Gooaye EP679: This Selloff Isn't Fundamentals — It's Margin Calls

A whirlwind does not outlast the morning;
a cloudburst does not outlast the day.
— Laozi, chapter 23

What the episode is about

Gooaye (股癌) EP679, 2026-07-15 — Taiwan’s most-listened finance podcast. In one line: the recent global shakeout wasn’t caused by the Middle East or inflation, but by cascading margin calls on heavily-levered Korean retail investors; with momentum strategies in a timeout and sector rotation running hot, the host’s prescription is smaller positions, hard downside control, and attention on the segments with real shortages — IC substrates and OSAT packaging.

There’s also a sharp aside on generic AI tools hallucinating in specialist domains and quietly burning tokens — relevant to anyone who works with AI, not just investors.

The original show: search “股癌” (Gooaye) EP679 on any podcast platform (~1 hour, Mandarin). These are my listening notes and takes, not a transcript.

Highlights

• Who is selling matters more than what happened. Korean retail commonly stacks leverage past 10x; a routine pullback triggers forced liquidation → liquidity stampede → global contagion. Distinguishing “leverage being cleared” from “fundamentals deteriorating” changes the entire panic-or-buy decision.
• Taiwanese brokers tightened margin rules on smaller caps after the volatility and scattered settlement failures — an administrative second squeeze on market liquidity.
• Momentum is in a timeout. The buy-strength playbook that worked for six months is getting whipsawed. Money rotated out of power components and passives into wafers, equipment, and substrates; leadership is narrowing and rotation is accelerating.
• Real shortages are still real. Substrates and CCL have been scarce for over a year with rising quotes and order transfers; OSAT packaging lines (including MOSFET back-end) are maxed out with price hikes queued. The passive-components selloff is valuation compression, not a fundamentals break — lead times are still stretching.
• Names discussed: TSMC as the storm shelter and its earnings call as the range-break catalyst; Powerchip’s disclosure that its power-device packaging lines have long been full; Chipbond-style OSAT plays surging in the rotation; Walsin Tech’s drop framed as post-run correction with MLCC supply still tight.
• The AI aside: the host found generic AI models hallucinating confidently in a specialist domain (pricing rare whisky), burning tokens on re-verification; using AI blindly on contracts or regulatory work is “genuinely dangerous.”
• Two operating rules: cut decisively when key moving averages break with no bounce, raise cash, don’t re-live 2022’s round-trip; and when you fall in love with a researched stock, let the market correct your attitude — no traction after entry means the market is grading your thesis.

Where my thinking goes

1. Attribution before reaction. A 5% drop from broken fundamentals and a 5% drop from someone else’s margin call are different events. The first demands a thesis review; the second is a stress test of your position sizing — if other people’s liquidations can shake you out, your size was wrong, not the market.

2. Momentum timeouts are discipline exams. Everyone is a genius in a trending market. Fast-rotation chop is where risk control shows. “Smaller positions, guard the downside” is the same medicine as the position-limit school in our stop-loss piece: preserving capital preserves the ability to come back.

3. The cure for falling in love with a stock is institutionalized accountability. Deep research breeds attachment — that’s human nature, not a flaw. So the fix isn’t “don’t get attached”; it’s a mechanism you can’t weasel out of: write down the entry thesis and its kill conditions, then grade yourself publicly when they expire.

How we apply this

Three direct hooks into our own work, noted for the record:

1. Attribute volatility before acting on it. Leverage-clearing selloffs trigger a position-size audit; fundamental selloffs trigger a thesis re-run. Two different SOPs — mixing them means doing the wrong thing at the wrong time.
2. Book “stories” and “shortages” on separate ledgers. Optical is hunting for new narratives; substrates and OSAT are physically short. One is flow behavior, the other is industrial fact. One ledger for both and you’ll dump facts when stories fade, or trust stories before facts deliver.
3. AI hallucination is a cost line, not an anecdote. Confident-but-wrong answers are the most expensive kind; verification time and compute must be budgeted upfront in any serious AI-assisted workflow — a lesson we re-learn weekly.

References

• The show: Gooaye (股癌) EP679 on Spotify / Apple Podcasts / SoundOn (Mandarin)
• On deleveraging cascades: post-mortems of the 2021 Archegos unwind — same mechanics, different scale
• Our related piece: Two Stop-Loss Philosophies

---

Listening notes and educational discussion, not investment advice. Stocks appear only as episode context — no recommendations. Quoted content belongs to the original show; go listen to the source.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.