Back-test KOSPI rotation strategies against the actual order book.

Most retail-grade back-testing tools model KOSPI trades at mid-price, with no regard for KRX lot-size constraints. A KOSPI 200 constituent trade at the wrong lot tier adds 20–60 basis points of slippage per fill — invisible in a simplified model, material in live execution. A rotation strategy rebalancing quarterly across 50 names accumulates this error every cycle. After four rebalances, a strategy that read as profitable on paper has absorbed hundreds of basis points in unmodeled friction.

The KRX dividend calendar introduces a second layer of systematic error. Korean companies concentrate ex-dividend dates in the last two weeks of December and first week of January — a structural feature of domestic fiscal year-end conventions and the KRX T+2 settlement cycle. A rotation signal that ignores 배당락 (배당락일, ex-date) clustering sells into price-adjusted closes at the wrong moment, distorting December and January return attribution by 30–80 bps per year.

The result: strategies that appear to generate alpha in back-test consistently underperform after live deployment on 한국거래소 (KRX). The gap is not random noise — it is a systematic artifact of models that apply US equity assumptions to Korean market structure. Finology was built specifically to close this gap.