Rating system

[Sumatakyo]

I just had to add that I really LOVE the system on Tygem.

A lot of people cherish their rank, and the fact that you never know when your rank will increase or decrease on KGS is somewhat... stressful

I've played hundreds of games on Tygem because it is so clear, and I know that I can lose a few games without losing a rank. I also tend to play more seriously throughout the entire game when I know that it might affect my rank either positively or negatively.

I like the variety of play on KGS, but the rating system on Tygem rocks

[Pippen]

What is the rank comparison from your experience between KGS and Tygem? I saw guys (e.g. "Inazuma") that are 1k on KGS and 3d on Tygem. Is that typical?

[Eizero]

Yep. KGS ranks are like 2 to 3 stones stronger than tygem ranks at dan level. I heard the difference between a weak tygem 9d and strong tygem 9d was 3 stones. I'm 3d on kgs, 6d on tygem.

[Pippen]

Yep. KGS ranks are like 2 to 3 stones stronger than tygem ranks at dan level. I heard the difference between a weak tygem 9d and strong tygem 9d was 3 stones. I'm 3d on kgs, 6d on tygem.

Oh, then that is an argument against tygem for me. I don't like those up-pimped ranking systems where you can get too easily to a dan degree^^. But the server looks like fun to watch pro games.

[petri]

I have ended up with a preference for simple system over a mathematically intricate system for playing on. There are huge amounts of factors (such as creating new accounts, deliberately sandbagging etc) that make for a far bigger impact than the intricate maths of theoretically "superior" systems. IGS and Tygem both have very simple systems yet I find myself preferring them every time I play on the different servers.

Well for strong players with relatively stable ranks these simple schemes may work. But for ranks say weaker than say 4k KGS they just fail. 13k in these korean servers can easily be 5 stones stronger than 13k, or even 7 stones stronger. Player getting better quickly, playing against other developing players will mess the system. Does not happen in KGS, nor would it happen with any ELO-based system, not even with original slow update algorithm. Besides mathematical max-likelihood ones beat any other system in real life in ability to predict outcomes of the matches so why not use them?

[Pippen]

IMO it would be better if it would reset after every 20 games group. That means: If you e.g. are 13W6L and you lose one more it shifts to 0W0L while keeping your rank and says: 1D up: +14, 1D down: -14 and then it starts all over again. I feel that would be clearer.

[Pippen]

After playing as a pretty stable 1d KGS, after over 90 games on Tygem and based on my playing experience I can say that a 3D-Tygem equals an average 1d-KGS and a 4d-Tygem equals a strong 1d-KGS or a weak 2d-KGS.

[cdybeijing]

After playing as a pretty stable 1d KGS, after over 90 games on Tygem and based on my playing experience I can say that a 3D-Tygem equals an average 1d-KGS and a 4d-Tygem equals a strong 1d-KGS or a weak 2d-KGS.

This is also my exact experience.

[lobotommy]

After playing as a pretty stable 1d KGS, after over 90 games on Tygem and based on my playing experience I can say that a 3D-Tygem equals an average 1d-KGS and a 4d-Tygem equals a strong 1d-KGS or a weak 2d-KGS.

I'm just curious how it translate to the "real world" EGF/AGA/other organization rank? What is a 1dan on kgs? EGF 1-5 kyu? AGA 1k/1dan?

[Pippen]

I'm just curious how it translate to the "real world" EGF/AGA/other organization rank? What is a 1dan on kgs? EGF 1-5 kyu? AGA 1k/1dan?

EGF <= KGS, means: EGF and KGS ranks are about the same with a small touch EGF's rank being a bit stronger/lower in comparison to KGS, maybe half a stone; kind of like a 1d KGS might be a strong 1k EGF. AGA and Tygem should be about the same. Both are way over KGS' ranks.

[Kirby]

I'm just curious how it translate to the "real world" EGF/AGA/other organization rank? What is a 1dan on kgs? EGF 1-5 kyu? AGA 1k/1dan?

EGF <= KGS, means: EGF and KGS ranks are about the same with a small touch EGF's rank being a bit stronger/lower in comparison to KGS, maybe half a stone; kind of like a 1d KGS might be a strong 1k EGF. AGA and Tygem should be about the same. Both are way over KGS' ranks.

I'd be wary of these direct comparisons, since so many factors are at stake. For the longest time, my KGS rank was several stones "over" than my AGA rank, mostly due to the amount of tournaments I played, and the same was true for awhile on Tygem.

It might be arguable that the "average strength" of a system can be compared to another, but even this isn't a great indicator, as some systems have greater variation and/or skew.

On top of all of it, I personally believe that there is a psychological aspect, too: if you believe that you should be able to beat rank X on a given system, you are more likely to

[Pippen]

What does "RP" mean and what's the use of it? I am a strong 3d/weak 4d at Tygem and I have 21184 RP right now.

[lobotommy]

EGF <= KGS, means: EGF and KGS ranks are about the same with a small touch EGF's rank being a bit stronger/lower in comparison to KGS, maybe half a stone; kind of like a 1d KGS might be a strong 1k EGF. AGA and Tygem should be about the same. Both are way over KGS' ranks.

I am a little sceptic about that (I'm 3k egf vs 2d kgs), but probably it varies for each country and my tournament's updates of EGF are too low in numbers. But I'm in doubt I could be a dan level, maybe 1 kyu, probably just 2k.

[skydyr]

EGF <= KGS, means: EGF and KGS ranks are about the same with a small touch EGF's rank being a bit stronger/lower in comparison to KGS, maybe half a stone; kind of like a 1d KGS might be a strong 1k EGF. AGA and Tygem should be about the same. Both are way over KGS' ranks.

I am a little sceptic about that (I'm 3k egf vs 2d kgs), but probably it varies for each country and my tournament's updates of EGF are too low in numbers. But I'm in doubt I could be a dan level, maybe 1 kyu, probably just 2k.

Just as people speak of a Korean style, for example, there seem to be styles in Western go that are common to certain countries, and it wouldn't surprise me if a European style might be better or worse at particular things that happen to easily exploit or be exploited by an American style, for example, which result in some of the rating differences (but not all).

Additionally, rating systems make an assumption about how broad a given rank is in terms of strength, so it's possible that a standard rank difference of, say, 2 ranks around 1d-2k may spread to more or shrink to less as one goes up and down the rank scale.