[Welcome to Sensei's Library!]

StartingPoints
ReferenceSection
About


Referenced by
ReferenceSection
MessagesToPeopleC...
MetaDiscussion
BigQuestionMark
WikiEtiquette
Handicap
SenteGettingTesuji
Oteai
OtakeFusekiChapOn...
Dan

Homepages
HuOfKGS
CraigDaniel

 

Rank
  Difficulty: Beginner   Keywords: Culture & History, Go term

see also: rank system

Apologies to Frs? for reviving version 60 but some useful info seemed to have been deleted. Nevertheless, this page could use a WikiMasterEdit. Go ahead, Frs? or anyone, but store what's beneath in Rank discussion? for instance. --Dieter

Why rank?

A main reason we go players sort ourselves into ranks is so we can have interesting games with strangers. Folks who play each other often know their relative strength without a system. (A less "noble" reason for ranking is pride or vanity.) Without go's rank and handicap system, weaker players would be totally frustrated and stronger players would be totally bored. With the rank and handicap system, a 3-kyu player can say "My rank is 3 kyu," and his new 3-dan opponent will know what handicap to give for a mutually satisfying game.

Dan Kyu Gup

Japan uses a system of kyu and dan ranks (or grades) for amateur players that has been adopted by most of the world. Korea calls its ranks gup. The Gup system starts with weaker players around 30 gup (note below), and goes down, so stronger players have a lower gup, with 1 gup being the strongest of all. The Kyu - Dan system is more confusing. It starts with the weakest players ranked around 30 kyu, and goes down as the players get stronger, with 1 kyu being a pretty strong player (similar to about 7 or 6 gup); then it starts over with the dan ranks. 1 dan is the first step after 1 kyu, and the dan ranks go up to 7 in most places. 7-dan players are the people who win the World Amateur Go Championship, for example.

Steps between ranks

The steps between amateur ranks correspond to the handicap difference. When a 6 kyu plays another 6 kyu, either player can take Black (over several games, they would alternate); there should normally be no handicap; and White would get komi to offset Black's advantage in moving first. When a 6 kyu plays a 5 kyu, the 6 kyu always is Black, and there is no komi (or only a half point to prevent jigo); so we see there is a one stone difference in the strength of a 6 kyu player and a 5 kyu player. When a 6 kyu plays a 4 kyu, it is a two stone difference in strength, so the 6 kyu not only takes Black, but puts two stones on the board before White makes a move. (0 or 0.5 point komi.)

6 kyu vs. 3 kyu, three handicap stones; up to 6 kyu vs. 1 kyu = five handicap stones; 6 kyu vs. 1 dan = six handicap stones; and so forth. In theory, a 20 kyu player should take a 19-stone handicap from a 1 kyu, and a 20-stone from a 1 dan. (In practice, handicaps larger than 9 stones are pretty rare. When the difference in strength between the players is greater than 9 stones, we just chalk it up as a Teaching Game, and accept that White will spend a lot of her time proposing better moves for the weaker player, just so she can stay awake.)

The Korean gup ranks work essentially the same way (a 7 gup player would need 3 stones handicap to have a reasonable game with a 4 gup player).

Note: I believe in korea they do not keep track of rankings until 18 gup. Hence there is "no such rank" as 19 gup, or 30 gup for that matter.


BillSpight: The traditional handicap scheme gives an advantage to White. In an even game, Black plays first and gives komi. If Black is one stone weaker, then he should take 2 stones and give komi. Or, reversing colors, White plays second and takes komi. If White is one stone weaker, then she should play first (i. e., play Black) and take komi. Either way, taking Black without receiving komi is an inadequate handicap for a one stone difference in strength. Similarly, taking 2 stones without receiving komi is inadequate for a 2-stone difference. And so on.

TakeNGive: Bill, is there a table somewhere that spells out handicap and komi combinations that give players of different strengths an even chance? The traditional scheme is the only one I'm familiar with.


Amateur instability

In the kyu ranks, it's not uncommon to find that a 20 kyu player might lose to a supposedly weaker 23 kyu. This can be due to unfamiliar playing style and uneven gaps in their understanding of go; and it can also be because in the weaker kyu ranks, improvement comes very rapidly.

Pro dan

Japanese professional ranks go from 1 dan to 9 dan, with 9 dan being strongest. In theory, by virtue of their superior training and rigorous discipline, professional players are always stronger than amateurs. In practice, a few ultra-strong amateurs 7 dans have occasionally beaten a few of the weaker professional 1 dans in even games. (That's what I heard from the folks who taught me the game; I have since learned that strong amateurs can sometimes beat strong pros.)


BillSpight: I read years ago (something by Sakata) that top amateurs could match pro 5th or 6th dans. Recently I have read claims such as the one above. Not that it matters, but I doubt it. :-)

TakeNGive: Particularly in light of the 2001 San Francisco Oza results, I have to say that very strong amateurs are as strong as professionals.


The steps between pro ranks are said to be smaller than between amateur ranks -- about 1/3 of a stone.


BillSpight: The traditional pro handicapping made a difference of 3 ranks per stone. However, with the new (post-WWII) pro rankings, there seems to be a 2-stone difference between 9-dans and shodans, which translates to about 1/4 stone per rank.


exswoo:so, if were to convert the pro ratings back to amateur ratings, a 9p would be more or less a 10d amateur? That's interesting..., and since the oft-repeated saying is that a grandmaster would need a 3-stone handicap to play against God, God must be 22p :)


Absolute rank

No such thing. The only way to establish such a thing would be to have a massive tournament, with every player in the world playing many games against other players from around the world, with nobody's skill improving. (The tournament idea sounds like fun, but by the time it was over, i would hope to have improved a stone or three!)

There is not a single rank-setting authority. Rather, there are several rank repositories (each national or continental go association is one such, as is each Go server that tracks ranks), and they do not always quite agree. So one person may be regarded as a 1-kyu in Osaka and as a 1-dan in Tokyo, for example.

Within rank repositories, there may be variation such that a person with a weaker rank can easily beat a person with a stronger rank. This is because ranks do not always keep pace with changes in players' abilities. If a small group of players within a repository mostly play amongst themselves, their ranks may seem stagnant even though the group as a whole is getting stronger. When such a player contests someone from outside the group, s/he will be surprisingly strong for her/is declared rank (which the outsider might regard as sandbagging). The larger the repository, the higher is the chance of isolated pockets of players whose rank isn't accurately compared to the general level of strength.

What about Igowin ranks?

Igowin is a popular, free, MS-Windows 9x9 version of a go playing program, The Many Faces of Go. More and more, players' first taste of our game comes from Igowin (kudos to programmer David Fotland for helping to promote go). One feature of Igowin is that it estimates what your rank is when you play it for several games.

How "accurate" is it? As long as you play only Igowin, its rank is good enough for your needs. But when you play other people, you will find that Igowin tends to overestimate your rank. (For example, on KGS I'm 10 kyu; based on AGA tournaments, I'm 11 kyu; but Igowin thinks I'm at least 6 kyu. Sometimes it mistakes me for a 3 dan.)

Suggested Equivalences between ranking systems

There is no universal rank. But here's a table of approximate equivalence for amateur ranks based on the table found on the [ext] NNGS help page. The list starts with total beginners, and ends with world-class amateur experts:

 AGA    CHINA  EUROPE    IGS    JAPAN    KOREA   NNGS
 ---    -----  ------    ---    -----    -----   ----
 27K     30K     29K     28K     26K     33Gup
 23K     26K     25K     28K     22K     29Gup
 19K     22K     21K     26K     18K     25Gup    22k
 15K     18K     17K     20K     14K     21Gup    18k
 11K     14K     13K     15K     10K     17Gup    14k
  7K     10K      9K     10K      6K     13Gup    10k
  5K      8K      7K      7K      4K     11Gup     8k
  3K      6K      5K      5K      2K      9Gup     6k
  1K      4K      3K      3K     *1D      7Gup     4k
 *1D      3K      2K      2K      2D      6Gup     3k
  2D      2K      1K      1K      3D      5Gup     2k
  3D      1K     *1D     *1D      4D      4Gup     1k
  4D     *1D      2D      2D    4-5D      3Gup    *1d
  5D      2D      3D      3D    5-6D      2Gup     2d
  6D      3D      4D      4D      6D      1Gup     3d
 6-7D     4D      5D      5D    6-7D      1Gup     4d
  7D    4-5D    5-6D      6D      7D      1Gup   5-6d

Additions, subtractions, and corrections, please.

- TakeNGive (10k)


Niklaus: Isn't the "unit" of the ranking system supposed to be handicap stones? So the only difference between the different scales should be where you set the bar for 30k (or 29k, since everybody below is 30k), and that difference should be the same on every level onwards? Most of the chart above is more or less consistent, but there are some areas (such as IGS double digit kyus and japanese high dans), where the "unit" seems to be more or less than one handicap stone. That doesn't make sense to me, because it makes the rating system a lot less useful.


Charles Matthews All the evidence is that server grades are not one stone apart, but closer. Where they are not used for handicap go much or at all, that's an artefact of the rating algorithm.


exswoo: The ranking system is purely relative since each system only have certain peoples from limited areas(both physical and virtual). So the ranking system works within that group because the stone differences do work out. What's happening is that the average strength of a player within each system is different, which gives rise to the major variances at each end of the spectrum.

From the table above, it's clear that the average player in each system is in the single digit kyu's, with the exact level varying about 3 or so stones.

Also, I'd like to question the Korean ranking table because while I can't vouch for the low end of the spectrum, I've played some 12 and 14 gups and they seem to only about 2 or 3 stones stronger than the AGA rankings.


Tobias Klaus?: IGS rating (at least in the single digit kyu and dan area) is a pretty exact match to the european rating right now. I readjusted the IGS numbers for that.


Harpreet: It seems the chart above did not reflect the IGS +3 inflation. So I adjusted the numbers for that.


Harpreet: Well, the chart is no good, it seems. I adjusted the ranks up for IGS inflation but if it's wrong relative to European ratings then it was wrong before as well. Now IGS is wrong relative to AGA (should be about the same). So perhaps this thing was fiction from the start.


I have heard that a Korean 1 Gup can be anything between 2 Dan and 7 Dan European/Asian and that 2 Gup is equivalent to 2k-1D. In the lower ranges, gup and kyu are approximately equivalent, but because Koreans don't have Dan grades, ranks don't go up quite as linearly as European/Asian. What is linear anyway: my experience is that handicap is not additive at all.

This is word having reached me and it can as well be myth.

--DieterVerhofstadt (1 kyu not gup)


Korean do give (sell) "official" amateur dan diplomas. Of course, you must be really strong (or rich ;-) to get it. One way is to pay to take a test (a handi game against a pro). I forget the pricing and the handicap but it gets more and more expensive as you progress up.


DaveSigaty: The table above is out of date, certainly for the server rankings. We should update it. It would be interesting to incorporate multi-ranking information provided by people to GTL and other people's actual experience. BTW how do ratings on KGS compare to IGS or NNGS? How do ratings on the Korean servers compare?

HolIgor: What about a server-server tourmament instead of flame wars.


A misconception I've experienced is that player A thinks something like: "I'm 6 kyu and have been able to beat player B. So player B can't possibly be 5 kyu, like he says." That isn't correct. The correct handicap should give each player a 50% winning probability, and a stone more or less will change that winning probability somewhat, but can hardly guarantee a sure win or loss (especially in the kyu ranges). The European Go Federation has taken the trouble to publish a statistic about winning percentages in even games between players of different rank [ext] here.

--Mark Wirdnam

SAS: These statistics are interesting, but care is needed in interpreting them, because the samples are necessarily biased. For example, consider the case of 5 kyu versus 7 kyu. Most (probably almost all) of the even games of this type in the EGF's database will be from McMahon tournaments, since most go tournaments in Europe are McMahon tournaments. For a 5 kyu to meet a 7 kyu in a McMahon tournament usually requires the 5 kyu to be doing badly and/or the 7 kyu to be doing well. So the sample of 5 kyu versus 7 kyu games will be biased towards the case of weak 5 kyu versus strong 7 kyu. I'm not sure how big this bias will be, but in any case the percentages given are likely to be nearer to 50% than they ought to be.


exswoo - All this talk about ranks got me interested in what is expected for How much a player should know at a given rank.


Hu of KGS: For a long time IGS was three ranks out of whack (away from equivalence) compared to all almost all other ranking schemes. I think it was the third quarter of 2002 that they were adjusted.

KGS ranks are held close to AGA ranks by use of anchors, but it would be good to put KGS into the table. While you are at it, move the misplaced "Suggested Equivalences" to a title where it belongs so that the whole table can be moved to the left margin.



This is a copy of the living page "Rank" at Sensei's Library.
(C) the Authors, published under the OpenContent License V1.0.