Given that there is no restriction on the strength of people who join the ASR I note that the people who join the ASR do so essentially in direct proportion to the population distribution of playing strength on KGS.

It is then noted, given the point system, that the distribution of players in the top rooms trends towards the natural distribution of population by playing strength on KGS.

To see that this is true assume that out of all league participants people play a random number of games between 0 and 39; or just assume there is some average number of games and then assume that win ratio is also random.

The most active people will tend to play around 15 or 20 games; at this level the difference between 67% and 82% is a 10-5 record versus a 12-3 record OR a 10-9 record, since win ratio based on wins is countered by double the number of losses and slightly more games; that is to say the system FINDS slightly weaker players who are also more active because more of them tend to join than active strong players.

CONSIDER; 60% of the players on KGS are between 8k and 1d. Therefore 60% of the most active players in the league are between 8k and 2d. Said another way, 60 shares of players are between 8k and 1d and 10 shares of players are 2d and up. Therefore it is six times more likely that a weaker player will be active, and therefore such players tend to drown out stronger players. In fact if we consider 30k to 2d we find that the factor is nine times, not six. But we are assuming players 10k and under are simply not strong enough to win at all.

Gresham's Law Revisited:

Recall the laws of probability; a 3d is a 3d on KGS because he has demonstrated approximately a 67% or greater win ratio versus a 2d when no handicap has been used. It's 81 or 82%, I think, for 2 stones.

Using statistics we see that if one 3d was in a room with six 1d players, he has a ~70% chance of losing to at least one of the players [thanks Joaz Banbeck -- I originally thought it was 22% x 6]

Do you know what this means?

It means that the band of 8k to 1d which is six times more likely to appear in the league than 3d and up, will DROWN OUT and statistically REMOVE higher rated players! 70% > 2/3.

Effectively, the activity reward of 1 point for a loss counteracts 2 stones in strength for a less active player.

It will systematically force higher rated players out of the TOP positions of the league because statistically they are SIX TIMES more likely to be "not as active" as the weaker players just below them. Their statistical expectation for winning IS NOT ENOUGH to counteract the fact that there are six times as many of them!

There is a hard limit to this! the average 6d is about 1 in 200 go players. In a 200 person league one will therefore expect one 6d, but this player will be drowned out by many lower rated players who are much more likely to be active. It is therefore a prediction of this that the strongest players in alpha given the current system, will be no songer than 3d or 4d.

If you think this is a problem the only way to get around it is to cut the points awarded for losing a game by, say, half, or start using win ratio/SODOS and mandate a high minimum number of games.

-

i'll skip the point where i ask politely what kind of drugs are needed for this one and try to find some other kind of answer...


If a strong player is less active then a weaker he might place below that one in the league.
That's a fact in ASR League and like said in other posts, it might be discussed if this is the best way, but right now it is the one used and i see more positive then negative effects...

Anyway, the strong guy will have problems getting first place if he plays less then others.
Now there are reasons, statistical facts that prove he WILL be less active (statistically).
And now, since it's not the fault of the strong guy, but of the statistic the league has to be changed, so with the statistical disadvantage the stronger guy gets the same chance for first place as others?

.

.

.

sry, i'm still starring at my monitor and can't belive it....

i guess, since statistical ppl get many different illnesses, chocolate should be banned (eating less isn't an option)
i wonder why never someone with a bigger weight is allowed to start 10m before the others at 100m sprint competitions, statistical he has lower chances to win.


well, honest...if a 6d wants to win he has to play enough, he won'z win cause he is 6d and the rest is weaker, the league is for playing and the more games are played the better, so this kind of "forcing" the strong ppl to play, cause weaker ppl can get good results by playing much (and the weaker ppl are forced too, they want the better result if stronger ones aren't that active)

You assume that people who join the asr do so in direct proportion to the distribution of strength on kgs- but I don't see why this should be. What reason does a 6d have to join the league? Pretty much all of the players are weaker than he is, and he won't get anything out of playing them. The prizes aren't that great for a 6d either, except the top of A class which would take several months to reach. It would simply be a waste of time when he can automatch against other 5-7ds. However, kyus and to a lesser extent lower dans can get something out of the league, which is why there are so many of us - kyus get teaching games with dans, dans get to stomp kyus in teaching games (well, I like to... ) and we can all compete for prizes.

Well it could be that I'm just too sleepy and not thinking straight.

My idea was to try and mathematically quantify exactly how many stones of strength are counteracted by someone being able to play (and lose) say 3 or 4 more games than the most active person below them.

It may be best if I could just show you what I mean, so i've decided to test my strategy live this month in the ASR. I am quite tired right now, but the plan is to take advantage of the statistical knowledge I came up with and determine how many wins I need, and then from that point just lose as many games as possible. If I can do this, then the meaning of my words will be clear. If not then I will be the first to admit I was wrong.

Other than that the main difference we have seems only to be of opinion.. you fully understand that hyperactivity is rewarded in this league. For example last month I had 21 points in beta I. But if I lost all my games I would have had more points. The real difference between us seems to be that I view it as a serious problem and you don't seem to mind it as much.

Dunno when people with 50% win ratios can surpass people doing 80 and 90% consistently, I just think it's broke.

-

If I understand the situation correctly, I think that the math is wrong. If you are saying that a 3D has an 81 to 82% chance of winning against a 1D, and that he plays each of the six of them once, then he has approximately a 70% chance of losing at least one of those games, not 22%.

Let us average 81 and 82 to 81.5. If there are six discrete events, each with a 81.5% chance of happening, the odds of all six happening is 0.815 ^ 6 or 0.293. The odds of that not happening are 1-0.293 or about 70%.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

That 67% win ratio doesn't look right; I mean a 3 dan could easily beat several 1-3 kyu's without losing, even if theres that 5% chance everytime, and I'm sure if he plays 21 1-3 kyus he'll still win them all (of course this isn't true 100%, but at normal circumstances a 3dan probably wouldn't lose to a 1-3 kyu); anyways lower dans get something out of this; the reward in alpha, and kyus get teaching games; I think it's just how badly one wants to get their own reward

That's not true either. He doesn't even have a 50-50 chance of winning them all. When you have a 5% chance of losing each game, the odds of winning 21 straight are 0.95 ^ 21 or about 34%.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

Well I guess player styles are more important, I mean if a 3d is really good at fuseki / yose, but really bad at fighting, a 1d player really strong at killing might even have upper hand.. so I don't think u can count with percentages ><

I not only don't mind, i also think it's good this way.

weaker ppl may try to get a good result by playing much.
stronger ppl are "forced" to play much (or will have less points then the weaker).

Having everyone play as much as possible is one goal of the league, it's not playing a few games and who is strongest will get the price.

The system still rewards winning more then losing, so it's not only activity, but activity is a major part of winning the league.




In that case the one with 80-90% wins just didn't play enough. he could have surpassed the one with 50% with less games, but still some are needed.

If ppl are equal in strangth the one more active will win (should be logical)
If they are not equal the weaker one can try to compensate it with more games. If losses would not give points ppl would not dare to play someone stronger, if they want to get a price/promotion, cause the stronger one may rival for that goal. So losses must give points if the league wants us to play and not to think about who we can affort to play and who not...
So a gap in strength can be compensated with more games, the bigger the gap the more games are needed and as long as there is a gab the stronger person can use it and play more too, so he will win in the end.

39 games is maximum for a month, i think 1 game/day is very active, but still something managable.
Sure some ppl are not able to play this much, but so they won't be able to promote/win prices, i don't see a problem here, it's about playing and playing will be rewarded.


//edit:



If it workes you only prove there are no ppl who play better then you and are more active, there is no number of games u hvve to win, only a number of points that is often enough to be promoted...statistical...

Hey, you claimed the numbers, not me. I just showed their logical conclusion.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

I wanna emphasize that if i see ANYONE losing games on purpose to prove a point i see it as my duty to kick you from the league. Losing on purpose is another way of cheating (you change the final outcome of the class) and will not be tolerated.

leagues, rules, players...they are never perfect, however we can tweak the system so we can improve on the league we have now and proving a point can be powerfull enough on the forum and the league should not be a testing ground.

It should be a place where we all play seriously and have fun doing it!

_________________
admin of the ASR league and KGS admin

It would indeed be nice if we could play serious games in the league but why waste all my energy on 16 or 17 serious games and win them all when someone will come along and play 25 or 30 games, and outweigh me?

The problem that Gabalon (for example) does not seem to understand, is that by that point it is no longer an issue of activity or playing more. 16 or 17 games is a lot. This month I'll just try to play 30 games if I can and I won't care if I win or lose. I'll probably end up with a 40 or 50% win ratio.

Look, I am just responding to what the system mandates as a #1 player. If you don't like someone like me "taking advantage" of the system then how precisely should I play in order to win the league WITHOUT taking advantage of the system?

Are you familiar with the term Nash Equilibrium? I would think so. If the Nash Equilibrium is that people play quickly and just try to get as many games as possible without a serious thought to their win ratio, then the problem is with the system and not with the attitudes of the players or the administrators.

-

The reason i don't see the problem you try to point out as one is, that i belife it is possible to play even 39 games per month and take them serious.

It's not black and white, more then 15 games per month = not serious playing.
Sure it's active and sometimes u won't play in your best mood or have it noisy around you, but it's impossible to ban "playing in less then the perfect state", but that doesn't mean one can't learn from such games or they are bad for the league.

usage...somehow u were able to play 48 league games 2 month ago (and u joined at least one week later, so 48 games in 3 weeks or so) and now u wanna tell us playing more then 15 games a month is hyperactivity and can't be managed if you have to play serious?

And for sure the #1 has to play many games, if someone is strong and other league mates are very active he will need 25 games at least.
But again, i don't see why this should be a problem, league is about playing and if someone wants to be 1st he has to do something for it. If someone doesn't want to play games i don't see a reason he should be 1st. Of course some ppl aren't able to play that much either cause i don't have that much time (in that case i don't think it's unfair if they are not 1st, i hope they are able to enjoy the games, but reaching the top needs activity, not only in Go) or they have problems finding opponents (time zone or other reasons... i think we should try to improve on that part, but it's not easy).

i'm getting a bit dissapointed here, so ill make one final push to my point.

The reason of the league:
the reason i made the league as it is now (i only active in the league since this year so don't even begin about older leagues) is because i think that people should treat online go more serious. By having a larger timelimit and actually making it possible to progress in the league and getting rewarded for it i think it is one of the most succesfull leagues in the history of kgs.

I think criticism is ok and should be heard but i dislike the fact youre willing to lose on purpose. By doing that you shoot down the goal of the league, playing long and serious games. How would you feel if a player played you in the league and you put alot of effort in trying to win while the other player just says to you that he is going to lose because of statistics and points? I would be demotivated and would not enjoy playing in the league anymore.

Think of people and not only stats...

_________________
admin of the ASR league and KGS admin

You misunderstood me I don't intend to lose. I just intend to play first, and win later.

I play to win, but I don't care if I lose, see? This is how the system works. Look. If I get another 93.2% win ratio on 13 or 14 games, and not being able to find a game in the league for literally 2 weeks I'll probably be so pissed I will quit the league. And I know that doesn't matter to a lot of people but it matters to me. So if the system tells me that I have to just play and play and not care if I win or lose (maybe get a 50% ratio) then that is how to play it.

Look just do the math yourself. Make 100 cases with 2 variables; V1 = win ratio, V2=# of games. See which stat is most important. Do it yourself. See what the league wants. See what it takes to win.

If I lose with a 50% ratio, then that will be quite another story; it's difficult to get that many games. But I've basically shown you mathematically the flaws of the league and some people are saying that it is actually better this way. No, it isn't. Please don't blame me for trying to promote.

Anyways as it stands it doesn't look like I will get enough games this month to win either. it's already the 10th and I have less than 10 games :/
-

If we assume that 3d got a 18% chance of losing to 1d we get INDEPENDENT probability (so you can't add them). With every next game probability of losing is the same (not considering mental state after series of loses).

So when our statistic 3d play one 1d player he got 82% chance of winning
If he take another 1d player he still got 82% chance of winning

We got states:
our 3d could lose both games, win one or win both of them:
chance of 3d losing two games = 18% * 18% = 0.18 * 0.18 = 3%
chance of 3d losing one game of two = 18% * 82% = 16%
chance of 3d winning both games = 67%

If you take those six 1d players we got 0.82^6 chance of winning ALL SIX games = 30%
This is approximate simulation, because to get more accurate results we should use probably Bernoulli scheme, but the difference is clear

But we are still considering 3d who is merely a 3d (that means he is in lower part of win/lose ratio required to advance from 2d to 3d). If we consider "strong" 3d the chances are considerably higher.

So if you consider even an army of 8k-2k it would be VERY hard to overtake a 3dan player (statistically) )

Actually, chance of winning one and losing one is 30% (ok, 29% rounded, but just making everything add up to 100%).

FWIW, getting one win against any opponent is the ASR holy grail. Going 0-3 against everyone gives you 22.75 points, going 3-0 gives you 45.5, but going 1-2 gives you 35.75, closer to 3-0 than 0-3. Reason being, 2-1 and 1-2 are both worth a total of 6 points (split between the players), whereas 3-0 and 0-3 are worth 5.25 split.

If you have lost twice to someone, you are at 1.5 points, but with 1.25 points at stake for being able to win that third game (instead of the 0.5 the opponent can get for racking up his third win against you). I'm not complaining about this being inbalanced in any way, but it's an important consideration for those after promotions

Representing it graphicly, and ignoring rounding errors, it looks like this:


Note that there are 2 ways to win one and lose one.

FWIW: The sum of any row or any column is 18 or 82; the total sum is 100.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

If you guys want to crunch the math, you forget the meaning of the league to weaker players.

In a system where winning is rewarded, and where you get the chance to play stronger people, I've noticed the ASR kyus play their games much more sersiously, take their time, and do some actual reading. As someone who is in more of a "teacher" role, this is something i'm glad to see. However, sometimes I notice a 6k will all of a sudden look more like a 2k in game simply because he avoids dumb mistakes. I think the math fails to realize that the big reason kyus are kyus is not taking games seriously, playing to fast, playing without thinking, etc, which are all quickly solved in a league format.

I think, usagi, you and i have a similar time zone problem. When I was in the USA and had nothing to do for a month, i played 51 games, which is active, but not a realistic pace to keep up. The problem I'm having here in asia is that i can't play anyone before 9pm at night, and often times i am out/busy/tired/etc... so the games i do play are bad quality games. I, of course, find this very disappointing.

While it's true, sure, you have to play many games against kyu players to get into alpha, i don't think it's a bad thing... i actually find playing kyus relaxing and a nice thing to do when it's too late to give it 100%.

THat said, I think the real issue is a time zone issue, at least for us, not some sort of strengths issue. It's nice to be in alpha and play other 3ds/4ds, but I think the stronger players won't bother, because it will essentially become just an endless string of T games for them.