I am foxGo 5D but I cant win 2 GS 1k player in a row, is this rank normal?
2 Likes
1d OGS is supposed to be between 1d EGF and 1d AGA level. It seems to be the case that Fox ranks around that level are about 3-4 ranks softer than EGF/AGA ranks. So I’d say it’s expected for a 5d Fox to win about 2/3 of their games against 1k OGS.
Also, 2 games is probably not a large enough sample to tell if your case is a clear discrepancy from that expectation.
6 Likes
It might be supposed to be like this. But in fact there are a lot 1k OGS players who are playing as 2d EGF. So I am not surprised that you are loosing against them.
7 Likes
thanks, I am registering for the lego season 20 as 3D as AGA, guess I am not 3D level, will there be handicap in the lego season 20 based on registered level or it doesnt matter?
2 Likes
Yes, this is not unusual - I’m also 5d-6d on Fox and around 1d on OGS. The ranks are a big mess around this level, because unfortunately there aren’t enough active dan players to maintain the accuracy of the ratings. I’ve easily beaten multiple 3d’s, and gotten crushed by multiple 2k’s.
5 Likes
NAOL uses yet another rating system, not OGS’s ranks! Tell them your rating for whichever system you have the most games and they’ll help you figure it out.
5 Likes
What do you mean “supposed to” here - does OGS intend to calibrate their ratings this way? It’s a bit off - I’m around 3d-4d AGA and all of my peers who play actively on OGS have told me that they are either 1k or 1d.
3 Likes
thanks, I dont want to get handicap in NAOL but I dont want to handicap player stronger than me neither…and if handicap does not exist in NAOL then I dont care
For reference, I first made Fox 3d a couple months ago when I was still OGS 4k. I registered on Fox as 4k i initially and shot straight up to 3d. Haven’t played on Fox since though so don’t know what rank I’d be there now.
1 Like
I doubt it. Substitute “supposedly” for “supposed to be” and I suspect you’ll have the intended meaning of that sentence.
2 Likes
on Fox you can start from 3D, and I can tell GS 2k is def better than Fox 3D
1 Like
I haven’t played on Fox for a while, but I was Fox 4d and now OGS 2k (was OGS 1k one week ago).
1 Like
5 years ago, OGS ranks were calibrated so that 1d OGS fell between 1d EGF and 1d AGA. Also see item 1 in 2021 Rating and rank adjustments
I assume it was OGS’ hope/intention that this calibration would keep up over the years, but I don’t think there are actual mechanisms in place to maintain it.
There are reports/indications that OGS ranks around 1d have gradually become tougher again. OP is the latest example.
English is not my native language. I don’t know which wording would have conveyed best what I intended to say.
I meant to say that 5 year ago it was an explicit goal for OGS to land 1d OGS between 1d EGF and 1d AGA. And I’d say this was actually the case after the 2021 OGS rating system update. Until June 2023 I did play somewhat actively on OGS, maintaining a 3d OGS rating (gennan). This was in accordance with my a long-time 3d EGF rating (albeit gradually weakening).
I’ve never played on Fox myself, so I rely on what other players say about how their Fox ranks compare to EGF/AGA/OGS ranks.
6 Likes
I do think a calibration is in order. I want to better understand the causes of the deflation before making any adjustments though. I have a couple of strong suspicions, but much analysis and simulation work is necessary before taking action to re-calibrate. This is in my plan for this year though fwiw.
20 Likes
I’ve grown kind of fond of the deflated ranks haha 
. It’s my personal goal to reach 5 deflated dan on OGS this year. If we can actually match EGF ranks, meaning bump everyone up by +1 to +1.5 ranks at most, I’d be happy.
1 Like
You worded it exactly right.
”supposed to be” implies intent
”supposedly” is more of a lore of general understanding of how things are
I did not realize that the specific link between ranking systems was intentional when I wrote that.
1 Like
It’s our objective to be in between AGA and EGF ranks, we don’t care about Fox (they tend to run much higher)
7 Likes
Glad to hear this is a priority - do you need any support? I have some (informal) experience studying online rating systems, primarily in the context of chess.
Unsolicited opinion: I think it would be a service to the larger go community if OGS explicitly pegged its ratings to the AGA and/or EGF. It would help stabilize OGS ranks, help online players make a smoother transition to OTB go, open up possibilities for online rated events, and maybe take a step toward consolidating online go in the west.
I don’t think it would even be necessary to choose between the two federations - since the system already maintains multiple ratings per player, simply calculate an AGA-pegged and an EGF-pegged rating for every user, and give each user a global toggle to determine which ratings are displayed for that user. I suppose the main challenge would be in designing a system which allows users to link their OGS account to their AGA or EGF account.
2 Likes
GitHub - online-go/goratings: This repository contains the (future) official rating and ranking system for online-go.com, as well as analysis code and data to develop that system and compare it to other reference systems. · GitHub is the repo we work from for ratings work if you’re interested.
That’s essentially the goal, when we do an adjustment we optimize on minimizing the error between what our (updated) ranks are and what their AGA/EGF rank is.
We don’t do individual rank pegging though, our primary goal with the rating and ranking system is to have ranks mean that given a 1 rank difference a 1 handicap brings the expected win rate back to that of a non handicap even ranked game. So first and foremost our rating and rank mapping system is tuned so that’s true, and then it’s a matter of labeling what the underlying rank numbers mean to labels like 1k and 4d, and that part is optimized to be comparable to the AGA/EGF.
3 Likes
I looked at a small sample of 1d vs 1d games in 2022 and in 2026, ordered by percentage of excellent moves.
The graph shouldn’t be taken too seriously since the sample is very small, but perhaps if we take a large sample and compare percentages of excellent moves in 2022 and in 2026 for each rank, we could measure the rank drift if it exists.
12 Likes