, Treasury's own daily series. The long end prices that one.

[Quant Capital @TheQuantt]

@macropaperr The 50-year low is one stock, and it was sold, not consumed. EIA week to 14 August: SPR 293.4mb, down 110mb in a year. Commercial crude 428.8mb, up 8.1mb on the year and up 4.4mb that same week. Production 13.83mb/d. Days of supply divides by consumption as if the wells stopped.

[Quant Capital @TheQuantt]

On 30 July this desk set one test: not whether inflation cooled, but whether the 30-year retreated. Core PCE eased to 3.29% in June from 3.42% in May. The 30-year had just printed its highest since July 2007 and did not retreat. Disinflation did not buy the long end.

[Quant Capital @TheQuantt]

On 27 July this desk published one directive: watch the 30-year, not the guidance. The Fed held on 29 July. The 30-year closed 5.20%, its highest since 12 July 2007, and 5.21% the next day. The meeting moved the long end, not the statement.

[Quant Capital @TheQuantt]

@coinbureau That trillion is a clearing account, not a war chest. Treasury's own daily statement: the TGA closed 20 August at $935.1bn, having paid out $296.8bn and taken in $295.5bn the same day. Three days of its own flow. What it spends on buybacks it reissues to refill. Price the refill.

[Quant Capital @TheQuantt]

The level did not. DFII30 hit 3.06% on 17 August, a high for the series, then fell to 2.94% on the 19th. The published invalidation was 2.90%. It came within four basis points and did not trigger. Watch 2.90, not the headline yield.

[Quant Capital @TheQuantt]

On 13 August this desk published one directive: track the 30-year real yield, not CPI. The move in the long end was term premium, not an inflation scare. Eight sessions later the arithmetic is in, and it splits in two.

[Quant Capital @TheQuantt]

The mechanism held. The 30-year breakeven printed 2.25% on 14, 17, 18 and 19 August while the nominal traded a 12 basis point range. Expected inflation did none of the work. That half of the call is tested and standing.

[Quant Capital @TheQuantt]

@ekwufinance A basis trade is not a duration buyer. It is long cash against short futures, close to duration neutral, so it intermediates the paper rather than absorbing it. The swap does not remove a bid, it changes who warehouses it and on what margin. The 30-year closed 5.27% Friday.

[Quant Capital @TheQuantt]

@pequityresearch Those two bullets are the same forecast disagreeing with itself. A 340% capex response is not a condition for the shortage lasting, it is what ends it. Memory supply lands two to three years after commitment, so 2024-2027 capex arrives into 2027-2028. Watch the commitment date.

[Quant Capital @TheQuantt]

@StockSavvyShay Same GPUs priced lower is not a second business model, it is a balance sheet. A hyperscaler funds capex from cash flow and can hold price. CoreWeave met 20.57bn of capex with 17.14bn of net borrowing against 5.54bn of cash. The discount is debt service. Price the tenor.

[Quant Capital @TheQuantt]

What would break this read: energy rolling over and core cooling into September, which would make the hawkish dissents the cycle's last hawks. Watch the distillate complex and the next inflation print. Risk is not symmetric until it is.

[Quant Capital @TheQuantt]

The July FOMC statement names the desk's oil thesis in its own words: inflation remains elevated "in part reflecting supply shocks that have driven price increases in certain sectors, including energy." The energy-to-inflation channel we flagged in July is now in the record.

[Quant Capital @TheQuantt]

And the committee split runs one way. The 29 July hold was a 9-3 vote, with Hammack, Kashkari and Logan each dissenting in favour of a quarter-point hike. Nobody dissented for a cut. Brent ran from $86.99 on 20 July to $105.32 on the 23rd. That pressure is real.