Friday, 9 October 2026 · New York Edition · 09:00 New York

Refiners print money. AI's IPO window just shut.

Signals

⚡ Convergence radar: Sell MGH.AX×3Watch NVDA×3

Oil refiners

WSJ says Valero, Marathon and Phillips 66 are set to crush the near-record June quarter because wars are shrinking global energy supply — and the tape agrees, with VLO and MPC up 4.6% and 4.8% in the prior session and all three within 1% of 52-week highs. The same paper's commodity desk puts Brent's September risk premium at about $22 a barrel, the second-highest monthly reading Goldman has on record. That's the tension: 9-10x forward multiples look cheap, but they're cheap on peak margins, so one ceasefire headline hits earnings and the multiple together.

VLO

Buy Valero Energy — WSJ names Valero first in the blowout-refiner trio; the stock gained 4.65% in the prior session and sits 1% from a 52-week high at 10.3x forward, so the call is peak-cycle cheap rather than distressed cheap.

“The earnings of independent fuel makers Valero Energy, Marathon Petroleum and Phillips 66 are set to crush the near-record results of the June quarter.”

$443.8 +4.65%
MPC

Buy Marathon Petroleum — Second name in the WSJ trio with the best week of the three at +9.7% and 1% off its 52-week high; 9.0x forward is the cheapest refiner and the most exposed to a crack-spread reversal.

“The earnings of independent fuel makers Valero Energy, Marathon Petroleum and Phillips 66 are set to crush the near-record results of the June quarter.”

$463.3 +4.77%
PSX

Buy Phillips 66 — Third WSJ name, printed at a 52-week high in the prior session and +115.7% YTD — least upside of the three at 10.5x forward, but also the least leverage to a ceasefire.

“The earnings of independent fuel makers Valero Energy, Marathon Petroleum and Phillips 66 are set to crush the near-record results of the June quarter.”

$281.6 +3.67%
XLE

Buy Energy sector — Sector proxy for two WSJ pieces — the refiner earnings call plus a September Brent risk premium of about $22/bbl; XLE is 1% off its 52-week high and +42.9% YTD.

$65.24 +2.97%
BNO

Watch Brent crude — WSJ says Brent's risk premium hit about $22/bbl in September, the second-highest monthly level Goldman has on record, yet futures still edged lower — direction now rides on war headlines, not inventories.

$63.64 +2.81%

AI data centres

Firmus scrapped what would have been Australia's biggest IPO in almost 30 years — an A$7.9bn raise at A$11 a share implying an A$43.9bn valuation — after investors refused to pay up for 46MW built against 912MW contracted. Nikkei (twice), The Information and WSJ all carry the same withdrawal with the same quotes, so the news itself is priced; the disclosure that most contracted revenue sits on sites that don't exist yet is the part worth keeping. Nvidia's backing, Coatue's money and a $10bn Blackstone debt facility didn't clear the book. Firmus is now looking at a Nasdaq listing instead.

EWA

Hold Australia equities⚡ — Australia loses its marquee tech listing and keeps a bank-and-miner index; EWA is +6.6% YTD at 0.44x book, so this is sentiment, not earnings.

$28.23 +0.00%
MGH.AX

Sell Maas Group⚡ — MGH owns 3.2% of Firmus and holds about A$1.1bn of work orders with A$373m already paid; the stock was halted after a near-30% drop and is 34% lower over the week.

$4.63 -6.65%
NVDA

Watch Nvidia⚡ — A scrapped IPO doesn't touch chip demand — NVDA fell 2.94% in the prior session but sits 5% below its 52-week high at 14.5x forward — though it proves Nvidia's name on the cover no longer clears a book.

$230.5 -2.94%

Cyber defence

The Information reports South Korean banks were hacked with a Chinese AI agent, and separately that Anthropic is widening a cyber defence program — two single-source briefs with no named banks, no dollar figures and no confirmed attribution. The read-through runs both ways: AI is now the attack tool and the defence product, which supports security budgets whoever wins. The market isn't paying for it — CIBR and HACK were each marginally lower in the prior session and sit about 3% below 52-week highs, while CrowdStrike trades at 163x forward. A named victim or confirmed attribution turns this from a brief into a budget line.

CIBR

Buy Cybersecurity stocks — Two single-source The Information briefs argue for bigger security budgets, but CIBR sits 3% below its 52-week high after a flat prior session — the theme is real, the catalyst is thin.

$106.2 -0.35%
HACK

Buy Cybersecurity stocks — Same read-through as CIBR, but the +58.1% YTD run already prices steady budget growth; today's briefs add narrative, not numbers.

$125.3 -0.37%
KB

Sell KB Financial — Korean banks are the named victim class; KB fell 2.16% in the prior session and trades at 8.2x forward with no remediation cost disclosed.

$122.0 -2.16%
CRWD

Watch CrowdStrike — Anthropic is both partner and competitor to incumbents; at 163.8x forward and 8% below its 52-week high, CrowdStrike is priced as if AI never eats its lunch.

$263.0 -0.92%

Japan tourism

Nikkei reports Japan's tourist spending windfall is shrinking because Chinese visitor numbers have plunged — a single-source item with no spending totals or visitor counts published. Department stores carry the most direct leverage to that spend, airlines the route exposure. The index-level hit looks modest: tourism is meaningful but not dominant in Japanese GDP, and EWJ sits 3% below its 52-week high, up 19.7% YTD. The next monthly inbound print is the first hard number.

EWJ

Hold Japan equities — Tourism is meaningful but not dominant in Japanese GDP; EWJ is 3% below its 52-week high and +19.7% YTD, so the index isn't the trade.

$97.32 -1.02%
3099.T

Sell Isetan Mitsukoshi — Isetan is the most levered listed name to Chinese tourist spend in Japan, but Nikkei gives no spending total, so the size of the hit is unknown.

9201.T

Sell Japan Airlines — Fewer Chinese visitors means softer load factors on China-Japan routes; no route or capacity figures were published to size it.

Chip materials

Sumitomo Bakelite will add chip encapsulant capacity in China and Singapore, per Nikkei — a deliberate choice to build next to customers rather than at home. No capex figure, timeline or capacity number came with the item, so read it directionally: materials suppliers are adding capacity for expected volume growth in Asia's two main packaging hubs. The chip tape isn't confirming — SMH fell 2.84% and SOXX 3.35% in the prior session, leaving them 10% and 14% below 52-week highs after +63% and +80% YTD runs.

4203.T

Buy Sumitomo Bakelite — Nikkei says capacity is going up in China and Singapore, a direct volume signal for the encapsulant business, though no capex or timeline was disclosed.

SMH

Watch Semiconductor stocks — Materials capacity additions imply expected chip volume growth, but the tape disagrees — SMH fell 2.84% in the prior session and is 10% below its 52-week high.

$607.3 -2.84%

Critical minerals

Mitsubishi Materials is buying tungsten from Vietnam to cut its reliance on Chinese supply, a single Nikkei item with no volumes, prices or project timelines attached. Small deal, large template: every non-China route that gets built erodes Chinese producers' pricing power and feeds the diversification trade this ETF tracks. REMX hasn't been paid for any of it — down 21.1% YTD and 1% off its 52-week low, 46% below the high. The next project announcement tells you whether that's value or a value trap.

5711.T

Buy Mitsubishi Materials — Securing Vietnamese tungsten cuts the biggest supply risk in the cemented carbide business; no volumes or timelines were given to size the benefit.

REMX

Buy Rare earth miners — Every non-China critical-minerals route feeds this ETF's thesis, but REMX is down 21.1% YTD and 1% off its 52-week low — the market isn't paying for diversification yet.

$60.58 -2.10%

Vietnam offshore

Vietnam's offshore oil and gas push has pulled Japanese, US and Russian companies into the South China Sea — a combination rarely seen in one licensing round. No blocks, counterparties or contract values were disclosed, so what you're buying is a geopolitical option rather than cash flow. VNM is down 13.1% YTD and sits 2% above its 52-week low, which says the market is pricing confrontation risk, not reserves. A Chinese naval or diplomatic response is the tail that isn't in the price.

VNM

Buy Vietnam equities — Offshore development is a growth driver for Vietnam, but VNM is down 13.1% YTD and 2% above its 52-week low — the market has already priced the confrontation risk.

$16.61 -1.54%
1605.T

Buy Inpex — Inpex is among the Japanese players drawn into Vietnam's blocks, adding long-dated reserves with no contract values disclosed to underwrite the upside.

Mexico autos

Nissan will sell a China-built Frontier Pro pickup in Mexico, using idle Chinese capacity to undercut rivals in North America's light-truck market. Neither volumes, pricing nor a launch date were disclosed, so the hit to Ford and GM is unquantified — both sell profitable pickups in Mexico and both trade cheap at 6.3x and 5.4x forward. The bigger point is that Chinese plants are becoming the export base for Japanese badges into North America. Ford is 31% below its 52-week high; GM is 11% below after a +5.1% week.

7201.T

Buy Nissan — Using Chinese capacity for a Mexico-market Frontier Pro cuts cost and absorbs idle plant, a rare positive for the turnaround — but no volumes, pricing or launch date were given.

F

Sell Ford — Ford sells profitable pickups in Mexico and a cheaper import attacks that franchise; it trades at 6.3x forward and 31% below its 52-week high.

$12.25 +1.07%
GM

Sell General Motors — GM is a large Mexico pickup seller with the same pricing exposure; 5.4x forward after a +5.1% week leaves little cushion for a price war.

$82.25 +1.56%

Gold

Gold rose more than 1% as a softer dollar and lower oil supported the metal, with potential Chinese demand cited as the swing factor — one WSJ item with no price level or demand figure attached. Positioning is the story: GLD sits 26% below its 52-week high and is down 4.9% YTD, and the miners are in the same place, GDX 26% off its high. That's a market that stopped believing the gold story rather than one crowding into it, which is where Chinese buying headlines hurt shorts most. Newmont at 11.4x forward is the cheap way to own it.

GLD

Buy Gold — WSJ ties gold's 1%+ gain to a softer dollar, lower oil and potential Chinese demand; GLD is 26% below its 52-week high, so this is early positioning rather than a crowded trade.

$378.6 +0.73%
GDX

Buy Gold miners — Miners give leverage to the metal and sit 26% below their 52-week high with the gold price only 5% off its 52-week low — the leverage is still cheap.

$86.72 +1.47%
NEM

Buy Newmont — Newmont at 11.4x forward and 15% below its 52-week high is the cheapest large-cap way to own a Chinese-demand bid, with the stock up 1.77% in the prior session.

$115.5 +1.77%

European equities

The FT runs an American bull case on Europe: siloed national markets, anti-competitive culture and low investment have stunted growth — and that's precisely the opportunity. It's opinion, with no data, named companies or catalyst, so treat it as framing rather than a signal. The market has done nothing with it: VGK and EZU are both up just 1.0% YTD and sit 8-9% below 52-week highs after flat prior sessions. A reform headline is the trigger; the column is only the set-up.

VGK

Buy European equities — FT's bull case is that Europe's fragmentation is the fixable discount; VGK is up just 1.0% YTD and 8% below its 52-week high with no catalyst in the column.

$85.43 -0.15%
EZU

Buy Eurozone equities — Eurozone markets sit at the heart of the fragmentation the FT piece flags; EZU is flat YTD and 9% below its 52-week high with no reform headline attached.

$65.63 -0.41%

Grid buildout

A WSJ opinion piece argues federal Washington, not state utility commissions, is the real bottleneck to building transmission lines. It's an argument, not reporting — no projects or companies named — but if the diagnosis holds, the beneficiaries are grid contractors and the copper they bury. Neither has been paid: PWR is 13% below its 52-week high after a -2.25% prior session but +55.9% YTD at 34.7x forward, and CPER sits 5% off its high, +12.8% YTD. Permitting legislation is the catalyst.

PWR

Buy Quanta Services — If federal permitting is the real bottleneck, grid contractors are the beneficiary; PWR is 13% below its 52-week high but +55.9% YTD at 34.7x forward.

$685.3 -2.25%
CPER

Buy Copper — Transmission buildout is copper-intensive and CPER is 5% off its 52-week high, +12.8% YTD, with no permitting bill yet on the table.

$39.46 -0.88%

Fixed income

Two WSJ items land the same way: a wealth column telling investors to match bond strategy to purpose — income, ballast or duration — and a report that Arini lost 16% while LPs kept adding to its $7.3bn flagship on the view that rising rates create opportunity. Nobody is capitulating. The tape says duration has been the pain trade: TLT is 2% above its 52-week low and down 10.5% YTD, HYG down 4.4%. Short-dated ballast over long duration is the cleaner seat until the inflation reassessment resolves.

SHY

Buy Short Treasuries — Short Treasuries are the cash-like ballast the WSJ piece describes and the least painful seat in bonds — SHY is 2% below its 52-week high, down 2.0% YTD.

$81.20 +0.05%
HYG

Hold High yield credit — Credit is the income end of the WSJ framework but carries recession risk; HYG is down 4.4% YTD and 5% below its 52-week high, while Arini's 16% loss shows how fast rate moves punish credit beta.

$77.14 -0.05%
TLT

Watch Long-duration Treasuries — WSJ's framing piece says own duration for recession defence; TLT is 2% above its 52-week low and down 10.5% YTD, so the trade needs a growth scare that isn't in today's coverage.

$77.87 +0.94%

Most original take

Matt Wirz · WSJ Markets · 8 Oct 2026

Arini, Hedge Fund Known for Bold Bets, Loses 16%

Arini's flagship lost 16% and investors put more money in. That's the opposite of the post-drawdown flows script, and it says LPs are paying for rate-volatility exposure rather than the manager's last twelve months. Set against a TLT 2% off its 52-week low and a HYG down 4.4% YTD, it says professional money reads fixed-income pain as the opportunity. The press reported the loss; the inflows are the actual story.

Read original ↗

Our view

The tape split in two today. Physical scarcity is being paid for: Valero and Marathon each gained roughly 4.7% in the prior session and sit within 1% of 52-week highs, Brent carries a roughly $22-a-barrel war premium, and copper trades 5% off its high. Promised capacity is not: Firmus pulled a $5bn Australian listing because 46MW built against 912MW contracted didn't justify a $30bn valuation. Meanwhile the safest asset on the page is the one everybody hates — TLT at $77.87, 2% above its 52-week low, down 10.5% YTD.

The case against that read is positioning, not logic. Valero is +168% YTD and Marathon +180%, both at 9-10x forward earnings — which is what peak-cycle always looks like, and one ceasefire headline takes the margin and the multiple down together. Nor has the AI trade actually repriced: Nvidia is 14.5x forward and 5% below its 52-week high, SOXX fell 3.35% in the prior session yet is still +80% YTD, and nobody who funded Firmus's buildout took an equity hit this week. A dead IPO is price discipline, not a stopped cycle.

What's missing is the money. Firmus had a $10bn Blackstone debt facility behind a pipeline that is 95% unbuilt, and no one is asking who funds the rest now that public equity has said no. Same silence in Asia: Mitsubishi Materials moving tungsten sourcing to Vietnam, Vietnam's offshore blocks pulling in Japanese, US and Russian partners, Sumitomo Bakelite adding encapsulant capacity in China and Singapore — all moves away from Beijing, and not one line on how Beijing answers. That's the story with a 2027 date on it.

The cleanest expression isn't a ticker, it's long physical bottlenecks against short financial promises. Refining capacity, grid copper, non-China tungsten, gold that Chinese households keep buying: scarce, and unloved next to anything sold on a slide deck. The other side is long duration 2% off its low, cyber vendors at 82-163x forward, and pre-revenue data-centre equity. Size the first as a trend and the second as a trade, and let the next ceasefire headline tell you which one you actually own.

Yesterday's signals, today

From the New York Edition on 8 Oct 2026 — 2/7 signals moved in the predicted direction.

Today's scan: 248 articles considered · 18 surfaced · across 14 publishers

Generated Fri 9 Oct 2026 8:04am New York · 37,973 views

Bloomberg Markets · CNBC Investing · CNBC Markets · CoinDesk · FT Alphaville · FT Companies · FT Lex · FT Markets · MarketWatch Top · Nikkei Asia · SCMP Business · The Information · WSJ Business · WSJ Markets

Next edition: Weekend Edition tomorrow at 10:00 London.

Methodology
Share this edition