I know a lot of people who privately agree with me that OGS has the worst ranking system. To me, ranks are designed to make it easier to find an opponent with roughly “even” strength. I have found that KGS, IGS, Tygem, and Fox all do a good job of this. Of course there is the occasional sandbagger and people can sometimes have an off day, but it’s generally easy to get a game that feels roughly “even.”
On OGS, it feels like a complete crapshoot. I often joke that I feel gaslit by OGS. I find it to be the most infuriating server to play on for this reason. When I play someone with the same rank as me, the feeling of difference in our skill is widely varied. Sometimes it feels even, but more often it feels mismatched in some way or another. I often play players who are multiple stones weaker than my OGS rank that feel stronger than people my own rank.
I imagine part of this is the “aggregated” ranking system that combines 9x9, 13x13, 19x19, live, and correspondence games. I am crazy or do other players feel this way? I just don’t think a rating system is good if it is that hard to find an even match.
It always was easy for me to find even opponent on OGS
I keep dropping to 6k, but I have 7 wins and 1 lose vs 7k players in the last 25 days (19x19, no handicap, live)
I’m sure you can find others who share your views but I really don’t think it’s a common position.
Also, if the ranking system was a “complete crapshoot”, you would expect your rank to vary significantly for no reason. Instead, over your last hundred games your rank was very stable, varying by 1k at most, which seems to indicate everything is working as it should.
… and you can dig deeper in to this to see whether your rank is stable “just because you get randomly opponents wildly less and wildly more skilled than you”.
We can see that this is not the case, because we can see that you win far more games than you lose against weaker opponents, and lose more than you win against stronger opponents:
Of course, OGS ranking system has areas of improvement, and it’s a topic that is constantly discussed on these forums, but I feel it does a pretty good job overall.
I think the OGS rating system is not like a crapshoot, but OGS ratings are more volatile than ratings on other servers. There are several topics about it (like this one).
As far as I understand, this volatility is by design. An upside is a quick settling towards a roughly correct rating for newly registered players. A downside is that ratings of established frequent players may fluctuate by a couple of ranks, while those may only fluctuate by like half a rank on other servers.
Yeah. My view on it is that many other servers artificially limit how easily one’s rank can drop, whereas OGS is more honest about how close the ranks really are to eachother
Its possible to choose rank of opponents to always keep win-rate around 50%.
Rank fluctuation is big only if you don’t care against which ranks you play.
Also sometimes you actually play worse than usual, then rank will temporarily go down, because its The Truth.
Yep. I appreciate OGS’s honesty in this regard. Many servers give players the false impression that playing stronger or weaker than your rank is very unusual, when if you look at the glicko-2 numbers on OGS, it’s quite clear that the win probability for a player 1 rank stronger than their opponent in an even game, is far from overwhelming:
A player 1 rank stronger than me is about 55 elo above me. I don’t see an expected win percentage, but Elo Win Probability Calculator estimates my win probability as still 41%. This could vary quite a bit depending on how different the OGS glicko-2 implementation is from the assumptions that site made, but the real value should back up the general idea that it’s not that big an advantage to be 1 rank stronger in the high sdk range
What is the reverse of volatility? Cristalization?
Didn’t play on kgs since some time, but it was hard to rank up especially after having played a good quantity of games, so when you felt playing stronger you had to create a new account to be able to rank up. Different stories.
It may also be worth pondering that if you don’t play evenly up and down, then you may also be getting the wrong idea about what your own rank (what your actual skill compared to other OGS players) is.
If you play down a lot, you get accustomed to winning, so when a person who is actually of your same rank comes along, it’s quite possible that it feels that they are “sandbagging” whereas in fact you have just become accustomed to being able to beat people due to playing down…
OP has played down in most of their games (~67%), but I have played down even more (~74%) and my winrate is ~67%, where OP’s winrate is more like ~55%.
I wonder if that has anything to do with both of us being at the high end of the rating distribution: there are simply more lower rated players than higher rated players, so we’re more likely to get lower rated opponents than to get higher rated opponents?
I suppose so yes, but it may take some effort on OGS.
In the EGF rating system most games are from MacMahon tournaments, so players will play up and down about evenly (except perhaps for the strongest and the weakest tournament players).
My EGF winrate is quite close to 50% (362 wins vs 370 losses).
50 Elo per rank seems about right for ~10k according to historical data from the EGF database.
100 Elo per rank seems about right for low amateur dan, 200 Elo per rank seems about right for high amateur dan and 300 Elo per rank (i.e. full handicap stone, ~3 pro ranks) seems about right for pros.
Those number are behind the EGF rating system since 3 years ago. OGS uses a less skewed mapping between Elo ratings and ranks, where 100 Elo per rank applies more to mid amateur dan than to low amateur dan.
Vertical axis is Elo per rank.
Horizontal axis is OGS rank as an integer, where 30 corresponds to 1d IIRC.
Red curve is OGS mapping.
Blue curve is EGF mapping.
I guess it’s a different discussion if you use handicap.
Personally, I haven’t started to think about that variation of the discussion, and I can feel a headache already when my mind glances in that direction
I thought OGS was renowned for being a place where handicap is not played
Haha, yeah, I also seem to remember, albeit vaguely, that some (many?) ppl here don’t like Handicap … I have no idea why that is so because appropriate(!) HC is, IMO, what gives both players a roughly 50% chance of winning/losing while having an ideal mixture of excitement, challenge, exercise …
Go is not only about winrate. Its also about cool fuseki. With handicap stones its impossible to paint something interesting and unusual in the opening.
Well, I’m only a 9k player, so take everything I say with a grain rock of salt.
about cool fuseki
Why should Handicap Fuseki be not cool?
Using Handicap doesn’t mean that ALL one’s games are Handicap games. IMO a “healthy” relation would be to play roughly 33% each of weaker players, stronger players, and peers, or make it 25/25/50 or something.
As I see it, playing Go using Handicap is like a good operating system which reveals its more nerdy options once you’re acquainted with plain vanilla and begin snooping around behind the curtains. What I mean is, the game gets more “interesting” and challenging all by itself as one progresses and takes fewer HC stones.
There are so many opportunities for interesting and unusual openings with or without HC, HC just gives the weaker player a jumpstart advantage.
