Wednesday, 14 October 2026 · London Edition · 07:30 London

Korea's free fall is the memory trade's first real crack.

Signals

Korean equities

Korea's market is in free fall, and the fear is specific: skepticism about semiconductor profitability. EWY fell 2.1% last session and 9.3% on the week, with the leveraged KORU down 26.6% over five sessions — a wipeout in a market that is effectively a leveraged bet on memory pricing. MarketWatch frames it as a sentiment-cycle problem rather than a hard-fundamentals one, which is the point: when a market's beta to chips turns this fast, positioning does the damage before any earnings revision does. The next test is any memory-pricing commentary out of SK hynix or Samsung.

EWY

Sell South Korea equities — MarketWatch says the market is in free fall on faltering semiconductor-profitability expectations, and the tape agrees — EWY is down 9.3% in a week and 25% below its 52-week high.

$166.6 -2.09%
KORU

Sell Leveraged Korea — The 3x wrapper has already given up 26.6% in a week and sits 77% below its 52-week high — leverage on a sentiment-driven selloff is the wrong side of the boat.

$14.72 -6.48%

Memory chips

One Chinese plan gutted the memory complex: CXMT says it will build a more powerful memory chip by end-2026, and the tape did the rest. CNBC names CXMT as the trigger — Micron fell 3.6% last session and 12.4% on the week, Western Digital dropped over 5%, SanDisk 4.6%, with SK hynix and Seagate both off more than 5%. The selling is indiscriminate across the whole group, which is itself the tell: this is a headline reaction, not a differentiated view on pricing. Micron trades at 4.6x forward earnings, 24% below its 52-week high — either the market knows something the multiple doesn't, or this is an air pocket. Watch for any CXMT capacity or HBM-timing confirmation.

MU

Sell Micron — CNBC pins the 3.7% session drop on CXMT's new-chip plan, but at 4.6x forward earnings and 12.4% down in a week, you're shorting a cheap multiple into a headline — low conviction by construction.

$953.2 -3.62%
WDC

Sell Western Digital — WDC fell over 5% last session on the same CXMT fear, but it's already 52% below its 52-week high with a 6.1% weekly loss — most of the downside is priced.

$380.7 -1.61%

Bank earnings

The banks delivered. Goldman printed $18.68bn of Q3 revenue against a $16.71bn estimate, JPMorgan raised its net interest income outlook, and UnitedHealth beat on both lines with $4.69 adjusted EPS on $112.69bn of revenue — a trifecta that says credit and the consumer are holding. The catches are smaller: JPMorgan's headline revenue and earnings still missed, and Wells Fargo beat on EPS but missed on revenue at $21.87bn. Direction is clear — the money-center complex is earning through a high-money-cost world. That is the opposite signal to the memory complex cracking in the same session.

GS

Buy Goldman Sachs — CNBC reports revenue beat by roughly $2bn on a $18.68bn print, driving a 2.8% premarket gain — a specific, dated beat that justifies the 12.3x forward P/E.

$891.9 -0.23%
JPM

Buy JPMorgan — Shares rose nearly 1% on an upgraded net interest income outlook, but the headline top-line miss caps conviction at medium despite the 13.2x forward multiple.

$331.4 -0.86%
UNH

Buy UnitedHealth — CNBC says adjusted EPS of $4.69 beat $4.14 and revenue hit $112.69bn, lifting the stock 2% — a clean two-line beat with the shares still 18% below the 52-week high.

$377.8 -0.20%

European gas

European gas hit an Iran-wartime high, and the drivers are stacked: a Washington-Kyiv standoff over Russian diesel, the approach of cold weather, and storage sitting unusually low for mid-October. WSJ and FT both flag the near-four-year price level, and fear is building of a global scramble for LNG cargoes. Cheniere is the obvious winner — LNG sits 5% below its 52-week high and is up 45% YTD, so the trade is working but not yet euphoric. A warm November would deflate the storage story as fast as the cold inflated it; watch weekly EU storage prints.

LNG

Buy Cheniere Energy — WSJ and FT both point to low storage and a scramble for cargoes, and Cheniere is 5% off its 52-week high, up 45% YTD, with a 13.8x forward multiple that still leaves room.

$286.9 +1.57%
UNG

Buy US natural gas — A global LNG scramble pulls US Henry Hub higher too, and UNG jumped 4.5% last session but is still down 5.4% YTD — headline-only sourcing keeps conviction low.

$11.41 +4.49%
USO

Buy Crude oil — Middle East attacks that threaten gas supply also bid crude, and USO added 6.4% on the week, though at 122% YTD it is only 6% from its 52-week high — late in the move.

$153.1 +1.36%

Long-end bonds

Two credible bond stories point opposite ways, and we're marking it a watch. The world's biggest fund manager has reportedly turned bullish on duration for the first time in six years — a six-year flip is a structural signal. Against that, a MarketWatch essay argues the household shift from housing into stocks is raising the cost of money, which is bearish the long end. TLT sits 3% above its 52-week low and down 10% YTD, so the bear case is priced while the bull case is not. When the press is this split and positioning this one-sided, you don't force a call.

IEF

Buy 7-10Y Treasuries — A broad turn in the bond view favours intermediate duration too, and IEF is 1% above its 52-week low with only a headline-level rationale behind the flip.

$89.54 +0.12%
TLT

Watch Long Treasuries — MarketWatch reports the No. 1 fund manager flipped bullish on bonds for the first time in six years, while a companion piece argues rising money costs keep long yields high — TLT is 3% above its 52-week low, so the bear case is already paid for.

$78.37 +0.41%

AI power / nuclear

Upgrading existing US reactors is the fast, cheap route to more nuclear power — and that capacity is running out, per Bloomberg. The implication is subtle: if you can't uprate your way to supply, the next marginal megawatt has to come from new-build, enrichment and more uranium demand. The whole chain is de-rated — CCJ sits 36% below its 52-week high, CEG 28% below, and NLR is 1% off a 52-week low, down 25% YTD. Scarcity of the cheap option is a tailwind for the expensive one. Watch for DOE uprate approvals; they'd mark the official end of the easy capacity.

CCJ

Buy Cameco — Bloomberg says cheap reactor upgrades are being used up, pushing demand toward fresh uranium-fuelled capacity, and CCJ is 36% below its 52-week high — though the piece carries no figures.

$86.15 +0.03%
CEG

Buy Constellation Energy — Existing reactor owners gain pricing power as the cheap upgrade option depletes, and CEG is 28% below its 52-week high, down 19% YTD, at a 22x forward multiple.

$295.8 +0.65%
NLR

Buy Nuclear fuels — Scarce uprate capacity lifts the whole nuclear chain longer-term, and NLR is 1% off a 52-week low, down 25.3% YTD — but the sourcing is a single bodyless Bloomberg newsletter.

$99.38 -0.82%

China IPOs

Hong Kong's SFC opened its biggest case ever — alleged IPO-rigging schemes involving over $590m in client assets — and suspended Sichuan Baicha Baidao, the operator behind the ChaPanda bubble-tea chain. Nikkei Asia has the figure the headlines leave out and names the exact entity. The read-through cuts both ways: tighter vetting protects the franchise but chills the listing pipeline that has been Hong Kong's best story this year. HKEX is the direct exposure, and a beaten-down FXI (down 14.2% YTD, 17% below its 52-week high) picks up one more regulatory overhang on a stack that's already heavy.

0388.HK

Sell HKEX — A $590m rigging probe can slow the IPO pipeline that drives HKEX revenue, but the link is inferred rather than reported — single-source and no HKEX-specific numbers.

FXI

Sell China large-cap — Tighter Hong Kong listing scrutiny adds to an existing regulatory overhang, and FXI is already down 14.2% YTD at 0.85x book — the trade is a marginal add to a crowded short.

$34.17 +0.09%
2555.HK

Watch Sichuan Baicha Baidao — Nikkei Asia says shares are suspended amid the SFC probe, so there's no trade in the stock itself — the signal is delisting and reputational risk for the listing cohort.

China sportswear

Nike's China revenue fell 26% in the latest quarter, and the NikeLab Shanghai flagship closed last week after a 12-year run — a brand retreating from a market it once owned. Nikkei Asia pairs both facts with the winners: Anta and Xtep are taking share, and Anta is now a top shareholder in Puma, importing its direct-sales playbook. NKE sits 50% below its 52-week high, down 44% YTD — the market has marked the damage, but a 26% China decline says the bottom isn't in. Puma at 88x forward earnings is priced for a turnaround that hasn't started.

2020.HK

Buy Anta Sports — Nikkei Asia names Anta as a share taker from Nike and now a top shareholder in Puma, but the piece gives no Anta revenue or margin figures — inference, not a quantifiable lift.

NKE

Sell Nike — Nikkei Asia reports a 26% China revenue decline and the NikeLab Shanghai closure, and the stock is 50% below its 52-week high, down 44% YTD — a specific, dated share-loss signal, if a well-trodden one.

$35.33 -0.06%
PUM.DE

Watch Puma — Puma is adopting Anta's direct-sales playbook, but at 88x forward earnings the multiple leaves no room for execution risk — the turnaround is fully priced.

€23.41 +0.99%

Korea crypto rules

South Korea blocked 27 unregistered exchange websites — KuCoin, BitMart and MEXC among them — after Korea Communications Standards Commission reviews on Oct 6 and 7. Only 28 virtual asset service providers are actually registered with the FIU, and 44 unregistered operators have been referred to investigators. The tell is behavioral: an estimated $110bn left South Korean domestic exchanges in 2025, so demand doesn't disappear when venues do — it moves. That's a quiet tailwind for compliant, listed venues. Coinbase sits 51% below its 52-week high and down 23% YTD.

COIN

Buy Coinbase — CoinDesk says Korea blocked 27 unregistered venues and the $110bn 2025 offshore outflow shows demand persists, but Coinbase is a US venue with only indirect exposure to the shift.

$181.8 +0.78%

Passive investing

A Bloomberg column by Merryn Somerset Webb and John Stepek argues passive investing is not the neutral default it looks like — a plain S&P 500 fund is a concentrated bet on US mega-cap leadership repeating the last decade. SPY sits 1% below its 52-week high, so the market is priced for exactly that. The implied fix is to spread out: equal-weight RSP captures the US without the mega-cap tilt, and emerging markets are the direct alternative. There's no data, no flow figures, nothing dated — this is an argument, not a signal, hence the low conviction.

RSP

Buy Equal-weight S&P 500 — The column frames indexing as a concentrated mega-cap bet, and RSP strips that out — up 10.6% YTD and 5% below its 52-week high with no cap-weight dependency.

$213.3 +0.05%
EEM

Buy Emerging markets — A next-decade handoff from US to cheaper markets would benefit emerging equities most, and EEM trades at 14.1x trailing earnings, 8% below its 52-week high — but this rests on an opinion column with no valuation data.

$65.74 -0.62%

Prediction markets

Kalshi's top inflation trader is a former theater director working out of his garage who, per Bloomberg, is beating Wall Street banks at their own macro game. The colour is fun; the signal is that event contracts are becoming a real venue for macro risk rather than a novelty. That's a slow structural tailwind for brokers who plug into prediction markets. Interactive Brokers, up 30% YTD and 11% below its 52-week high at 27x forward earnings, has the retail plumbing to capture flow migrating off bank desks. No volume or P&L data in the piece, so conviction is low.

IBKR

Buy Interactive Brokers — Bloomberg's Kalshi profile signals retail macro flow migrating away from bank desks, and IBKR — up 30% YTD — has the plumbing, but the article offers no volume or P&L data to size it.

$87.49 -0.46%

Most original take

Merryn Somerset Webb and John Stepek · Bloomberg Markets · 14 Oct 2026

US Stocks Won the Last Decade. What About the Next One?

Merryn Somerset Webb and John Stepek argue that passive investing is not the neutral choice investors assume. A plain S&P 500 index fund is, in effect, an active and concentrated wager that US mega-cap leadership repeats the last decade. With SPY 1% off its 52-week high, that wager is already priced and crowded. The implied fix is deliberate spread — equal-weight exposure or non-US markets — rather than defaulting to the index. It's an argument, not a data-driven signal, but it reframes the most common portfolio decision there is.

Read original ↗

Our view

Today's tape split along a clean line. Korea is down 9.3% on the week via EWY, its leveraged proxy KORU down 26.6%, Micron down 12.4% and Western Digital down 6.1% — all triggered by one Chinese announcement that CXMT will build a better memory chip by late 2026. At the same time Goldman, JPMorgan and UnitedHealth beat, and US gas exporters are bid as European benchmarks print an Iran-wartime high, with Cheniere 5% off its 52-week high. The money that spent two years treating 'semis go up' as a law of nature is being tested, and it's rotating rather than fleeing.

The bond market is the day's second split. TLT is 3% above its 52-week low and down 10% YTD — yet the world's biggest fund manager has reportedly turned bullish on duration for the first time in six years. A MarketWatch essay argues the opposite: household wealth has rotated from homes into stocks, and that concentration is raising the cost of money, which keeps long yields high. Two credible views, opposite directions, one bond. The bear case is priced; the six-year flip is not.

The case against all of this: the memory selloff rests on a plan, not a product. CXMT said it will build a chip by end-2026; nothing has shipped. Micron trades at 4.6x forward earnings, Korea has sold off at this pace before and bounced hard, and one slipped quarter from CXMT unwinds the whole short violently. You also cannot call this risk-off when JPMorgan lifted its NII outlook and UnitedHealth beat on both lines. Our semiconductor caution is a positioning call, not a fundamental one — and if the equal-weight RSP, up 10.6% YTD and 5% off its high, keeps grinding, it's just a rotation, not a top.

What's missing from every version of this story: nobody connects CXMT to the actual memory demand curve. The press still frames memory as a commodity cycle when high-bandwidth memory is now an AI supply-chain bottleneck — and that distinction decides whether today is a dip or a top. That is why the cleanest expression isn't a single ticker. It's dispersion: long the banks and the gas complex earning through a high-rate world, funded out of the crowded semiconductor beta everyone still holds. Watch whether the selloff stays contained to Korea and memory, or starts pulling SOXX off its own highs.

Yesterday's signals, today

From the London Edition on 13 Oct 2026 — 3/5 signals moved in the predicted direction.

Share this edition