I don’t understand the overall go ranking. I am 21kyu in blitz and 17 kyu in live and correspondence. still I am 14 kyu in overall. How is that possible ? How does it work ?
The three time-control ratings (blitz, live, correspondence) are, as you would expect, based on your performance in those time settings vs your opponents’ ratings in the individual settings. ‘overall’ is essentially a fourth rating that is calculated against your opponents’ overall rating. It is not an average of the other ratings. Depending on which types you and your opponents play more vs your performance, this happens, because math. Chances are, you’ve done better against people who play those time settings less.
E.g., if you are 20k live and 18k overall and beat a player who is also 18k overall but rarely plays live, and is thus still 25k, your live rating will barely increase while your overall will get a healthy bump.
It’s also good to note this is the only server with 4 rankings every other server just has one.
We used to only have one, but that wasn’t considered representative in many cases.
Another thing to consider with respect to the Overall rating: the Overall rating is based on all games. I think the calculations are the same as for each of the separate ratings. This will have it moving faster than the others. This will work both ways, but since folks tend to improve more than not, Overall will tend to lead.
1 Like
Also, any time you’d like your ratings true’d up to match one another, it’s easy for us to do that, just ask the mods. Well, when I say anytime, don’t do it every week, but you know what I mean. ;o)
2 Likes
If you beat someone by 30 points or 2 points, will it have the same effect on your ranking?
Yes, the effect will be the same, the point difference doesn’t matter. The Elo system only takes into the account whether the game ended in a win, a loss, or a draw. Hardly surprising, given that Elo had originally developed his system for chess.
I understand… In this case an 11 kyu player wins 6 games against a 14 kyu player and it won’t have any effect even if he wins by 40 points? But i guess the other person would just resign… I don’t know but still It would be nice if it took the point difference to account as well. 10 points = 1 stone stronger etc…
An 11k player is expected to beat a 14k player, but when he does, the ratings are still adjusted: the 11k will get some Elo points and the 14k will lose some. Just not so many points. On the other hand, if a 14k beats an 11k, that’s unexpected, so the adjustments will be more dramatic.
That’s not how things usually work in go. There are some forms of go where point difference matters (mego and bangneki gambling, Hahn pointing system, some handicap selection schemes), but normally, a win is a win and a loss is a loss. 
Oh man I couldn’t disagree with you more on point differential ranks, especially here. Point differential isn’t the point of Go, and encouraging that would terribly skew ranks, because in lower ranks, aggressive players would advance faster, until they hi opponents that can properly respond to and punish massive overplays. Plus, point differential only works in more explicitly competitive games, which would encourage people to be more cutthroat on the server, leading to less pleasant games, even getting to the point where resignation could be frowned upon because of whatever impact (or not) it has on games. This isn’t even to mention the fact that score differentials often inflate artificially. Let’s say white is losing mid-game, they’re going to have to start making riskier plays to come up with enough territory to make a comeback. This may end even worse for them on points, and a game that ends B+40 might be a completely inaccurate reflection of skill since it may have been more like B+15 if white was playing for score differential instead of trying to win.
What you’re looking for, which seems less common here for some reason, is a handicap game. As I recall (someone please correct me if I’m wrong), handicaps are factored in when calculating ratings.
3 Likes
I thought when you win by 40 points it would mean the player is 4 stones stronger so it would make the win 4 kyu higher, if it’s against a 11 kyu, +40 points would mean the win is against a 7 kyu. It seemed logical… I didn’t think the way you explained before and you are right. That kind of point based system might have very negative effects on the way people play… Thanks for the extensive explanation…
2 Likes
Sure they are. As an example: I’m currently 9k on DGS and recently I’ve played a couple of 9x9 games against a 3d player. The first one was with a normal handicap (2 stones and 6 points of reverse komi), winning it improved my rank by 0.11k. The second game was with a minimal handicap (I’m black, no komi). I’ve lost and this result was SO expected, there was no noticeable change to the rankings at all.
1 Like
it makes sense that win margin isnt considered. After all recognizing your own lead and playing accordingly (a little safer) reflects strenght as well… often more so, than mindless attacks do imho 
Absolutely right
It wouldn’t be a good way to work ranking seeing as how you will beat one 11kyu one day anywhere from .5-40 points and lose to the same 11kyu the next day by .5-40 points
Well, if one’s play is that unstable (and for kyu players it often is) there is no good way to rank it reliably, no matter which system is used. That’s why I’m not overly concerned with my exact rating value. I know I’m somewhere in the weaker half of SDK, hopefully progressing towards the stronger half, that’s enough for me. When (if ever) I’m dan, I might to start caring about which dan. Caring about which kyu I’m doesn’t seem to be worth it.
.5-40 points is an extreme example but its not uncommon for dan players to lose to each-other within komi range both ways to the same person
You’re not wrong, at least that’s what I’d heard as the conventional wisdom as well, but the only times I’ve seen it work in reality is in scenarios like informal club and teaching games.
I’ve used that method for initial settings in a club situation where we had no other indication.
While higher ranked players are more consistent in their play, there are times with a large group lives or dies at the end of a long sequence where the difference is a sente at the moment the group needed another liberty.