Capital follows protection, not potential

Condensed from an Idiotopedia essay of 29 July 2026, a preview of Mad Mad World: The Backroom Economy, the final book in the Mad trilogy.

For about a year, Korea ran one of the strongest equity rallies anywhere. On the figures in a widely watched analysis at the time, the KOSPI rose close to 200 percent, Samsung's shares more than 500 percent and SK Hynix's more than 1,000 percent, until the two chipmakers made up 56 percent of the index. Then it unwound in 21 days. Around 1.2 million accounts received margin calls, brokers liquidated positions automatically, and each forced sale pushed prices low enough to trigger the next.

The easy reading is retail mania. It is the wrong one. The roughly 14 million small investors behind the rally were acting rationally inside a structure that left them few other routes. With the cost of living in Seoul beyond what a salary could close, the only leverage an ordinary investor could legally buy was a leveraged ETF or a margin line, both licensed, approved by risk committees and sold with fees attached.

Who holds the protection

The thread through the book is that capital follows protection, not potential. Institutions hold the protection: collateral, access to policy, first claim on liquidity, and committees that slow their exits. Retail holds the potential, which is a polite word for the risk that nobody upstream wanted to keep. Korea tested that at national scale and settled it in three weeks.

Three lessons for anyone allocating capital

Leverage turns a drawdown into a liquidation. Once margin debt, leveraged ETFs and short-dated options become the main source of flows, a market loses its ability to find a price on the way down. Forced selling drives prices lower whatever the fundamentals say, which forces more selling.

Concentration is a single risk sold as diversification. With two stocks at 56 percent of the KOSPI, every passive inflow was a bet on one thing: demand for high-bandwidth memory from US hyperscalers. The S&P 500's reliance on a handful of AI-linked companies has the same shape, and on the figures cited at the time, US margin debt stood near 4.5 percent of GDP, above the peaks of 2000, 2007 and 2021, before counting single-stock leveraged ETFs and same-day options.

Watch capital expenditure guidance, not end demand. The same analysis put returns on incremental AI spending at roughly 10 percent, down from about 40, while share prices went vertical. Valuations rose as the returns underneath them fell. As those returns approach the cost of capital, suppliers' earnings depend on the largest buyers choosing to keep spending, and one lower capex line on an earnings call can reprice a whole supply chain.

Closer to home

Indonesia has no pair of chipmakers dominating its index, but a few banks and conglomerate-linked names play the same role, some with free floats so thin that a single negotiated trade can set the price for the whole company. Millions of brokerage accounts have opened since the pandemic, mostly held by investors under 40 who take their cues from Telegram groups and promoters. Their leverage is harder to see than Korea's, spread across online lending apps, crypto accounts and family money. Leverage that goes unmeasured still has to be sold.

The difference is in how the market breaks. Korea's fell and was done in 21 days. Indonesia's auto-rejection limits mean the IHSG tends to queue rather than crash: limit down, no bids, and holders unable to exit for days. The loss is the same; it is paid more slowly. When trading was halted in March 2025, the response within a week was to loosen buyback rules, which made selling harder without asking why so many wanted to sell.

The question to ask first

Before holding an asset, ask who carries the risk if the exits close. It costs less than any model, and in markets built like these it is usually the more useful answer.