Will handicap game ranking adjustments ever be fixed? Go forum
18 replies. Last post: 20131124
Reply to this topic Return to forum
klaashaas at 20130205
I think it handles a win in handicap games the same as a win in even games. For your rating, Crelo, it doesn't matter if you beat me in a 9H game or an even game.
Beating me in a 9H game should (I hope :) be more difficult than beating me in an even game and you should be rewarded with more points.

Crelo at 20130205
Well, this means in fact that the equal games are not OK :) If we presume the handicap right, winning at 9 stones agains a 9 ranks weaker player should be the same as winning an even game against a player with the same rank.
Winning an equal game should value less for the stronger player. Or winning with 9 stones with a difference of more than 9 ranks.

klaashaas at 20130205
I think equal games are okay. If you and I play an equal game, I lose very little rating points, and you win very little points. This is they way it should be, IMO.
If I win a 9H game against you, I get rewarded with a huge amount of rating points (as if I would beat you in an even game). That's not fair. Beating you with 9H is not as much of an achievement as beating you in an equal game (or, when beating you when I play white and give you 9H, for that matter).

MarleysGhost at 20130205
> If I win a 9H game against you, I get rewarded with a huge amount of rating points (as if I had beat you in an even game).
This is not fair. But what /should/ happen?
If the handicap is accurate, the probability of winning is 0.5 for both players. That would indicate the winner gains 16 rating points and the loser loses 16 points (whatever that is in terms of k or d for Go).
Another possibility is that the winner gains n points and the loser loses n points, where n is the number of points the higherrated player would gain by winning a nonhandicap game.

Crelo at 20130205
An accurate handicap is the same as an even game between equal players.
When calculating the rating after the game, the difference between players should be corrected by the handicap, then used the normal formula.
For example 1 dan wins against 5 kyu with 3 stones handicap. The difference in ranks is 5, minus the stones is 2. So the 1 dan gains points like he wins against a 2 kyu (2 ranks). The 5 kyu lose points like against a 3 kyu (+2 ranks).
If a 10 kyu wins with 8 handicap against 2 dan, 11 ranks difference, corrected 118=3. The 10 kyu wins points like a win against a 7 kyu. The 2 dan lose points like a lost against a 2kyu.
A 3 kyu wins with 7 handicap against a 2 dan, 4 ranks difference, corrected 47=3. The 3 kyu get points like a win against a 6 kyu. The 2 dan lose points like a lost against a 5 dan.

klaashaas at 20130205
Ah.. 'even' is a misleading term :) An even game could be a game with no handicap, or a game where players have about the same (even) winning chance.
When I talk about 'an even game', I mean a game with no handicap, regardless of the players strength.

Crelo at 20130206
On an even game (no handicap) both players have the same chance before the first move. Only RoRoRo the Bot does not agree :)

Marius Halsor at 20130206
If you really think so, Crelo, I'd be happy to make som bets! :) (For instance, I'd be very willing to make an even bet that I'd win against a random 1600 player in Hex…)

Crelo at 20130206
Well, Hex is another game, too young yet. In Go, some beginner can find an old board in his grandfather attic and get an ancient ghost. Then win any game :D

ypercube ★ at 20130206
Crelo, you say: On an even game (no handicap) both players have the same chance before the first move.
Then why do we have handicaps? Why do we bother with kyus and dans and ELO points?

Crelo at 20130206
Now seriously, I agree that the rating formulas should be consistent. But I also think all games should be played even. The rating is too volatile on Internet servers and not something to take seriously.
While I like to play handicap games I consider them as teaching games. What does a handicap tournament prove? Who is the strongest player? Obviously no.

pfafulous at 20130207
Handicap tournaments allow weaker players to stay motivated by winning aometimes. It's frustrating to get destroyed every game. Handicaps allow opportunities to learn from stronger players, and feel challenged by weaker players. I enjoy the mix of nonhandi games for championship, and balanced games for monthly.

Crelo at 20130207
Well, handicap tournaments are usually not won by weaker players but by sandbagging players. It's frustrating to get destroyed by giving handicap to a strong player :) I think is much better to make tournaments between players with close ratings than handicap tournaments.

xed_over at 20131124
I like the Monthly tournaments. I like playing against both weaker and stronger players. The championships don't come around often enough. But for a tournament, I think the handicap pairing should be 1 (in other words, players 5 stones apart should play on 4 stones).
While I agree that it can be “frustrating to get destroyed every game”, it doesn't have to be. I like to compare it with other popular sports, such as Baseball, where a great batting average is still an extremely low percentage of hits – that's a lot of striking out. Get used to losing :)
I find it more frustrating to give handicap to stronger players (whether or not they may be accidentally sandbagging), but I try to take these as learning experiences and not be frustrated by them. Though I would like to fairly win a monthly tournament once in a while :)