Name: How to Be Right and Lose Everything
Uploaded: 2026-07-18T13:15:06+00:00
Duration: 2347 s

Speakers

Patrick Boyle — Host / Hedge Fund Manager and Finance Professor

Summary

The video examines South Korea's KOSPI as both the world's best-performing major market and a severe bear market. It attributes the boom-bust to extreme index concentration in Samsung and SK Hynix, single-stock leveraged ETFs, and aggressive retail leverage. It also explains how Korea's weak won, AI capex dependence, and Victor Haghani's coin-flip experiment illustrate being right on a trend but losing through bad bet sizing.

KOSPI soared on Samsung and SK Hynix, then fell about 27% from its June peak.
The index became a concentrated, leveraged AI-memory bet with two stocks near 60% weighting.
Single-stock leveraged ETFs created mechanical buy/sell feedback and forced selling at the close.
More than one million Korean retail accounts faced margin calls, with many wiped out.
The won is weak despite a record surplus because export dollars stay offshore and retail buys US stocks.
Korea's market depends heavily on US hyperscaler AI capex, and TSMC's capex/margin warning raised AI ROI concerns.
Victor Haghani's biased-coin experiment illustrates how excessive bet sizing can ruin a positive-edge strategy.

Ideas

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 0:55

Concentrated, leveraged KOSPI fragile despite AI boom.

South Korea's KOSPI became the world's best-performing major market, but it is also in a severe drawdown because the index became an extremely concentrated, leveraged bet on Samsung Electronics and SK Hynix. Those two stocks reached nearly 60% of the KOSPI, single-stock leveraged ETFs create mechanical pro-cyclical flows, and foreign diversified funds can be forced to sell if weights rise further. The market is therefore fragile to any slowdown in AI memory demand or a reversal in leveraged flows, making it dangerous to own despite the underlying companies' real earnings.

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 14:50

Leveraged single-stock ETFs amplify forced selling.

Single-stock leveraged ETFs tied to Samsung and SK Hynix promise twice the daily return but must reset exposure every day, buying into strength and selling into weakness. This short-gamma feedback loop amplified volatility, and on turbulent days rebalancing became a huge share of SK Hynix trading. Banks hedge via total return swaps and cliquets, and rising insurance costs plus regulator freezes and account requirements show the structure is fragile. Retail investors suffered mass margin calls and forced liquidations.

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 27:23

Won weak despite record trade surplus.

Korea runs a record current-account surplus, yet the won is one of Asia's weakest currencies. Samsung and SK Hynix leave export dollars offshore for overseas expansion, while Korean retail investors buy US tech stocks, both reducing demand for won. The weak won imports inflation and forces the Bank of Korea to raise rates into a falling stock market; Seoul's reluctance to allow a freely convertible offshore won due to 1997 IMF trauma keeps this structural. This supports a bearish view on the won.

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 31:10

Memory demand depends on hyperscaler capex.

Samsung Electronics and SK Hynix are highly profitable, world-class memory suppliers at the center of the AI boom, controlling much of global HBM supply. But their success rests almost entirely on the capex budgets of four US hyperscalers, and memory chips are a viciously cyclical industrial commodity where prices can fall off a cliff if demand softens. If AI monetization disappoints and hyperscalers slow spending, these stocks would be exposed to a severe downcycle, with leveraged ETF flows amplifying the downside.

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 31:27

AI infrastructure may earn poor returns.

AI infrastructure has been heavily subsidized: frontier labs charged flat monthly fees often below cost, but investors now want returns, pushing usage-based pricing that may reduce usage. If AI becomes more like commodity electricity, value may flow to users rather than the model builders and infrastructure funders; hyperscalers could capture little return and respond by delaying data-center spending. That makes hyperscaler capex a key monitorable risk for AI infrastructure and memory demand.

Patrick Boyle Host / Hedge Fund Manager and Finance Professor 32:07

TSMC capex and margin pressure risk.

TSMC reported record profits and raised guidance, yet its shares fell because it warned of higher capex and margin pressure. The market reaction suggests investors are starting to question the cost of staying ahead and the ultimate return on AI infrastructure spending, rather than just rewarding growth. This is a warning setup for TSMC and AI-infrastructure names.

How to Be Right and Lose Everything

Summary

Ideas

Up Next