Ranked games - different Lvl Players - NO handicap

When I play against a 2 Rank stronger Player in an even game - without Handicap - and lose by only 7.5 Stones, why is my Rank than still dropping?
Should I not get better Ranking points instead because of the result of the game suggests I am stronger than actual ranked and not weaker just because there were no Handicap stones in Place?
Is there no leveling system for the case : Ranked Game WITHOUT Handicap?

So you lost a ranked game and now you are surprised that your rank is dropping?

This is working as it should be.
If you win a game, you gain points. If you lose, you lose points. The amount of points is calculated according to the Elo system.
More information on Elo: http://en.wikipedia.org/wiki/Go_ranks_and_ratings#Elo-like_rating_systems_as_used_in_Go


Losing by 0.5 or 368.5 points is equivalent, a win is a win regardless of the margin. This principle is applied throughout the Go world (save a few rarities). So when you lose, even to a much stronger player, you never increase your rank. But the amount your rank decreases when you lose to a stronger opponent is small compared to the decrease when you lose to an equal opponent.


I am only surprised as I lost a game 12k against 10k with no handicap stones by only 7.5 pts. And still dropped down to 13k. This is just not logical to me.

And so mainly this tells me to watch out for games against weaker opponents with no handicap stones involved so I am sure to win the games and get a better ranking. And also to avoid games against stronger players that I am sure will lose.
This sounds a little bit stupid to me as it should not matter if you play against weaker or stronger players.

So maybe OGS is not as good as I thought it would be.

I feel there’s a problem with your proposal @tsitua. I think what you’re saying is that losses by a narrow margin should count for something. But remember there are two ways to end a game: scoring and resignation. As a dan player, I often resign if I’m losing by a few points with no way to catch up. Your proposal would penalize me for resigning, and it would penalize both players if I played until scoring, e.g. wasting our time.

Also consider that there is no Go server that gives rating points in the way you wish for. You might modify your statement to “maybe Go servers are not as good as I thought they would be.” But if you stop to think about why this is the case then perhaps you will warm up to the idea in the end. :smile:


The thing is, if you lose against stronger player in even game, you lose only little of your rank. If you win against him you get whole lot of a ranking points.

And if it is you who is the higher ranking player in a non-handicap game, by winning you can get only few ranking points and by losing you lose them a lot. This is kinda like the “prize” of winning somebody stronger than you and the “penalty” of losing somebody weaker than you, so this system encourages player to challenge (or accept open challenges from) stronger players, instead just picking 6 stones weaker players and beat the living hell out them.

When using automatic handicaps, it is like a game between two equally strong players and rank changes accordingly


But I was the weaker losing by little against a stronger player AND lost a lot of points. And if I win against a stronger player i just get a few points in return. So it looks like i would have to win 10 games for 1 lost game.
But I could be wrong. Are there any others with similar experience?

If you mean #1784471 you lost ~13 rating points…

1 Like

I tried looking up the game in question. as near as I can tell, the reason you lost 12 k is because you were close to being pushed back anyway. You can tell by the rank graph in your profile how “solid” your ranking is. the closer the bar is to the left, the closer you are to being pushed back one stone, and vice-versa for the right.

OGS’ ranking system isn’t unique among go servers. I’m sorry that you feel cheated in this instance, but if you leave here looking for a system you find more fair, I fear your search will go on forever.


If it’s 13 points, as KoBa said, that’s not a lot of points to lose. If the game had been even (equal ranks) or handicapped appropriately, it would have been more like 30 to 40 points. Had you won the game as was, you would have gained more than 40 points.

1 Like

The points you lose depend on the rank difference, as it was said before more than one time. But 2k isn’t that much of a diference; in fact, it seems to me that 2k diference is almost an even game. 13 points looks even like a small loss. If you lose to somebody 5k bigger than you, then you’ll lose only 3 points or so.
And your idea of fighting smaller rank people fails on the same way that the one relating stronger players. You’ll win few points if you fight somebody 3k lower than you, and there’s still some chance you might lose. It would be like betting money on 80% win chance, you’ll win several times few money, and then lose a lot when the other 20% is victorious. Not to mention that if you keep fighting weaker players you won’t improve a lot your go skills.


Ok, now I played a rated game against an 8k with Handicap, did win by resign, but my points did not change at all. Is this also normal???

Hi @tsitua,

The reason you did not gain any points from this victory: your opponent was a provisional player.

1 Like

On a slightly related note, the Wikipedia article on the Elo system has some really interesting bits. Specifically:

From a modern perspective, Elo’s simplifying assumptions are not necessary because computing power is inexpensive and widely available.
Moreover, even within the simplified model, more efficient estimation techniques are well known.

And then this:

Subsequent statistical tests have suggested that chess performance is almost certainly not distributed as a normal distribution, as weaker players have greater winning chances than Elo’s model predicts.

If this is true, maybe it would be interesting to try out some variations on the rating system and see if there is a noticeable difference.
Are any of you mathematically enclined enough to explain some of those ‘well known’ better estimation techniques?
If some of us were really knowledgeable about statistics, maybe we can figure out a way to get a new, more realistic performance distribution empirically from available OGS data and adapt Elo to use that? Better yet, make it self-correcting over time like the rank points :smile:


I unfortunately do not know of these well known techniques, but I would be intrigued to learn about them if someone here knows of a reputable source for it.

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.