Ya, teaching games are different than normal games, it’s not about winning and trick plays, but giving lessons. And playing against solid shapes already existed on the corners is a good way to teach students the way of influence of course.
I think that’s what I said, so I agree with you, the trick fighting move at dan level handicaps, matter less, that would require all-around better to be able to play high handicaps, from opening to using influence, better at reading semeai, even yose. So, ya, the handicap stones pass a certain level like 5 or 6 would almost make it impossible to win where both sides know how to play solidly when the weaker party reaches “dan level”.
If you are able to grasp all that and execute them well, I believe you already are about dan level. I wonder how sdk players would play with very high handicaps games even if they don’t study them. Can you still stuck in the basics and sdk or ddk and still got familiar with handicap games? Or are they actually hinder their play if they are so used to handicap assistant?
Oh yes indeed, yeah after 6 handicap it does get shaky.
What I ended up doing is weighting things, basically handicap 1, 2, 3 were the primary weights and “worth more”, by the time I added handicap 9 it was factored in but only at like 10% the weight of handicap 1 if I recall.
When I was still an active player in a regular go club (a few decades ago), casual handicap games were the norm in clubs here. Non-handicap games were usually only played between players who were close in rank. Many clubs in the West are fairly small (less than 20 players come to play), so handicap allows club members to play challenging games against a wider variety of opponents.
In my childrens club I play handicap games almost exclusively (including large handicaps on small boards). Games amongst the children in the club are usually handicap games, so I can use the game results to rank them relative amongst eachother and all children get to experience winning and losing.
In my experience, handicap works very well for that at all levels, on smaller boards as well as on larger boards.
In my experience, handicap games generally don’t progress like white overplaying all the time and trying to kill some big group. A handicap game can progress like that, but it doesn’t have to, especially when white has a lot of experience giving handicap.
When I’m white and the handicap is correct, I can rely on my opponent playing many moves that are unneccessary or too slow or losing points, moves that don’t work, bad exchanges etc. So I can play fairly calmly and properly, trusting on black’s initial lead to become smaller and smaller as the game progresses. Such handicap games tend to end up being quite close, with me or my opponent winning by less than 10 points.
If white needs to use every dirty trick in the book to have a chance, I would say the handicap is too large and a promotion of the weaker player may be due.
In the Netherlands, there are some handicap tournaments where white gives the traditional handicap according to rank difference. Sometimes a kyu player wins such a tournament.
BTW, I tried to find some high handicap games on currently listed correspondent games, and I think there is some bug in filtering. Have to uncheck the unranked, or non-handicap games would mix in with them.
Being the top rank is different than being a skill ceiling, though. An amateur that can give the average 9d a 2-stone handicap is still a 9d. If an Amateur played at the level of Shin Jinseo, they’d still cap at 9d, even though they’d be something like 1,000 Elo ahead of their peers. That’s the convention for amateur rankings almost everywhere: there’s an arbitrary skill level that 9d is pegged to, and that’s the highest rank that’s used.
It’s also what’s used on OGS. If you look at Sadaharu’s rank history, they were 9d at 2700 and 9d at 3000 because 2700 is the point where the OGS system stops counting Dan ranks, because tradition.
@anoek What at are the chances that with the rating adjustments and all done, that blitz games can also be counted toward your blitz rating? (Specifically in 19x19)
Both the automatic sitewide blitz tournaments and the auto match blitz count as live games.
I thought it was just cosmetic, that the game history just put the wrong icon there but actually it just doesn’t count any of those games as blitz in the blitz rating column for 19x19.
(I also think, that it retroactively discounted games that were counted before as blitz too.)
The expected winrate with (H-0.5) depends on the level of the player. It should correspond to the winrate calculated from about half the rank Elo width around the player’s level.
So in the 1d/lower dan range, where the rank Elo width is about 100, black’s expected winrate should be about 43% in a 1 handicap game (½ rank = 50 Elo ~ 43%) between players separated by 1 rank.
But around 15k, the rank Elo width is only about 50, so black’s expected winrate would be about 46% (½ rank = 25 Elo ~ 46%).So for black to win only 43% of 1 handicap games around 15k, the players should be 50 Elo apart after correcting for the ½ rank handicap. So the players should be separated by 1½ ranks, not 1 rank.
So if black has 43% winrate in a 1 handicap game against a white 15k, you might conclude that black should be ranked 16k, but I would argue that black should be ranked 16.5k.
I didn’t really notice that black’s win rate decreases in the EGD as handicap gets higher. But maybe that is because I corrected black’s expected winrate for under-handicapping by 2 stones, which is common in European MacMahon tournaments (enabling me to include this data, because I wanted to keep as much handicap game data as possible). But ofcourse it’s also quite possible that I made an oversight.
Do I understand correctly that you excluded those underhandicapped games from your sample?
Sorry, I have only barely skimmed this thread, so it might have already been discussed, but I just wanted to note that @S_Alexander created another thread looking at how people’s ranks changed here:
Here is another visualization and some observations
This is very interesting. And I feel there is some kind of Go cultural difference. Learning in a place where there are lots of players and kids of similar level with rank segregation in different classes, there is almost no need for handicap games to become the norm, even if there are handicaps, it’s usually within 3 (since a class is divided every 3 ranks). And when I think of high handicap games over 5 or 6, I can only remember them being teaching games. They tend not to play out to the end, and only being given by Go teachers, and students would know they lost when the teacher told them it is over (by pointing out the mistakes and reviews during the process). And our in-school practice tournaments are trying to mimic real local tournament rules and settings, and they tend not to be handicapped.
And usually, it is the dan level ranking tournaments that would use handicaps, since actual dan level players would be much fewer to held a 30+ players swiss tournament, hence mixed dan level matches would sometimes be needed to invite enough players to join. And by that level, most dan players would already study handicap games and wouldn’t try to over-play (and usually would already have some experience play teaching games with the lower level class students). Still, there are rarely high handicaps (5 or 6 above) in actual tournaments.
I get a feeling that a lot of the “dan player” here might not actually be as strong as AGA or EGF high sdk players who participate lots of tournaments already, and our kyu players are actually much closer in their level with each other since we tend to study with the same textbooks and learn the same basics, the difference is the number of practices to make use of this knowledge. And our beginners (15k and below) are actually very far spread out in their strength and actual knowledge. They rank up within class by taking exams and learn one section of the basics at a time, before actually able to finish a game, they might already be good at life and death, before knowing anything about fuseki.
I am fully aware of that, and if you read any of my other posts, I have mentioned
as well as many times distinguished between “rank ceiling” (the highest skill level achieved by humans) and “skill ceiling” (the highest skill level achievable (sometimes considered to be theoretically perfect play)
And it is true, that 9d no longer carries the weight of the original – that is, only for the meijin godokoro – but the text presented is still clear and true: it is simply the rank of the best (amateur) players currently playing.
another thing I will mention is that I don’t believe there is a way to quantify a “skill ceiling” in terms of elo/glicko, as it would change depending on how good the average population is.
I’m referring to the data from Winning Statistics | E.G.D. - European Go Database which as I understand it, is still the soon to be old ratings? The decrease I’m referring too, if I’m understanding things correctly, is that decreasing trend of the top most numbers, the 42%, 39%…32%, 28%. One of the primary things I was optimizing for was to get those values to all be similar, so more’less 43% accross the board (with the inevitable exceptions).
43% for no komi is about what we should expect. Then it should go down for 2 stones due to the weaker player (on average) being less familiar with opening hoshi. Then the 5th, 7th and 9th handicap stone have a big jump because they are tengen.
I think trying to fit for 43% on all of these would be a mistake.
edit:
Blacks winrate decreasing for 8 and 9 stones makes some sense too because they often involve handicap style fuseki that might be learned in SDK range. So a kyu giving 8 or 9 stones is likely to use it vs an unsuspecting opponent but a dan giving 8 or 9 might run into opponents who know handicap fuseki.
I don’t really have a handle on the discussion at the moment. I can’t tell what is just semantic/definition or what has any measurable difference (if 1 rank really means one rank or 1.5 ranks when assigning ranks to rating based on 1 stone (0.5 komi), 2 physical stones etc).
Is the above table you’re commenting on supposed to contain the winrate (percentage) for handicap games? If one doesn’t aim to keep the winrate the same for multiple handicap differences, surely the handicap isn’t proper then? If I am 5 (or 5.5 - it probably doesn’t matter) ranks weaker than another player and we play a 5 stone handicap game, but I’ve only 9% chance of winning, that’s hardly a proper handicap, and therefore, surely, it shouldn’t enter into the rating calculation as an even game (as in rating adjusted before fed into the elo/glicko/whatever system).
There are fundamental differences between each handicap so you can expect some variation. The 43% number is somewhat important because that basically defines how wide one rank is. And 43% roughly correlates to AGA/EGF/KGS ranks. That table is showing that a 5 handicap game might give black a 34% or 35% chance to win (rather than 43%) which makes sense based on the reasoning in my last post.
Right, they should be treated differently from even games.
For what it’s worth, I biased the optimization heavily on handicaps 1-3, moderately on 4, and then it dropped off pretty quickly. The weights I used (which were totally just shooting from the hip so to speak) were 1: 1.0, 2: 0.9, 3: 0.7, 4: 0.5, 5: 0.3, 5: 0.3, 7: 0.3, 8: 0.1, 9: 0.1. So we we didn’t particularly optimize for higher handicaps nearly as much as the more common handicaps, but they were a consideration, and I think it worked out decently.
I didn’t even notice this handicap panel of the EGD statistics page before.
But I don’t quite understand what I’m seeing there. At the bottom of that page it says “N.B.: Games with handicap greater than the difference of grade between opponents have been ignored”. If that were true, then how can there be numbers below the diagonal H = Gr. Diff?
Also, I wonder if handicap games crossing the EGD 20k “purgatory” were filtered out. If you have a handicap game where black is an EGF 20k, I can understand that black wins less than expected, because black may well have been weaker than 20k in reality. If he registered and played as a 25k, the EGD just updates his rank to 20k.
I’ll look up some other statistics, but I have to work now.
Hi @gennan - since you seem experienced with IRL ranks, i have a question:
I am about to enlist in the BGA open tournament which is to take place this month. I have never played an official EGF tournament before and i am wondering what rank to assign myself. My new OGS rank is 6k - so would you suggest to take around 10k for registering? I feel i am a little stronger than 10k, i was around that before the ranking update… it just feels a little off registering at 10k while the account that i am going to play with shows a significantly higher rank. What are your thoughts? When is the EGF update applied and how do the new OGS ranks relate to the current EGF ranks?
