Rank - worldwide comparison

Table of contents Table from a 2018 OGS survey Graphs from a 2015 OGS survey Table of approximate equivalences between ranking systems Problems with the table presented above Rating methods More ratings links Footnotes

Table from a 2018 OGS survey

https://idex.github.io/go-rank-survey/go-survey-results#A-table-of-rounded-mean-ranks

Graphs from a 2015 OGS survey

OGS conducted a survey in November 2015 , where users gave their ranks on several servers and organizations.

• results scroll down a bit and you can see scatter plots comparing different rank systems.
• survey announcement

Another OGS connected survey in March 2018. (Credit: Sebastian Pountney, sourced from AGA e-journal 2018-04-18)

• results

Table of approximate equivalences between ranking systems

Please help to improve this table by adding data to our /DataCollection and feel free to edit it if you have more accurate information than currently presented.

However, since universally valid ranks do not exist; this is and will always be a table of approximate rank equivalences. Approximate means that a rank in one system can be several ranks above or below a rank stated for another system; the latter rank need not even be the mean.

The table starts with beginners and ends with amateur experts.

Problems with the table presented above

Be very aware that the information in this table is at best an approximation, and is not very reliable.

• People are living in different time zones - some play comfortably in the evening while their opponents play in the middle of the night.
• Older data points used to create the table have not been removed when the relevant ranking system changed.
• The AGA has no ranks at all, just numerical ratings. There is no official translation between numerical ratings and ranks, and there are no ranks above 7d.
• EGF ranks vary internally in Europe. A large range of methodologies exist to allocate ranks, ranging from Anarchy to Strict Implementation of the GoR formula. Selecting a country from this page at the EGD will display a graph comparing that country's ratings versus standard GoR.
• Likewise ranks are awarded differently in provinces in China, and it is undoubtably the case in Japan and Korea too.
• WBaduk ranks vary internally between the different servers.
• The whole list expresses strength differences only with integers, hence might be off by half a dan grade (China)
• Blitz internet ranks and ranks derived from slower time limits online are often markedly diverse.
• Turn based servers can suffer from novices improving far faster than the rating system can cope with. (This is also a problem in EGF territories which use a strict implementation of GoR with no rating resets.)
• Some people will play online while drinking
• Some people will play online while distracted by children, siblings, or parents.
• Ranks from face-to-face games are based on comparatively small numbers of games (often played in tournaments) which are taken more seriously. Ranks from online servers include large numbers of games played more casually.
• The internet can also contain sandbaggers, players who are deliberately playing under their strength in order to beat up weaker players. The IGS rank of 30kyu used to be notorious for this.
• There is no data for South America or Australasia. ^[2]
• AGA ranks vary internally in the USA, one coast is said to be stronger than the other.
• Depending on where a player learned to play go and where he mainly plays, it is possible that he is significantly stronger or weaker on real gobans than on virtual ones. A difference of 5 stones and more are not unusual.
• Ranks can be determined in different ways in the same country. For example there is an official Nihon Ki-in rank in Japan that is determined by tournament play, not by paying a pro for an evaluation. Those ranks are comparable to AGA and EGF ranks. There are also go club ranks in Japan and ranks obtained by paying a pro. Such ranks could be weaker or stronger than those of other countries.
• Some ranking systems converge very slowly to a player's actual strength. This is a problem with the IGS ranking system for ranks weaker than about 5k, because these players are often improving faster than the ranking system is converging. Consequently, within the rank of "8k" on IGS there will be players whose strengths differ by three or more stones. The KGS ranking system converges very quickly, so its ranks are more consistent, except for frequently playing players, whose ranks can move very slowly.
• OGS ranks have been adjusted since this chart was produced; these comparisons are no longer valid. For kyu players, OGS ranks now seem to be similar to KGS and IGS ranks, but of course we need real data.

Rating methods

• AGA - ratings information at AGA rating information
• China - here is a link(chinese)
• Japan - See Points Rating System
• Europe - grades vary by country, see http://gemma.ujf.cas.cz/~cieply/GO/rgdrep.html. for a comparison of ratings with GoR. Comparisons of GoR do not take into account isolated player populations. Also see Ian Davis / Gor
• WBaduk - See World Rating System
• IGS - B C, beginners class, is used for players below 17k and guest logins. see the IGS-PandaNet rating system for more information
• DGS - Link to the Dragon rating system
• KGS - Here are the Details of the KGS Rank System.
• OGS - Online Go Server has made the following changes to their ranking system:
  • July 2014 - corrected its ELO/GoR ranking system (targeting a 20k+ rating system to be more compatible with KGS/IGS/EGF/AGA): details in the OGS forum
  • August 2017 - Switched to Glicko2 rating system. details
  • January 2021 - revised ranks again: details

More ratings links

See also:

• /DataCollection – Our attempt to gather data
• Rank - first comments
• The Forum for this page
• RatingHistogramComparisons.
• DragonGoServer has a rank-converter that is based on the above conversion table.
  • It can be used for rank translations between AGA, China, Europe, Japan, Korea, IGS, DGS, IYT and some more servers or ratings; see function "convert_to_rating()" in DGS server source code.
  • ErikO: The conversions in "convert_to_rating()" are in most cases taken directly from the rank translation table at NNGS (type help rank).
  • JUG: The rank conversions and interpolations had been adjusted to closer match the above conversion table (Jun-2012)
• You may find a subjective estimation of your EGF rank after solving this test: http://play.baduk.org/ (test results are discussed in a special chapter at /discussion)
• Rank microcosm
• New rank comparison list (from 2010 data of various servers/lists)

Footnotes

[1] BC (Beginners Class) is a blanket initial rank at IGS. Unranked people & guests start as BC for their first 5 games at least. The first numbered IGS rank appears to be 17kyu currently. There are also some stronger players who choose to have a BC rank.
[2] xela: As I understand it, Australian ranks are set by consensus within each go club. There are not enough tournament games for accurate calculations in the style of American and European ranks. It's generally believed that Australian kyu ranks sit somewhere between the American and Japanese ranks. High dan ranks are possibly a bit stronger (Australian 7 dans seem to hold their own against other 7 dans at the WAGC), and so it's likely that the spacing between dan ranks is greater than one handicap stone per rank. Possibly someone else will have more accurate data on this.
[3] The European Go Federation rates tournament results of European players with the EGF rating system. A European 7D grade is claimed to be approximately equivalent to a professional 1D grade by the EGF. 20K is the lowest rank used for the EGF rating system and usually the lowest rank used for pairings in European tournaments. Some European players claim a lower rank.
[4] This column represents the rankings as used by the Hankuk Kiwon, for information on the old gup ranks, see rank - gup (Korean) explained
[5] Ranks on Wbaduk (formerly Orobaduk, CyberOro) depend on the server you play on: On the Japanese server the ranks tend to be better than on the Korean servers.
[6] Now that the Tygem English server is open, perhaps we can include Tygem ranks here when more data becomes available. As a general rule of thumb you can assume that Tygem ranks are very similar to those of Wbaduk (this also holds for the number of sandbaggers in the DDK ranges)