Does OGS give out handicaps for 19x19 the same way it gives them out for 9x9? For example, let’s say a 9k versed a 5k. The 9k would get four handicap stones on 19x19, but what about 9x9? I would see this as a huge advantage for white.
after i saw your question, i clicked on 9x9 games and see if there were any handi stones
and the result is that no matter how strong the other player is, no handicap stones were given
this is just based on what i saw
most likely is handicap stones can be given but more often than not handicap stones aren’t necessary
you can imagine you’re 1k giving 9 stones to 10k on 9x9, how do you win? lol
Handicap in 9x9 does exist, although many games are played without it(such as my game against a 6d).
I think I have given two stones to a 25k when I was a 20k, but I don’t remember much about this. They also sometimes give reduced komi instead.
In my childrens club I calculate handicap on 9x9 by dividing the rank gap by 6.
The system I’m familiar with says half the stones for 13x13 and a quarter for 9x9
I suppose you mean a quarter for 9x9, so 1 stone per 4 ranks.
I think that scheme is quite tough for white.
I saw a japanese pro talking about handicap for a smaller board size, specifically 13 x 13:
“13路では置き碁より逆コミの方が面白いゲームになると思っていまして” (translation: I believe, in 13 board, reverse komi makes for a more interesting game than on-board handicap.)
His article does not mention a method for calcing such reverse komi for 13 x 13 though.
For 19 x 19, the following article links to 2 ways to calcing reverse komi, one is a table made from game data, the other a formula using ogs rating.
In theory, someone versed with data mining could make a similar thing for smaller boards as well.
the following is a discussion about handicap vs reverse komi. there is an interesting comment wrt their effect on a game.
I asked KataGo to estimate the komi value of handicaps on different board sizes.
This is what it says (Japanese rules and standard placement of handicap stones):
h1 = 5.5 komi
h2 = 15 komi
h3 = 24 komi
h4 = 45 komi
h5 = 71 komi
so roughly 15 points per full handicap stone
h1 = 6.5 komi
h2 = 18.5 komi
h3 = 30 komi
h4 = 44 komi
h5 = 51 komi
h6 = 63 komi
so roughly 11 points per full handicap stone
h1 = 6.5 komi
h2 = 19 komi
h3 = 31 komi
h4 = 46 komi
h5 = 59 komi
h6 = 74 komi
h7 = 87 komi
h8 = 102 komi
h9 = 112 komi
so roughly 13 points per full handicap stone
Yes, I did ^^. I should stop posting when I’m sleep addled.
You should be aware that KataGo has no training experience with handicaps greater than 5 on a 19x19 board, or greater than 2 on a 13x13 board, and none at all on a 9x9 board.
So it’s figuring out everything entirely by pure guessing and extrapolation, and based on its initial shallow reading from the position you gave it with so many stones that it has never seen before - so it’s not necessarily accurate for larger handicap sizes.
I found the same system mentioned in the 1st comment of this blog article (“ハンディは９路盤は４段級差１子 […] 13路盤は２段級差１子”). The commenter apparently is a highschool/Go teacher.
Almost identical system, shifted by 0.5 kyu/dan difference:
In Nijmegen where I play we use the following table for handicap / komi.
- klasseverschil: class difference (2 classes is 1 kyu)
- voorgift: handicap
- komi voor wit: komi for white (negative komi: komi for Black)
Example you gave: 9kyu versus a 5 kyu means class difference 8;
19x19: handicap = 5; komi for white = 6
13x13: handicap = 2; komi for white = 2.5
9x9: handicap= 0 / komi for white = -2.5 (komi for black = 2.5)
On a 9x9 board only class difference >= 10 gives two handicap stones or more.
(So if there is 5 or more kyu difference: handicap.)
So the Nijmegen system uses 15 classes (7.5 ranks) per full handicap stone on 9x9?
I would expect that to be advantageous for white.
Did the Nijmegen club ever collect statistics on its 9x9 handicap system (in particular higher handicaps like 3+ stones on 9x9)?
Following link shows the complete handicap / komi table:
I know that this table has been used for at least three decades, but to be honest I don’t know the test results (if black did win significantly more than white or vice versa).
Yes, results of all competitions (since 1983) are collected and can be found on the website:
I give raw novices (playing their first game ever) 8 stones without komi on 9x9, giving about 50-50 odds in my experience.
I am class 15, so according to the Nijmegen table, that would put those raw novices at about class 110 (~45k), if I calculated that correctly. I do think that’s close to the truth.
Glad that you don’t come with more questions, because I wouldn’t have the answers. I did not create that table, Erik did.
Not sure if Erik is on OGS.