EGC 2015 Weekend Tournament

[RobertJasiek]

The Weekend Tournament of the European Go Congress 2015 had 504 players in the list. It was a 5 rounds McMahon with the tiebreakers SOS-SOSOS, a supergroup of 24 players with a rating of at least 2542, a bar-1 group with ratings 2450-2541, a bar-2 group with ratings 2350-2449 etc, 0.5 points (rounded down) for missed rounds, pros called "8d", unknown-rated players getting default ratings (such as 2700 for 7d or 2600 for 6d, and those top Chinese players proved their strengths in this tournament) and AGA Rules.

In violation of the EGF General Tournament Rules §3.2.1 ("By default, players in the top groups of a tournament must play all rounds. In a McMahon tournament the top groups include any supergroups, the group above the bar, and the group just below the bar. A player starting in a top group may skip a round only in exceptional circumstances like a serious a medical problem proven by a doctor's certificate or a family member's decease."), top players were allowed to drop rounds. 8 players of the supergroup did so, usually as soon as their losses indicated that they could not compete for the top places any longer. For example, Remi Campagnie, Tanguy Le Calve and Benjamin Drean-Guenaizia dropped the tournament after their first round's loss. For a 5 rounds McMahon with SOS as the first tiebreaker, this is very bad for two reasons:

a) Dropping rounds greatly affects the order of the top players with 4 wins following the winner, who has 5 wins.
b) Other participants, who would have the intention to compete for top places seriously, are prevented from their chance because they cannot be in the supergroup (or top group) of limited size.

Such ought not to occur. Top players must play all their rounds seriously!

The top places were as follows, where "NA" means "number of opponents not playing all rounds":

Code: Select all

Place Name              Rk Co MMS Wins SOS NA

1 Kim, Young-Sam        7d KR 31  5    145 3
2 Dinerstein, Alexandre 8d RU 30  4    146 0
3 Chan, Yi-Tien         7d TW 30  4    144 3
4 Lisy, Pavol           8d SK 30  4    143 1
4=Ma, Bin               7d CN 30  4    143 0

The players sharing place 4 also had the same luck in the SOSOS lottery (drawing the number 714). The relative order of the first two places was decided by their mutual game in round 5. The relative order of places 2 to 5 (the shared place 4) or, had Dinerstein beaten Kim, of the other group of top 4 wins players was decided as much by opponents missing rounds (and the tournament organisers allowing this at all) as by the tournament-strengths of the opponents performed in their played games.

Young-Sam Kim is well known at European Go Congresses. Surprisingly, there were no other Korean 7d competitors this year. Kim's tournament win could have been predicted by his rating 2791 while the second-highest rating at the start of the tournament was only 2711.

**************************************************************************

Players below the top with 4 or 5 wins:

Code: Select all

CZ 13
DE 10
FR  9
UA  8 (of which are two win 5 wins)
JP  5
AT  4
HU  3
NL  3
CN  2
CH  2
ES  2
SK  2
XY  1 (9 other countries)

Among those with many such players and not too many participants, there are, you guess it, CZ 13, FR 9, UA 8. The same countries (apparently) relying on the EGF rating system to set ranks and therefore conservative ratings. (Poland had only 10 below-top players in this tournament and this may explain why its players did not score exceptionally.)

The bar-1 group consisted mostly of 5d players and two 6d, who did not show their greater strength in this tournament. The bar-2 group consisted mostly of 4d players with these rating exceptions:

Code: Select all

Place Name              Rk Co MMS Wins

29 Jasiek, Robert       5d DE 28  4
30 Cheburakhov, Andrej  5d RU 28  4
50 Rehm, Robert         5d NL 27  3
56 Halchenko, Mykhailo  5d UA 27  3
85 Shakhov, Kim         3d RU 26  2

I am not surprised by players #29 (who, on the Friday evening before the weekend tournament, got rid of his cold affecting his play and, e.g., two 1/2-point-like losses during the first week of the Main Tournament) and #30 but #56 was stronger a couple of years ago. IOW, the ratings' wisdom does not in general exceed the players' knowledge about their own strength expressed in their self-declared rank. All we can say about ratings below the very top players is that there is some good correlation on average (although maybe shifted by 1/2 or 1 rank in countries with self-declared ranks). However, for the bar-3 group and below, the correlation is much weaker because various ranks can be found within a range of 100 rating points.

Does using ratings for forming McMahon groups provide an advantage or disadvantage compared to not using ratings at most other congresses (i.e. except for Bordeaux)? Neither removes the rank difference of ca. 1 between countries not using versus using the rating system. Neither avoids exceptional players finding themselves in the wrong start McMahon groups. Which system (self-ranking or rating) is better cannot be answered easily, but IMO a combination of both systems applied carefully on a per-player basis might reduce the number of exceptions. However, for big tournaments with several hundred players, such would require meticulous organisational preparation.

Oh - my losses in round 5 and in the 9x9 KO stage against Mateusz Surma: I had not seen his play before, but now I understand why he could become an EGF professional. He has a very profound knowledge and calmly assesses every position's aspects. His play is not flawless but he ensures to let his mistakes impact a game less than his opponent's. In the 9x9 game, he won mainly because he demonstrated that one should not make an arbitrary second black move of the game but choose it after considering the best replies.

http://egc2015.cz/results/wallweek

[RBerenguel]

Robert, do you have records of your (or others') 9x9 games from this tournament? I always like seeing high quality 9x9, and the EGC site does not have any records (so far?)

[RobertJasiek]

I have not recorded any side tournament games this year (only a few of my own 19x19 games). Maybe I could reconstruct some from memory but lack time to do so. During a few earlier years, I recorded some 13x13, 9x9 or lightning final games and part of those were printed in the congress bulletins, so you might start collecting them...

[RBerenguel]

I have not recorded any side tournament games this year (only a few of my own 19x19 games). Maybe I could reconstruct some from memory but lack time to do so. During a few earlier years, I recorded some 13x13, 9x9 or lightning final games and part of those were printed in the congress bulletins, so you might start collecting them...

Pity. It's tricky to get hold of those, since I'm not so in contact with most Spanish players who go to EGC's. I guess I should come there and record them myself once I'm eliminated

[Calvin Clark]

For a 5 rounds McMahon with SOS as the first tiebreaker, this is very bad for two reasons:

a) Dropping rounds greatly affects the order of the top players with 4 wins following the winner, who has 5 wins.
b) Other participants, who would have the intention to compete for top places seriously, are prevented from their chance because they cannot be in the supergroup (or top group) of limited size.

Such ought not to occur. Top players must play all their rounds seriously!

Interesting. I don't think the AGA has such rules, but I can see how if drops are common, some bad things might happen.

[skydyr]

For a 5 rounds McMahon with SOS as the first tiebreaker, this is very bad for two reasons:

a) Dropping rounds greatly affects the order of the top players with 4 wins following the winner, who has 5 wins.
b) Other participants, who would have the intention to compete for top places seriously, are prevented from their chance because they cannot be in the supergroup (or top group) of limited size.

Such ought not to occur. Top players must play all their rounds seriously!

Interesting. I don't think the AGA has such rules, but I can see how if drops are common, some bad things might happen.

As I understand it, in the AGA tournament placements are distinguished by SOS and further methods, but prize money is supposed to be split between participants with the same record in the same band. So, if there are two 3-1 players in the open division as the best results, one gets first and one second, but they each get half of the combined 1st and 2nd place prize money.

[Celebrir]

I don't think it is good to disallow dropping. Sure, there are those who drop because they have no chance on winning anymore, but there are also those who have a serious reason to drop. Besides I completly understand dropping if you can't win anymore and you have something more intresting to do. For the most of us Go is a hobby and not a job so we wan't to have fun. The possibility of beeing forced to play unfun matches might make some players to not even start to play.

On the other hand, how would you enforce a no-drop-policy? You can't force a person to sit down and play a game you only can keep them registerred in the tournament which can lead to a lot of No-Shows and therefore wins without playing.

About the concerns:

a) Dropping rounds greatly affects the order of the top players with 4 wins following the winner, who has 5 wins.
The No-Shows which might result from non-dropping might affect this order even more. There might be persons who actually only got 2 wins in the first rounds against relativly weak players and got byes in the following rounds to lose against the winner in the final game. It is true that dropping influences the standings, but this would be even worse.

b) Other participants, who would have the intention to compete for top places seriously, are prevented from their chance because they cannot be in the supergroup (or top group) of limited size.
Those who dropped because they have no more chances on winning did compete seriously, but they failed. So whoever would have wanted to compete for the top-ranks has to be stronger than the person who dropped or he/she would fail as well. Since we don't have prelimnary rounds all we can determine initial strength with is the rating.

[RobertJasiek]

No drops are enforced by

1. clear announcements

2. missing dropping procedure (for top players) and organisers enforcing this when somebody tries to drop nevertheless

3. tournament rules encouraging seriously played games

4. signed tournament agreements by the participants

5. penalties for illegally dropped games, such as prohibited participation in the same (kind of) tournament (or its top group(s)) during a few subsequent years

It has worked well in the past: only very few top players dropped rounds in violation of the rules.

[Celebrir]

No drops are enforced by

1. clear announcements

2. missing dropping procedure (for top players) and organisers enforcing this when somebody tries to drop nevertheless

3. tournament rules encouraging seriously played games

4. signed tournament agreements by the participants

5. penalties for illegally dropped games, such as prohibited participation in the same (kind of) tournament (or its top group(s)) during a few subsequent years

It has worked well in the past: only very few top players dropped rounds in violation of the rules.

1. That is something really really really rarely happening at Go tournaments but that could be worked on.

2. I don't think that it is good to create a two class structure depending on whether you are on position 10 or 11 with your ranking. The honly possible difference I could imagine is between pro and hobby player, because pros get payed and it is their job to play.

3. I don't see were dropping goes against seriously played games. Quite the opposite because the player who is forced to play has not the same motivation to play serious.

4. Are we speaking about hobby players or professionals here? I can understand that one has to sign an agreement if some games are streamed to the internet, but for playing this seems to be overkill and at least go against the spirit of the game.

5. I guess what you mean here are No-Shows? The damage is still done to the current tournament and I'm not sure whether those players even would like to return under those circumstances.

And that it worked for other tournaments does not mean that it will do so in the future, especially if the tournaments become bigger.

Besides I still don't see why you think that drops are that bad. Your statistic shows that a player with the same points and 3 "NA"s can have more and less SOS than a player with 0 "NA", so it doesn't seem to be that much of a deal for the final results.

[RobertJasiek]

Enforcing no drops for top groups and, in quite a few years, signing tournament agreements by top group players, worked well for tournaments as big as the EGC main and weekend tournaments from ca. 1996 - 2013. It is no overkill, because it (and the sportsmanlike condition in the tournament rules, and the rating of played games) has had the desired positive effects incl. serious play by everybody playing in the top groups.

1 (or 1/2) SOS difference decides some places 2 to 5. A range from 0 to 3 NA means that NA has an impact on a player's SOS of 0..3, i.e., with an average significance GREATER than that of the minimal SOS difference deciding the places. Hence the places are not decided with numerical significance but are decided more by the NA noise. The NA noise is too great by a factor of ca. 10. Since SOS is affected too much by noise, (top) places with the same MMS should be shared (or dropping rounds prohibited for top players so that NA does not have any (significant) impact).

[HermanHiddema]

With the current size of the top group (24-32 in recent years), the distinction between places 2-6 (those at 4 points) is meaningless anyway. Because of the large top group size, players in those places have not played each other much if at all, and then SOS is mostly a lottery anyway, regardless of dropped rounds.

The advantage of a larger top group is that more players can compete for first place. Its downside is that other top places are random. If you want a meaningful distinction for more than just the first place, the top group should be 8-12 players for a 5 round McMahon tournament like this (and the bar-1 group should be small too)

[Celebrir]

Enforcing no drops for top groups and, in quite a few years, signing tournament agreements by top group players, worked well for tournaments as big as the EGC main and weekend tournaments from ca. 1996 - 2013. It is no overkill, because it (and the sportsmanlike condition in the tournament rules, and the rating of played games) has had the desired positive effects incl. serious play by everybody playing in the top groups.

1 (or 1/2) SOS difference decides some places 2 to 5. A range from 0 to 3 NA means that NA has an impact on a player's SOS of 0..3, i.e., with an average significance GREATER than that of the minimal SOS difference deciding the places. Hence the places are not decided with numerical significance but are decided more by the NA noise. The NA noise is too great by a factor of ca. 10. Since SOS is affected too much by noise, (top) places with the same MMS should be shared (or dropping rounds prohibited for top players so that NA does not have any (significant) impact).

I was talking only about tournaments of EGC-size. This years main tournament had 762 participants, which was as far as I know, a new record. But not "big" for me (I helped with a 3600 player event this year. Not go, but similiar in structure to the point that I already used go tournament software for small private events). go in general seems to be growing faster the last years then before. I would hope we can break the 1000 person mark soon in the next years. That is what I'm thinking of in size. And for this size of a tournament it is just a terrible thought and we have to create the grounds for such big events now in my opinion.

Besides I don't get why you say that someone who drops is unserious or even not respectful (in violence of sportmanship). He contended seriously until the point where he sees that he has no more chance to succed. This principal is called "resigning" and very well known as respectful in a game of go. Why should it be forbidden in a tournament of go? If the tournament is influenced to much by this, then we should change the tournament structure.

But I don't believe that is the case. Let's see the opponents by ranking

2.Dinerstein 6d 6d 1p 5d 7d 0NA
3.Chan 5d 7d 6d 7d 5d 3NA
The others never played the actual winner so were in no spot to content for the top places
4.Lisy 6d 6d 5d 1p 7d 1NA
4.Ma 6d 1p 5d 6d 6d 0NA

Dinerstein had the more difficult opponents and Chan had 3 players with a bad day (if that was the reason they dropped), so Dinerstein is the correct 2.place. Lisy had one more difficult opponent, but one of his opponents had a "bad day", so it is more or less equal and they are sharing the 4th place. Everything working as intended as far as I see it.

This all is leaving out the question of the results if you play a match you don't want to play (which I would not like to do out of respect for the opponent who has to play me in this case)

[RobertJasiek]

Usually, resignation is a per-game concept. Using it as a per-tournament concept is a principle possibility but then it should not influence the tournament results by side effects. The intention of a player's resignation from a tournament does / should not intend to influence tiebreakers. So, if tournament-resignation should be allowed, tiebreakers must not be used or not be opponent-score-dependent.

Dinerstein had the more difficult opponents and Chan had 3 players with a bad day (if that was the reason they dropped), so Dinerstein is the correct 2.place.

If Chan had not had 3 NA but only opponents playing all rounds, he would have had greater chances for greater SOS. Dinerstein's greater SOS was thus also the luck of getting 0 NA. (Besides I agree with Herman.)

Everything working as intended as far as I see it.

Intention?! Rather it think it is far more likely that the tournament's organisers did not properly reflect Herman's and my objections as to splitting places with final top MMS-1 (places 2 to 5) by SOS at all in this tournament.

[Celebrir]

It seems we have two discussions going on here:

Using SOS as tiebraker: I agree that in both cases (with drops and without drops) there is luck involved and it is questionable to use them if there would be good way to split all benefits between all players with the same number of points (e.g. only cash prices)

Allowing drops: I don't see that this changes the chances of luck enough to justify a no-drop-policy. If there were only drops because of bad performance that is it: You played against someone who had a bad day and this is reflected in your SOS.

Also when I said "working as intended" I meant "working as the creators and the most players expect it to work". I don't know how much the tournament organizer thought about this.

[Matti]

I started another thread http://www.lifein19x19.com/forum/viewto ... 45&t=12219 about, whether we could allow top players to drop rounds.