Handicap and komi

How much points komi needs to be to make handicap games back to even?

For example, with 9 stones, does white need 100 points komi for it to be even game? 110?


Depends on the level of players and the style of game they play, but each stone being worth twice komi is a good first approximation (they probably aren’t linear). So 9 × 13 = 117. When I’m giving 9 stones and reverse komi to DDKs I give 10 points extra per rank which leads to fairly close games, which is a bit stingy compared to twice komi per stone, but also I’m not playing full power to murder them. You can ask KataGo how much it thinks each stone handicap is, but bear in mind it is answering a different question to how much komi you and your 5k friend need to make a 9 stone handicap game 50-50 win.


I’ve seen estimates from a bit more than 110 to slightly less than 150, iirc.

I don’t think I’ve ever seen a value less than 110 claimed.

@gennan didn’t you make a post on L19 about this?

Yes, I did: katago Handicap games • Life In 19x19

At that time, up to about 16 stones handicap, KataGo’s evaluation for the point value of handicap (under Japanese rules) was close to

points value = (H - 0.5) * 14

This formula gives 119 points advantage for 9 stones handicap, while KataGo 1.3.1 estimated 9 stones handicap to be a 115 point advantage.

That post is already 18 months old, so newer versions of KataGo may give slightly different values.


In may 2020 I did it again, and at that time my results were closer to

points value = (H - 0.5) * 13

1 Like

Nick Wedd ventured a number in 1991, in BGJ #82 (Komi, pg. 18).

I don’t know what ruleset was being used by the BGA in 1990 – they originally used Japanese rules and later switched to AGA. His komi estimate looks like an area scoring value.

… a move early in the game is worth just over 15 points, and so (…) komi should be 7.5 or 8 points. (…) If I play a 6 dan, I start with 9 handicap stones, worth on the above argument about 130 points

This quote is not entirely in context, though. He goes on to say that Kageyama suggests a lower value of komi.