Preferred vs common shares

Why this can work

Many Korean companies have preferred shares that trade at a discount to the common shares. When the gap gets unusually wide or narrow, it has tended to drift back.

What can go wrong

The gap can stay wide for years (a preferred share's discount reflects voting rights and liquidity), and thin preferred shares cost more to trade than the backtest's averages suggest.

Rule

Price ratio 2+ standard deviations from its 1-year average → buy the cheap one, short the other; exit near the average or after 60 days.

Backtest

At ₩1억 a trade, Nov 1, 2021 to Oct 30, 2026.

Only 0 trades so far: treat these numbers with care.

Capacity: No edge after costs
Avg costs: —
Avg fill days: —
Turnover a year: —

1003 signals → 0 trades (1002 left out by filters, 0 over the position limit, 0 still open, 0 too big to fill, 0 exited at delisting).

Equity curve (% of book)

Shaded: out-of-sample, from Apr 30, 2025.

Trade returns

How this is calculated

Each trade pays commission (your setting), half the typical spread (half a price tick) on the way in and out, the securities transaction tax when a stock is sold or sold short, and borrow fees for a short stock (by how easy it is to borrow). Index futures pay a small fee and no tax or borrow.
Price impact uses the square-root model: impact = k × daily volatility × √(order ÷ 20-day average trading value), k = 1.0 by default, on the way in and out, for each leg.
Days to fill = trade size ÷ (average trading value × your participation %). Orders that take more than a day are spread over those days at each day's typical price (high + low + close ÷ 3), in and out.
Capacity is the trade size at which the setup's average return after costs falls to zero, if each trade were done within a day.
Everything is point in time: filters use only what was known on the entry day, financial statements count from their reporting deadline, and delisted stocks stay in (a stock that stops trading is exited at its last price).
Short selling was banned in Korea (all stocks Nov 2023–Mar 2025, most stocks from Mar 2020). During bans a short stock leg moves to index futures when futures are your hedge; otherwise the trade is skipped. Each setup says how many.
In-sample is the first 70% of the history, out-of-sample the last 30%. Walk-forward trades each half-year only if the setup made money over the previous two years.
Returns are per trade, as a % of the trade size. Sharpe is per trade, scaled by how often the setup trades. Drawdown and the equity curve are in % of your book.

By year

Year	Trades	Profitable	Avg net

Walk-forward

Each half-year is traded only if the previous two years were positive.

From	Traded	Avg net
Nov 1, 2023	No	—
May 1, 2024	No	—
Oct 30, 2024	No	—
Apr 30, 2025	No	—
Oct 29, 2025	No	—
Apr 29, 2026	No	—
Oct 28, 2026	No	—