The strength differences between professional levels is...

hyperpape · **#21**

The Lee Sedol match was online. The Japanese match is longstanding, you can find an article on Senseis Library. I assume it is played in person.

Bill Spight · **#22**

pookpooi wrote:

Bill Spight wrote:

These days pros no longer give handicaps to other pros. Maybe the 2-3 stones you have heard comes from Ing's ranking system.

Now what I need more from these information is that, after given such a handicap, does winrate balanced to around 50:50? (I don't think so)

Certainly up to WWII the win rate was balanced. That was the idea of handicaps, to balance the win rate. And Ing in 1980 plainly intended the handicaps to balance the win rate.

True, there were exceptional cases where the lower ranked player was markedly stronger than his rank, and up and coming players had to prove themselves by winning more than 50% of the games, but once players had plateaued, the handicaps reflected differences in strength.

xed_over · **#23**

pookpooi wrote:

hyperpape wrote:

Lee Sedol beat a group of Western pros down to a three stone handicap, and the Japanese professional vs. amateur Meijin contest often sees the professionals winning with 2 stones.

These are both very weak signals, and your question is still not perfectly defined. Still, I think that >= 2 stones is a safe bet.

Yeah, that's the kind of information I want to hear more. I hope those matches are not play on the internet cause some people might say that they're not playing seriously because the match happen online.

While our newly crowned, first ever, American pros had earned the title of "professional", they were still a ways off from being as strong as their Asian counterparts. That was part of the point of the match, to test just how far off they were.

I'm not sure you'll want to use these particular examples in your quest.

xed_over · **#24**

While we have tendencies to rank people and list some as better than others, other cultures not so much. At last year's US Go Congress, Maeda Ryo 6p of the Kansai Kiin told us during one of his lectures that they don't consider lower ranking pros to be any weaker than higher ranking pros. They consider them all on an equal level.

Of course, we all know that actual results speak louder than that, but that's the attitude (and respect) they have toward their professional players of all levels.

Elom · **#25**

The relation of a single stone of difference to a specified rating difference seems to be a measure of convenience.

I don't know how this works too well, but it appears that the winning percentage of a 1kyu against a 3kyu would be less than that of a 7d against a 3d. It's generally assumed that the stronger the players, the higher the winning percentages of the same difference in stones, and the more certain the strength level the stronger the players.

However, if we assume that the winning percentage of a player a certain number of stones stronger is the same across rthe board, then we may come across a slight problem.

If we were to take the winning percentage of a 1k v a 3k, and assume the same for an 8d against a 6d, for example, what could occur is the the rating system believing the higher winning percentage of the 8d against the 6d translates to a higher stone difference between the 8d and 6d than the 1kyu and 3kyu, resulting in a slight inflation of the stronger players rank in dan levels. Therefore, a kgs 10d may not necessarily be 3 stones stronger tyan a kgs 7d in real terms. and using the quota, "100 points of rating for every difference of one stone" may not be quite accurate, as 100p difference refers to a winning percentage, I believe...

Secondly, The Go associations in China, Korea, Japan Taiwan, Europe and North America have different systems in recruiting new professionals. In fact, the Kansai Kiin nominates professionals differently from the Nihon Kiin, and I'm not sure if Taiwanese players must pass the general Chinese professional qualification to become a pro in Taiwan.

Ali 1p and Pavol 1p would not have appeared to have unusually weak strength for a 1p pro in Japan by any case (although they played relatively few official games), according to what I have heard, and taking from that, the majority (at least half) of those taking the recent Winter examinations could say the same-- lthough this doesn't mean the passers were not clearly stronger than the other pro-level players. And if we were referring to 1p in Korea or China, it would be a completely different story.

What may make this more complicated is something I saw on a Chinese weiqi site, along the lines of: "If A > B, and B > C, is A really > C?". It could actually be the case that a certain player may appear weaker than all of the other players in a certain circle of pros, but hen suddenly outperform them in a broader circle of pros.

What is meant by, "Asian counterparts"?

Bill Spight · **#26**

Elom wrote:

The relation of a single stone of difference to a specified rating difference seems to be a measure of convenience.

Most modern rating systems are built on even game matches and winning percentages. IMO differences in strength measured in handicap stones are quite accurate, even with vast differences in strength. And it is possible to have ratings based upon handicap play, but I doubt if the theory has been much developed.

For instance, if a shodan and a 2 dan play even, maybe the 2 dan will win 60% of the time, but if a 10 kyu and a 9 kyu play even, may be the 9 kyu will win 53% of the time. Still, in both cases if the stronger player plays White and gives komi, the results will be roughly even. Also, even with the great strength difference between shodan or 2 dan and 10 kyu or 9 kyu, the rank difference will indicate a reasonable handicap.

IMO, a rating system designed to predict winning percentages in even games is not as appropriate for go as one designed to predict 50-50 handicaps.

Bill Spight · **#27**

xed_over wrote:

While we have tendencies to rank people and list some as better than others, other cultures not so much. At last year's US Go Congress, Maeda Ryo 6p of the Kansai Kiin told us during one of his lectures that they don't consider lower ranking pros to be any weaker than higher ranking pros. They consider them all on an equal level.

Of course, we all know that actual results speak louder than that, but that's the attitude (and respect) they have toward their professional players of all levels.

I expect that Shusai had a certain regard for Go Seigen, but when he gave him three stones I doubt if he thought of him as even close to his own level.

EdLee · **#28**

Quote:

They consider them all on an equal level.

Of course, we all know that actual results speak louder than that

If they do, then they have an interesting way of showing it by the teaching fees they charge:
in Japan and China at least (not sure about in Korea), various pros routinely command much higher teaching fees --
50% or higher than other pros. An interesting idea of equal.

Mike Novack · **#29**

EdLee wrote:

Quote:

They consider them all on an equal level.

Of course, we all know that actual results speak louder than that

If they do, then they have an interesting way of showing it by the teaching fees they charge:
in Japan and China at least (not sure about in Korea), various pros routinely command much higher teaching fees --
50% or higher than other pros. An interesting idea of equal.

Ah, but that is a very different matter. Just because pro A is stronger at playing go than pro B does NOT mean that pro A is a better TEACHER of go than pro B. As long as the teacher is a significantly stronger player than the student, how much the student benefits will depend on the teacher's teaching skills, not go skills.

EdLee · **#30**

Mike, yes, that's common sense.

I wish I had taken photos of the price signs at the Tokyo Go salons in 2011.
The point is that the higher prices are based (partly? primarily?) on their ranks (or current titles), which, as we both agree, are not necessarily corelated to teaching levels.
And yet they were apparently priced as if they are.
So no, it's not a very different matter.

Uberdude · **#31**

ez4u posted some nice statistics of Japanese pros ranks versus ratings (from marumaru's site). It didn't make any connection to handicaps, but iirc the average rating of a 5p was lower than a 4p. And there were those young upcoming pros with low ranks who were rated above most 9ps.

Edit: here is the link viewtopic.php?p=175541#p175541 and Kyo Kagen 2p had a 83% chance to win against an average 9 dan.

hyperpape · **#32**

xed_over wrote:

While our newly crowned, first ever, American pros had earned the title of "professional", they were still a ways off from being as strong as their Asian counterparts. That was part of the point of the match, to test just how far off they were.

I'm not sure you'll want to use these particular examples in your quest.

1. I subtracted a stone.
2. Taranu participated, and while he's out of practice, I believe he earned his professional status in Japan and promotions without special dispensations.
3. Andy just won four in a row against lower ranked Kansai professionals.

I feel comfortable wagering that the three players who faced off against Lee are not (on average) more than one stone weaker than the bottom 10% of active professionals in Japan.

Edit: added "on average"

ez4u · **#33**

EdLee wrote:

Mike, yes, that's common sense.

I wish I had taken photos of the price signs at the Tokyo Go salons in 2011.
The point is that the higher prices are based (partly? primarily?) on their ranks (or current titles), which, as we both agree, are not necessarily corelated to teaching levels.
And yet they were apparently priced as if they are.
So no, it's not a very different matter.

Here are the current prices charged by the Nihon Kiin if they dispatch a professional as lecturer to you club, study group, etc. This is their advertisement in Go Weekly. The left-hand column shows the rank and the prices are to the right, e.g.: from shodan/2-dan for 25,000 yen (for a 3-hour scheduled lecture) up to 9-dan for 50,000 yen. Of course pros are much more flexible for privately arranged lessons without the Nihon Kiin standing between. ;-)

I wasn't present to hear Maeda's comments but I have to speculate that he was referring to the reform of the preliminary tournaments that eliminated the formerly privileged position of the higher ranking pros. In the good old days they were automatically seeded into the later rounds without having to hobnob with the kids and less successful colleagues. That went by the board around the time that the Oteai was reformed. I don't know whether it was officially connected or not. JF could fill us in if he were still posting actively. :sad:

I don't know any professional that thinks all J pros are equal, except maybe administratively.

gowan · **#34**

I don't know how it works in Korea and China but in Japan pro ranks never decrease. That is no one is demoted in rank. Thus the rank at any point in time represents the highest rank the player achieved up to that time. Actual playing strength might be lower than the current rank or, in the case of advancing players, the current rank might be lower than the actual playing strength. Ratings (numerical) are closer to the current actual playing strength if ratings increase or decrease according to playing results.

Among a community of go players of a variety of playing strengths handicaps are intended to create a level playing field and a player might have have different handicaps with two players who are equal in strength. For example B and C play on even (same rank) and A gives two stone handicap to B and four stones to C. Note that in a club, for example, handicaps change according to various rules. One common one is handicap changes by one stone after three consecutive wins. When pros used handicaps in Japan the handicap changed when one player got ahead in the series by four wins; this was how the playing conditions changed in the famous Go Seigen uchikomi ten game matches.

EdLee · **#35**

Quote:

I don't know how it works in Korea and China but in Japan pro ranks never decrease. That is no one is demoted in rank. Thus the rank at any point in time represents the highest rank the player achieved up to that time.

This high water mark system exists not only in pro Go but in many other areas as well.
During the teens and early 20's, a person can graduate from a 4-year university with a bachelors degree, go on to get a PhD and eventually retire as a professor emeritus. But between the ages of 90 and 94, she may not be able to complete the identical 4-year curriculum to obtain the same bachelors degree any more. (Or maybe she can.

)

Elom · **#36**

Bill Spight wrote:

Elom wrote:

The relation of a single stone of difference to a specified rating difference seems to be a measure of convenience.

Most modern rating systems are built on even game matches and winning percentages. IMO differences in strength measured in handicap stones are quite accurate, even with vast differences in strength. And it is possible to have ratings based upon handicap play, but I doubt if the theory has been much developed.

For instance, if a shodan and a 2 dan play even, maybe the 2 dan will win 60% of the time, but if a 10 kyu and a 9 kyu play even, may be the 9 kyu will win 53% of the time. Still, in both cases if the stronger player plays White and gives komi, the results will be roughly even. Also, even with the great strength difference between shodan or 2 dan and 10 kyu or 9 kyu, the rank difference will indicate a reasonable handicap.

IMO, a rating system designed to predict winning percentages in even games is not as appropriate for go as one designed to predict 50-50 handicaps.

Oh yes, I mean, using a direct conversion of a rating difference in rating systems (like the EGF rating system where 100 points of rating is often assumed to correspond to one stone in strength) to correspond to a stone in strength is were skewing may occur in the top-bracket of strengths. Because a one stone difference doesn't refer to winning percentages by themselves, but imply that if A gave B xxx stones, the winning percentage would be 50%

.

Paying mind, different rating systems may give slightly differing winning percentages between A and C relative to the winning percentages between A and B and B and C in any case, as well-- the main purpose is simply to produce a ranking table, so this small disagreement doesn't seem not so vital concerning rating systems-- unless one maps rating point directly to stone strength without any type of coefficient based on the actual rating of the players themselves :blackeye:

.

ez4u · **#37**

I just wanted to add this chart from mamumamu's site. It shows the average change in rating by age in his domestic database. On average, pros in Japan have stopped adding rating points at about the age of 30, held relatively steady until about 40, and declined more rapidly from then on. Both ends of the chart have only a few people in the numbers but from 20-50 at least he is averaging across more than 100 (and sometimes nearly 300) people.

Apologies for the size of the image. I did this from the iPad and did not have the right tool handy to resize it.

The strength differences between professional levels is...

Who is online