nyasekai

Graveyard / Basis and spreads / Binance perps

Kimchi premium trading

Can you trade Bitcoin using the price gap between Korean exchanges and the rest of the world?

Family
Basis and spreads
Verdict
Near miss
Cause
No edge
Buried
10 October 2026

The idea

Korean exchanges often price coins above the rest of the world (the kimchi premium). A sudden jump or drop in that premium can signal retail enthusiasm or panic, so trade Bitcoin on Binance when it moves.

How we tested it

DataTest 1: the average Upbit premium of 6 coins, a shock of ±2 percentage points, trade the BTC perpetual and exit when the premium normalises. Test 2: the same pooled over BTC, ETH and XRP perpetuals with a funding-rate regime filter. Test 3: per-coin relative premium shocks of ±3 points across 219 Upbit coins, fade and continuation.
PeriodOctober 2023 to October 2026 (tests 1 and 2).
CostsNot stated.
HoldoutIn-sample and out-of-sample Sharpe. The out-of-sample slice had to contain at least 20 trades.
ControlNone stated.

Result

Test 1: Sharpe full / in-sample0.88 / 0.70
Test 1: profit factor full / in-sample1.32 / 1.24
Test 1: out-of-sample trades19 (20 required)
Test 2: filtered Sharpe full / in-sample0.62 / 0.46
Test 2: out-of-sample Sharpe, filtered vs unfiltered1.46 vs 2.17
Test 3: fade Sharpe−1.05
Test 3: continuation Sharpe−2.58

Why it failed

The one promising version missed the minimum trade count by one trade, and the regime filter that was supposed to improve it made it worse. When pooled across coins the edge turned out to be BTC only, short only, and only in choppy markets. The per-coin version lost in both directions and in every regime.

What would make it worth retesting

A fresh pre-registered test with more trades (ETH or XRP perpetuals, or 10-minute Upbit data) and a different regime filter.

Past results of a backtest say nothing certain about the future.