Binomial interval, Cochran's Q
How big is home field in the KBO?
Every final this season, split by who batted last. The league-wide edge, each club's home-road gap, and whether those gaps differ by more than chance.
Computed on request from the record book (Oct 13, 2026) and 720 finals in the game log.
The finding
Home clubs have won 362 of the 703 decided games this season, a .515 winning percentage with a 95% interval of .478–.552. The interval includes .500, so 703 games are not enough to confirm a home edge on their own. The long-run MLB figure of about .540 sits inside that interval, so the KBO's edge could be just as large.
Club by club, the gap between home and road winning percentage runs from Doosan's +123 points to Lotte's −92. But each club's gap rests on about seventy games a side, and Cochran's Q finds no evidence they differ by more than that noise (p = 0.81). On this evidence, no club's home field is demonstrably special.
Home W%
.515
95% interval
.478–.552
p = 0.43
Home runs per game
4.96
visitors 5.18
Clubs differ? (Cochran's Q)
p = 0.81
Q 5.2, df 9
Home minus road winning percentage, by club
Each club's home-road gap with a 95% interval. The league row is the average gap a club would show from the league-wide home edge alone.
Home and road records side by side
Above the dashed line, a club has been better at home than on the road.
| Club | Home | Road | Gap | 95% | R/G home | R/G road |
|---|---|---|---|---|---|---|
| Doosan | 40–28 | 33–38 | +123 | −41 – +288 | 4.69 | 4.86 |
| SSG | 34–33 | 31–41 | +77 | −89 – +242 | 4.96 | 4.85 |
| Kiwoom | 28–43 | 22–47 | +76 | −83 – +234 | 3.42 | 4.44 |
| KIA | 42–29 | 37–34 | +70 | −93 – +233 | 5.10 | 5.59 |
| LG | 42–31 | 36–34 | +61 | −102 – +224 | 4.75 | 5.30 |
| Samsung | 43–29 | 40–29 | +18 | −145 – +180 | 5.55 | 5.65 |
| KT | 45–24 | 45–25 | +9 | −150 – +168 | 5.68 | 5.78 |
| Hanwha | 27–41 | 30–42 | −20 | −182 – +143 | 5.90 | 5.51 |
| NC | 32–41 | 32–37 | −25 | −189 – +138 | 4.68 | 5.31 |
| Lotte | 29–42 | 35–35 | −92 | −255 – +72 | 4.90 | 4.53 |
Method
All 720 regular-season finals; ties are excluded from winning percentages. The league interval is a Wilson score interval and the p-value tests against .500. Each club's gap is the difference of two proportions with a normal-approximation interval. Cochran's Q weights each gap by the inverse of its variance and asks whether they scatter around their pooled value more than their own standard errors allow:
Q = Σ wᵢ (dᵢ − d̄)², wᵢ = 1 ÷ SE(dᵢ)², Q ~ χ²(clubs − 1)
Run margin per game, home minus visitor: −0.22 (95% −0.58–+0.14).
Home clubs do not bat in the bottom of the ninth when they lead, so runs per game understate home offence; winning percentage is the cleaner measure. A club's home-road gap also folds in a roster built for its own park, and LG and Doosan share Jamsil, so their home games are less of a home edge than anyone else's. The MLB comparison is a long-run average quoted for orientation, not computed here.