Strategy

ETF discount to NAV

An ETF's price usually sits within a fraction of a percent of the value of what it holds (its NAV). When one trades at an unusually deep discount, market makers have tended to close the gap within days.

Rule: Discount to NAV 0.3%+ and 2+ standard deviations below its own 60-day average (Korean equity ETFs with ₩100B+) → buy the ETF against the index; exit when the gap closes or after 5 days. Premiums the other way.

Out-of-sample = from Dec 22, 2025: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Leveraged ETF rebalance fade

Leveraged and inverse ETFs must buy after up days and sell after down days at the close. When their estimated trade is large next to a stock's normal volume, the closing push has sometimes faded over the next days.

Rule: Estimated rebalance trade above the notable share of the stock's 20-day trading value → the opposite way against the index for 2 days (short legs through stock futures where they exist).

Not enough history yet: 70 trading days of data and 9,566 trades. The test fills in as data builds up.

Out-of-sample = from Sep 28, 2026: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Futures basis convergence

KOSPI 200 futures normally trade a little above or below the index by their financing and dividends. When the front future gets unusually cheap, the gap has tended to close before expiry.

Rule: Front-month basis 2+ standard deviations below its 60-day average → buy the future, sell KODEX 200; exit near normal, after 15 days or before the roll.

Out-of-sample = from May 19, 2025: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Stock futures implied borrow

A stock future priced well below the stock (after financing) means borrowing the stock is expensive or a big dividend is expected. Heavy borrow demand often comes with crowded short sellers.

Rule: Nearest stock future implies a rate 8%+ a year below the CD rate (10+ days to expiry) → short the stock through its futures against the index for 20 days. Dividend season can trigger this too.

Out-of-sample = from Dec 24, 2025: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Futures roll spread

Before each quarterly expiry, positions move from the front month to the next. When the spread between them strays from its fair value (financing less dividends), the roll has sometimes pulled it back.

Rule: 5 trading days before expiry, next-minus-front spread 0.3+ points from fair value → buy the cheap month, sell the rich one; exit the day before expiry.

Out-of-sample = from Apr 18, 2025: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Expiry pin to max pain

Some traders believe the index drifts toward the level where option holders collect the least (max pain) into a monthly expiry. This tests it.

Rule: 3 trading days before a monthly expiry, the index 1.5%+ away from max pain → the front future toward max pain until expiry (one-sided).

Out-of-sample = from Jun 5, 2026: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.

Volatility spikes

When the fear gauge (VKOSPI) jumps far above normal, the index has sometimes recovered over the following weeks as volatility calmed down.

Rule: VKOSPI 2+ standard deviations above its 60-day average and above 20 → buy KODEX 200 for 20 days (one-sided).

Out-of-sample = from May 19, 2025: the last 30% of the history, which the rules weren't tuned on.

Total, worst drop and the curve treat each trade as 2% of the book with at most 20 open at once (signals beyond that are skipped). Average and profitable count every trade.