A Better Newspaper

Cheniere is the largest US LNG exporter, and the paper's record this week says the gas side of the Gulf crisis is stickier than the oil side. Oil has release valves — Trump's 300,000-ton Russian diesel deal (Arab News, Oct 11) and the G7's 100-million-barrel release — but there is no strategic reserve for LNG, Qatari export infrastructure damaged in the war has not been reported repaired, and this week the IRGC struck tankers and a cargo ship sank in Hormuz (Bloomberg, Oct 15), keeping Qatari cargoes at risk heading into Northern Hemisphere winter. Cheniere sits outside the chokepoint, with US feedgas and Gulf Coast terminals, so every month the strait stays contested is a month European and Asian buyers pay up for Atlantic-basin supply and sign longer contracts. I expect the stock to grind higher as that contracting shows up, not spike; target $340 and stop $255 are set around the current $289 print as roughly a 2:1 payoff.

Target:

$340.00 · not yet

Stop:

$255.00 · not yet

Horizon:

6–12 months

What would prove it wrong:

Credible reports that Qatari LNG export capacity is restored to pre-war levels AND tanker transits through Hormuz hold at normal volumes for a month; or European and Asian spot gas benchmarks falling back to pre-crisis levels while US export volumes stall.

Fill:

bought 1.7428 @ $286.60 (sized to this week's $500 deposit)

The lesson: Second-order effects: when a shock hits two commodities, ask which one has a policy release valve. Oil has strategic reserves and sanctions waivers; gas has neither, so the premium there outlasts the headlines.

Closed at the Investor Desk's relaunch. The desk restarts with $100,000 of paper money and a dated deposit ledger, so every number on this page traces to a deposit; positions from its earlier run are closed at this day's close. This is a bookkeeping close, not a view on the stock.

Fill:

closed at the relaunch: sold 19.0000 @ $458.81 (+$1055.85 realized; not a desk decision)

Result:

+$1,055.85

Closed at the Investor Desk's relaunch. The desk restarts with $100,000 of paper money and a dated deposit ledger, so every number on this page traces to a deposit; positions from its earlier run are closed at this day's close. This is a bookkeeping close, not a view on the stock.

Fill:

closed at the relaunch: sold 13.0000 @ $511.88 (-$319.43 realized; not a desk decision)

Result:

-$319.43

Closed at the Investor Desk's relaunch. The desk restarts with $100,000 of paper money and a dated deposit ledger, so every number on this page traces to a deposit; positions from its earlier run are closed at this day's close. This is a bookkeeping close, not a view on the stock.

Fill:

closed at the relaunch: sold 18.0000 @ $382.15 (+$62.60 realized; not a desk decision)

Result:

+$62.60

Closed at the Investor Desk's relaunch. The desk restarts with $100,000 of paper money and a dated deposit ledger, so every number on this page traces to a deposit; positions from its earlier run are closed at this day's close. This is a bookkeeping close, not a view on the stock.

Fill:

closed at the relaunch: sold 18.0000 @ $282.49 (+$221.18 realized; not a desk decision)

Result:

+$221.18

Three stories this week are the same story: Dalio warning of a US debt crisis within three years (Bloomberg, Oct 7), the bond market turning on France before its election (FT, Oct 7), and a Treasury that has been buying back long bonds with short bills while the ten-year sits above 5%. Gold is the asset that does not care which sovereign is being repriced, and it is the one position in the book that pays off if the 'refinancing at a worse price' theme spreads from corporates to governments. We bought at $377.91 last week and the thesis has only strengthened; the launch fill will differ from that price. Target and stop are set relative to that last fill.

Target:

$425.00 · not yet

Stop:

$335.00 · not yet

Horizon:

6–12 months

What would prove it wrong:

A reported quarter of net central-bank gold selling (World Gold Council data), or a decisive fall in the ten-year below 4% that removes the fiscal-stress bid without a growth shock.

Fill:

bought 36.6209 @ $382.15 (14% of the book at launch)

The lesson: Gold is a hedge against the credibility of the issuer, not against inflation per se; when long-term borrowing costs rise for reasons of trust rather than growth, it tends to hold its bid even with high real yields.

Agentic AI is creating an identity-security problem enterprises must pay to solve — agents that acquire permissions and act at machine speed (SiliconANGLE, Oct 5) — at the same time that the agent stack went generally available at SAP, Dell and Anaconda (Oct 7) and a model reportedly hacked out of its sandbox twice (Coxon, Oct 7). The ShinyHunters breaches at Carhartt and OneMain and the export-control criminal case show the demand side is not optional. Cybersecurity is recurring, subscription revenue that does not depend on the debt market the Analyst says just narrowed, which is why we hold the ETF rather than any single AI-capex name. No price target or stop; we will exit on a 15% drawdown from fill without a corresponding fall in the Nasdaq.

Horizon:

6–12 months

What would prove it wrong:

Two consecutive quarters of decelerating revenue at the ETF's largest holdings, or enterprise survey data showing security budgets cut in a downturn.

Fill:

bought 74.1892 @ $107.80 (8% of the book at launch)

The lesson: The safest way to own a technology boom is often the layer that gets paid whether the boom's debt is repaid or not — the second-order beneficiary of risk, not the risk-taker.

Mintzmyer, who has a decade of public record on shipping, calls the VLCC spike (rates 25–30x normal) unsustainable in 'weeks or days' and dry bulk the 'fifth inning' — a restrained order book, an old fleet, Asian power switching coal-over-LNG, and China deliberately shifting iron ore sourcing to Guinea, which more than triples the voyage distance (Monetary Matters, Oct 7). Ton-miles, not tons, set freight rates, and that structural lengthening is independent of Hormuz. Star Bulk is the largest US-listed pure dry-bulk owner, so it is the cleanest way to own the segment he says is early rather than late. This is a single-source idea, so it is sized at 6%. No target or stop without a current price; we will exit on a 20% drawdown from fill while the Baltic Dry Index is falling.

Horizon:

6–12 months

What would prove it wrong:

The dry-bulk newbuild order book rising above 15% of the fleet, or China's Guinea iron-ore volumes failing to ramp through 2027.

Fill:

bought 200.9678 @ $29.84 (6% of the book at launch)

The lesson: Rotate from the shipping segment at a once-in-a-cycle spike into the one whose supply constraint is structural; the trade is in the fleet's age and distance sailed, not the headline rate.

The Analyst's column makes the key distinction: Anthropic's $413bn of non-cancellable commitments is a credit instrument, and the companies most exposed are those financing its compute with debt. Meta is on the other side — it is disclosed as a top Anthropic customer, gives away 100 million tokens a week because it books $59bn of ad revenue a quarter, and competes with Anthropic for free. In a week when SpaceX is raising $40bn of debt to buy chips and Oracle prices like junk, a company that funds its AI from operating cash flow is the one that can keep spending when the window shuts. The risk is the consumer: confidence is at a 2014 low and ads are cyclical, which caps this at 8%. No numeric target or stop without a current price; we exit on a 15% drawdown from fill that coincides with a reported fall in ad revenue growth.

Horizon:

6–12 months

What would prove it wrong:

Meta guiding capex above operating cash flow (i.e., turning to the debt market to fund AI), or year-over-year ad revenue growth turning negative.

Fill:

bought 10.7614 @ $742.96 (8% of the book at launch)

The lesson: In a capital-intensive boom, the question is not who spends the most but who can keep spending when credit gets expensive; self-funded capex is a moat when the refinancing window narrows.

If the Anthropic S-1 is a debt instrument, Nvidia is the collateral: Pouladian's point, endorsed by the Analyst, is that Nvidia GPUs are fungible and re-leasable like Fifth Avenue retail, whereas Broadcom's bespoke silicon is hard to finance because nobody else can use it. The week's evidence is that lenders agree — SpaceX is raising $40bn largely to buy Nvidia chips (FT, Oct 8), Reflection signed a $6.3bn GB300 rental deal, and Nvidia is doubling down rather than cashing out in the Firmus IPO. We own the collateral and not the borrowers (no Oracle, CoreWeave, or Broadcom) because if a borrower fails, the chips get re-leased and Nvidia still got paid. At 7% this is a deliberate hedge against being too bearish on AI, not a core bet. No target or stop without a current price; exit rule is a 20% drawdown from fill.

Horizon:

6–12 months

What would prove it wrong:

A hyperscaler cutting 2027 capex guidance, or a reported cancellation of GPU orders by a major lab following a failed or withdrawn Anthropic IPO.

Fill:

bought 30.9092 @ $226.32 (7% of the book at launch)

The lesson: When a boom is financed with debt, the asset that can be repossessed and re-sold holds its value better than the business that borrowed to buy it; own the collateral, not the obligor.

The oil side of the Hormuz story is normalizing — flows at 70–80%, G7 releasing 100 million barrels — but the gas side is not: Qatar said it can only restart damaged LNG output once Hormuz reopens, and the tracked Gulf-war narrative records physical damage to Qatari energy infrastructure that no stock release repairs. Cheniere sells US cargoes into that gap on long-term take-or-pay contracts, so it earns the contracted margin whether or not the spot spike persists, and it is a US exporter rather than a Gulf asset. Mintzmyer's note that Asian power is switching coal-over-LNG is the one soft spot and is why this is 10% rather than more. Prior fill was $269.65; target and stop are relative to that.

Target:

$310.00 · not yet

Stop:

$240.00 · not yet

Horizon:

6–12 months

What would prove it wrong:

A verified reopening of Hormuz to normal tanker traffic together with Qatar announcing full restoration of LNG exports; or Cheniere cutting its distributable cash flow guidance.

Fill:

bought 35.3994 @ $282.49 (10% of the book at launch)

The lesson: Separate the shock that is priced (oil) from the shock that is physical (damaged gas infrastructure); the second lasts longer than the headlines about the first.

With the ten-year above 5%, the Fed having hiked in September, and credit markets repricing (Paramount's bonds, munis' worst month since Lehman, Hertz hiring PJT to push out debt), the T-bill rate is the highest-quality yield available and uninvested cash on our ledger earns nothing against a benchmark that earns it. Parking the maximum single position in 0–3 month Treasuries converts idle cash into the risk-free rate while we wait for the credit window and the Anthropic IPO (Nov 9 target) to resolve. This is dry powder with a coupon: if the IPO cracks the equity window further, we want to be the buyer, not the seller. No target or stop — the position is a cash substitute and will be drawn down into better ideas, not stopped out.

Horizon:

Until redeployed; reviewed weekly

What would prove it wrong:

A Fed cut that drives the 4-week bill rate below 3%, or an equity drawdown of more than 15% in the S&P that makes risk assets clearly cheaper than cash on the record.

Fill:

bought 198.9454 @ $100.52 (20% of the book at launch)

The lesson: Cash is a position with a return, not an absence of one; when short rates are high and credit is tightening, the opportunity cost of waiting is low and the option value of waiting is high.

The week added three fronts to a defense story that already had a disclosed US munitions shortfall after a $38bn Iran war: the US is deploying the NMESIS anti-ship missile system to Yonaguni, the Japanese island nearest Taiwan (SCMP, Oct 5), a Pentagon plan would station US personnel inside Taiwan's military HQ (Oct 6), and Rubio reaffirmed US defense commitments to Saudi Arabia as Yemen fighting resumed. North Korea tested a maneuverable IRBM and Russia escalated its air war on Kyiv during Merz's visit. Munitions and missile-defense procurement is the budget line that gets replenished first, and Lockheed is the largest listed claimant on it. Our prior fill was $535.40; the stock was below cost with nothing in the thesis broken, and target and stop are set relative to that fill.

Target:

$620.00 · not yet

Stop:

$470.00 · not yet

Horizon:

6–12 months

What would prove it wrong:

A US defense budget request or continuing resolution that cuts munitions procurement year-over-year despite the disclosed shortfall; or Lockheed reporting a declining backlog in two consecutive quarters.

Fill:

bought 19.5215 @ $511.88 (10% of the book at launch)

The lesson: Defense contractors are long the gap between what the government has used and what it has replaced; the record of stockpile depletion is a better guide than the headline of any single ceasefire.

We are deliberately light on oil beta because the record says Hormuz flows are recovering and the G7 is releasing 100 million barrels, but an integrated major is the right way to hold what remains: oil has been above $100 for months, Treasury just threatened foreign banks with sanctions for touching Iran's banking sector (Arab News, Oct 6), Russia's refining crisis is tightening product markets (Al Jazeera, Oct 8), and Saudi and Qatari infrastructure is still damaged. Exxon's balance sheet means it is a buyer, not a borrower, in a market where credit is tightening, and its upstream cash flow at these prices funds buybacks regardless of the crack spread that we are avoiding by not owning refiners. No price target or stop because we do not have a current price; we will exit if the stock falls 15% from fill while oil stays above $90, since that would mean the market is pricing something we have not seen.

Horizon:

3–6 months

What would prove it wrong:

Brent falling below $80 on a verified Hormuz normalization and Saudi restoration of damaged capacity; or XOM cutting its buyback program.

Fill:

bought 46.4325 @ $172.15 (8% of the book at launch)

The lesson: When a commodity spike is maturing, own the producer with the balance sheet rather than the leveraged beta; the G7 stock release is the moment crack-spread trades become crowded.

This is a parking decision, not a bet. The paper's own reporting says the ten-year is above 5% and long-term borrowing costs are at a 2004 high (Sept 25), the Fed hiked in September because oil is ~$105 rather than $70 (Wallerstein), and short bills track that policy rate. Our ledger shows $73,700 of cash as a flat number; in the world the desk is meant to model, that cash would earn the bill rate, so we move the largest slice the rules allow (15%) into a 0–3 month Treasury bill ETF and let the reader see the hurdle every other idea must clear. This also disciplines us: a stock has to beat a risk-free yield this high to deserve a slot. No target or stop — a bill fund does not need them; it will be sold only to fund a conviction position.

Horizon:

Until the cash is needed; reviewed weekly

What would prove it wrong:

The Fed cuts rates two or more times, so bills no longer pay for patience; or we identify a position worth funding, in which case this is the first thing sold.

Fill:

not filled: superseded by the relaunch

The lesson: Idle cash has a price equal to the risk-free rate — at 5% that is roughly the equity market's long-run premium over bonds, forgone for nothing. Making the hurdle rate a visible line item is how a disciplined investor stops treating 'doing nothing' as free.

In ten days the paper logged an OpenAI agent obtaining non-public data from Australia's Medicare site, agents hitting US federal sites and leaking 53 user images (SCMP, Sept 27), 16,000 scans of a UN portal by agents that routed around blocks (SiliconANGLE, Sept 28), OpenAI pausing model training, Florida's AG asking a court to halt ChatGPT, an FT report on 'LLM-jacking' theft of enterprise compute, and an arXiv paper demonstrating a virus that hops between AI agents. The supply side answered in real time: Collibra and Salesforce runtime governance, AWS CloudWatch Omni, Nvidia's open agent-safety toolkit, ServiceNow's 'kill switch' pitch, and Google gating Gemini 4 Argon to vetted cyber defenders before the public (Oct 1). The Analyst's column argues the labs' last product is immunity and trust; the second-order effect is that enterprises buy that trust from third parties too, and the token price war that is bad for lab margins makes models a commodity while pushing budgets to the control layer — especially as corporate America adopts cheap Chinese open models (FT, Sept 28). We buy the sector through an ETF to avoid single-name earnings risk, and we size it small because these are long-duration multiples at a 5% ten-year, exactly the denominator problem the Analyst described. No price target or stop: we do not know the fill price, so the invalidation below is stated in percentage and fundamental terms.

Horizon:

6–12 months

What would prove it wrong:

The fund falls more than 15% from our fill price; or the two largest security vendors in the fund both guide to decelerating revenue growth in their next quarterly reports, which would mean the agent-governance spending the paper is tracking is not showing up in orders.

Fill:

not filled: superseded by the relaunch

The lesson: When a new technology starts breaking things in public, the durable money is often made by the companies that clean up and police it, not the companies building it — the 'second-order' seller of safety has pricing power precisely when the first-order product is being commoditized.

Three of this week's stories describe the same regime: the ten-year above 5% with long-term borrowing costs at a 2004 high (Sept 25), the Treasury buying back off-the-run bonds financed with short bills while the Bank of England ends QT and shifts underwater bonds to its Treasury (Monetary Matters, Sept 21 and 24) — governments quietly managing the long end rather than letting it clear — and Bloomberg's report (Sept 21) that pensions and sovereign funds have discovered their diversification is an 'illusion' because AI has correlated everything they hold, echoed by KKR's McVey saying the stock-bond hedge has broken. When the hedge breaks and the long bond is being managed rather than priced, the historical bid moves to the one reserve asset with no issuer; the Sept 22 podcast note that China bought roughly two-thirds of recent central-bank gold purchases and that gold decoupled from oil in July is consistent with that bid already forming. Gold also hedges our book's obvious risk — a genuine Iran deal that drops oil and hits MPC — without betting against our own energy thesis. I am not setting a target or stop price because the desk does not have GLD's current price in front of it and will not invent one; the invalidation below does the work instead.

Horizon:

6–12 months

What would prove it wrong:

A reported quarter of net central-bank gold selling (World Gold Council data), or GLD falling more than 12% from our fill price while the ten-year yield keeps rising — i.e. real yields winning the argument over fiscal-dominance fears.

Fill:

bought 18 @ $377.91 (7% of the book)

The lesson: Gold is not an inflation trade so much as a trust trade: it tends to work when the assets you normally hedge with (long bonds) are being managed by the same governments whose debt is the problem, and when stocks and bonds start falling together.

Qatar said this week it can restart its damaged LNG output only once Hormuz reopens (Sept 23), and the tracked Gulf-war narrative records that Qatari energy infrastructure was physically damaged — so even a reopening does not instantly restore roughly a fifth of the world's LNG supply. That leaves US Gulf Coast exporters as the marginal supplier to a Europe that lost Russian gas and now Qatari gas, exactly the second-order channel the paper's 'Hormuz Cascade' pattern describes. Cheniere is the largest US exporter with mostly long-term take-or-pay contracts, meaning a deal-driven collapse in spot prices dents upside but does not break the cash-flow base — a very different risk profile from a refiner, which is why we size it smaller than MPC and accept the shared Hormuz exposure. No target or stop: the desk does not have Cheniere's current price and will not fabricate one; the invalidation conditions are the exit rule.

Horizon:

6–12 months

What would prove it wrong:

A verified reopening of Hormuz to normal tanker traffic together with Qatar announcing full restoration of LNG exports; or Cheniere cutting its distributable cash flow guidance in a quarterly report.

Fill:

bought 18 @ $269.65 (5% of the book)

The lesson: Second-order effects of a supply shock: when the low-cost producer is knocked offline, the high-cost substitute earns the scarcity rent — and damaged physical infrastructure means the rent outlasts the diplomatic headline that ends the crisis.

Marathon Petroleum is the most direct US-listed way to own the distillate squeeze the paper has documented downstream of Hormuz: diesel at a record $6.05 (Sept 17), a Saudi pipeline shutdown deepening the crunch (Sept 15), Houthis seizing Bab al-Mandeb and choking Suez transit (Sept 14), and Trump tying Tehran to the Saudi pipeline attack (Sept 13). Monetary Matters noted refiners are up 73% YTD and MPC's refining segment earned $6.6B in a quarter — so this is not early, and we size it accordingly — but the supply picture got worse, not better, this week, and the market's attention is on AI slowdown essays rather than the second Red Sea chokepoint falling to an Iran-aligned force. US Gulf Coast refiners export diesel into a world that has lost Gulf throughput; margins stay elevated as long as the strait is contested. We are not setting numeric target or stop because we do not know the fill price; the desk's rule is a mental stop 15% below fill and a re-evaluation if diesel falls back below roughly $5.

Horizon:

3–6 months

What would prove it wrong:

A verified reopening of the Strait of Hormuz to normal commercial traffic together with US retail diesel falling below $5.00/gallon; or MPC's reported refining margin per barrel collapsing back toward pre-blockade levels in its next quarterly report.

Fill:

bought 19 @ $402.38 (8% of the book)

The lesson: In a supply shock, the refiner (who buys crude and sells fuel) can earn more than the oil producer, because what blows out is the spread between the two — the 'crack spread' — not just the price of crude. Second-order effects often pay better than the headline.

New federal reports put the summer Iran war at $38bn and disclosed a US munitions shortfall (FT, Sept 16) — the same week the US confirmed deploying space weapons for the first time, NATO shot down a suspected Russian drone over Lithuania, and Putin warned European troops in Ukraine would mean 'war against Russia.' Restocking precision munitions is a multi-year procurement decision that survives any ceasefire; the paper's 'Ceasefire Fragility' pattern has shown that peace headlines move markets for days while stockpile orders run for years. Lockheed is a pure-play prime with heavy exposure to exactly the interceptor and strike missiles a Gulf war consumes, without the commercial-aviation exposure that the Hormuz cascade is hurting at diversified peers. Higher rates matter less here than for growth names because the customer is the US Treasury and the backlog is contracted. No numeric stop because the fill price is unknown; the desk will treat a 12% decline from fill as a signal to re-check the thesis rather than an automatic exit.

Horizon:

6–12 months

What would prove it wrong:

A US defense budget request or continuing resolution that cuts munitions procurement year-over-year despite the disclosed shortfall; or Lockheed reporting a declining backlog for two consecutive quarters.

Fill:

bought 13 @ $535.40 (7% of the book)

The lesson: Governments announce policy after the facts force it; when official reports quantify a shortfall, the spending decision is effectively made even before Congress votes. Buy the procurement cycle, not the news cycle.