Is there a rating board of top ranking players on OGS server?

Most Go servers have a top rating board or list that can sort by ranking, is there a similar function or pages for OGS?

And I cannot seem to find any players with 7d or above even in correspondent games. Is it scaled like the real-world amateur dan?

Which real-world rating: US, Europe, Japan, China, Korea, New Zealand, …? Each of them is on a different scale. Even online ranks aren’t in sync. But the answer to this question is probably no in any case.

OGS rank difference between two players is 1 if a game with handicap 1 gives both players a chance of 50% to win the game (assuming the weaker player plays as black).

OGS doesn’t have such a feature. You can find a one-time snapshot of the top rated players is the forum: Top OGS players (more serious this time)
You can read the discussion the if you are interested in why such rankings do not work.

2 Likes

No, it’s goes up to 9d+. It’ll be hard to find active people in that range though. Bots, botters and people who tricked rating system.

I was just thinking that armature dan here I know only goes to 7 dan realistically, so maybe it is the case for OGS. I don’t know it could go up to 9d+

And from my understanding from the discussion, I only gathered that it’s less reliable because too few live games and correspondent games have some weird dependency between them? Not too clear about how it is calculated thought.

But since everyone has a current rating value, isn’t it just a matter of sorting to list them regardless of how accurate they are in practice?

  1. If you maintain such a list, the top is full of bots and other rank manipulater.
  2. Ratings fluctuate. 2 players may change places several times a day.
  3. The error bars are much wider than the rating separation between 2 players. Placing one of these players on top of the other doesn’t have any meaning. Either is as strong as the other as far as we can say.
  4. “You do the top X list the wrong way!”
2 Likes

This account is 8d from correspondence games, but I haven’t played any rated games since 2014 so this rank is based on a very high winrate versus lower and mid dans years ago, rather than being able to beat the 8d bots that exist now.

3 Likes

Aren’t bot accounts grey? so should be able to keep them separate in different category, Not sure how many rank manipulators existed on OGS, I think they existed on every servers nowadays. Maybe just aren’t enough high level skill players on OGS to outnumbered them? Or like on IGS, there are verified Japanese pros play teaching handicap games everyday to anchor players at SDK and low dan ranking.

Some servers don’t just use ELO rating number to sort them one-by-one, but rather bracket them into ranking (within the rank just list alphabetically). Players strength shouldn’t change that much once accumulated enough played games. And there are grace period and overlapping rating buffer zone before advancing or demoting to a new rank.

However, KGS does maintain and constantly updating its top ranking list, and rating within rank.
http://www.gokgs.com/top100.jsp
I wonder what’s their method to keep it. Anyone knows how “stable” the list is?

1 Like

Here the KGS Top 5:
KGS rank graph for larc
KGS rank graph for cryx
KGS rank graph for agcStupGod
KGS rank graph for koyuki
KGS rank graph for Nexus7k1

The rating of some of them changed by several ranks within a few days.

You cannot outnumber the top 10. For finding the strongest player it doesn’t matter how many legit strong players are in the pool. Everyone with a decent KataGo backing him up will be stronger than any human and therefore on top.

I see only the last one jumped in Jun, and blank record, other’s looks pretty stable. the x-axis is unit in months, not days. And the y-axis have different scales, within the same line range is the same rank I believe.

And no one said about top 10, KGS maintain top 100, or IGS maintain a top 1000 I believe. And I don’t think they would use AI assist at that level with so many players watching their matches constantly and pride maybe? And like IGS with real human pros on site, it would be even less incentive.

Maybe the problem is with correspondent matches are the majority of game played on OGS contributing the rating?

And my original question is just to if this function existed on OGS, and might be a cool feature to keep some good players to play on OGS more often. Where it might get some perks or have a chance to play matches with pros or show match, where it would get notice with more high quality games to attract more viewers. A Go server might not just for playing games, but also a spectators’ arena some sort.

1 Like

We do have leaderboards, which are based on recent success in automatic sitewide tournaments instead rating. Tbh i like those leaderboards more than having a rating-based top list, since its kinda easy to say that strongest users on OGS are the strong AI’s, and the ~30 professionals whom have accounts here.

I doubt many strong players play it

Main title tournaments and ladders interest strong players for what I see. And after the soon promised revamp of the AST they may attract too.

So a global leaderboard based on all these could be a nice incentive.

The other thing about KGS is that your rating can continue to change even if you’re inactive. I believe it continues to update your rating based on how your opponents progress.

There were people reporting that they stopped playing for a while and then their rating went from kyu up through some dan ranks.

https://www.gokgs.com/help/rank.html

I read that KGS also has a type of anchor system
https://www.gokgs.com/help/anchor.html
Instead of like IGS with real human pros, they seems to be picked from player pools.

What is the promised revamp for the AST?

I tried several times for the automatic sitewide tournaments, and only once it got enough players after actively invited someone to fill the remaining empty spot.

I like using main title tournaments as the attraction and have leaderboards associated to them. But are they all played as correspondent games? Maybe we could have some joint tournaments with winners from the OGS site joint with other real live events from real-world Go associations’ tournaments? There are some blitz live tournaments would invite armatures to have a side track tournaments, or filling in the ranks. It might be a way to benefit both? If somehow can find good sponsors to them…

2 Likes

There were a couple of former 9d players in the Through the Years: Long Correspondence tournament, and many other 7d+ players as well (which considering the number of players invited doesn’t really surprise me that much). Some of them had held that rank for quite some time, but I say former because a lot of them seem to have found the number of games they were participating in (probably multiple concurrent simul tournaments) just a tad too much, and timed out of many games.

Another way of finding a lot of top ranked players (not all of them, I don’t think you could do that without trawling the API) is to look at the group OGS Title Tournaments — which is rather “top heavy” compared to the other very large groups, which usually target kyu players since that’s what most players in general are. Each group does have a player list sorted by rating/rank. I don’t think there’s any other group with near as many top ranked players though, and even if you manage to find most of those groups, there’d probably be a few players that aren’t in any of them.

I checked the top players listed in the title tournament and found that most of them have been inactive for years. And most titles seems to be hold by Sadaharu , but with just 3d rank at the moment, but listed as 9d+. (there is a big drop last year). What was going on back then? Was there something changed in the ranking system last year?

Starting roughly August 14th they timed out around (I don’t really want to count them), say 200 correspondence games or so. Most of them were annulled, but this type of thing would have a negative effect on rank regardless of any "most"s. Seems to have happened to many of the other stronger players in that long correspondence tournament I mentioned too.

They all timed out around the same time? I saw gaps in months before any new games been played, like everyone all went on their vacation together?

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