Page 5 of 7

Re: Hnefatafl Internet Championship

Posted: Fri Sep 23, 2016 7:26 am
by Tuireann
BigWurm91 may be new here on aage's site but he's been playing online longer than I have, first on the late kingstafl and now on PTO, and I consider him every bit as good as his rating now reflects. He and I played(and reviewed) about a dozen games of Tawlbwrdd a week in the month leading up to the tournament just as we are playing Copenhagen now to prepare for asgarthr. His success reflects the amount of board time he put into preparing for this tournament.

I think if anything the loss of players in that group made the biggest difference. His rating before the tournament was something in the high 1400s I can't imagine it made that much of a difference in the initial group distribution. Less players make for less chances at mistakes(even some of the great players made some silly mistakes) and that doesn't help weed out the competent vs the truly skilled players when you only have 5 sets vs the 7 sets everyone else plays. Perhaps 3 groups with more players to advance would have introduced more drop out tolerance while not creating too many additional games. Or some form of double elimination.

Re: Hnefatafl Internet Championship

Posted: Tue Sep 27, 2016 5:23 pm
by altti
regarding Valhalla finalists; I was the only one directly affected by bigwurms appearance in the Vinland grouping, as I was eliminated. I hold no hard feelings nor am I surprised that a player unknown to the site apparently upset the outcome. I have been waiting for the day that I would get ousted from that top grouping as I haven't spent enough time studying the game with others. The more people that are attracted to the site the better the play will get.

Re: Hnefatafl Internet Championship

Posted: Thu Oct 06, 2016 6:29 am
by Hagbard
Distribution of players to Championship Tournament 2016 Round 2 groups.

Players in the small forum are discussing the 2016 Round 2 distribution. Here's a copy of the discussion:

2016-10-04 23:09:51 HELFRICH: Curious that I cannot find myself in any category of the final round, or am i misinterpreting the rules?

2016-10-05 01:00:15 crust: Helfrich, you and Vonraider seem to have been left out. I'm sorry, I assumed you had left the tournament, but it seems that a mistake has been made. I hope Hagbard will be able to fix it; it looks like you and Vonraider should both be in the Asgarthr group

2016-10-05 11:58:48 Hagbard: About Helfrich/Vonraider: No mistake. Both will be added as soon as possible, one to Asgarthr and the other to Mithgarthr. But it will take one or two more results of Round 1 games to know who goes where.

2016-10-05 12:46:23 crust: Oh that makes sense - sorry Hagbard, I should have realized that!

2016-10-05 12:49:15 crust: I saw Helfrich had finished all his games, but of course his placing in round 2 depends on the results of Vonraider's remaining two first-round games

2016-10-05 17:17:19 HELFRICH: I would like a clarification on the tie breaker situation. I won seven games by your count, nine by my count. Your removal of two games wins does not seem 'cricket' to me. It is unfair that I or others in my first round grouping should be penalized because some non-serious entry didn't finish any of his games. That player should not have been 'wiped' from the tourney records either: historical records count for future tournaments. As I noted in an earlier posting, some sort of qualification should occur for entry into this championship cycle (like perhaps having played at least 25 games). If you all don't like enacting a qualification, then there should be no penalty.

2016-10-05 21:23:25 Tuireann: It did not penalize you because if you had 2 more points so would everyone else in your group and the standings would be identical at 13,11,11,9(you),5,2,0X with a maximum score of 14. Only the top 2 players from each group advances so its not as if you have to compete against players from other groups with more possible points. The historical records don't count for future tournaments by any reasoning I understand because the group distribution was done by ELO. And the tie breaking is neustadtl score which didn't apply to you because you didn't tie anyone.

2016-10-05 23:47:02 crust: there is an issue, though. Helfrich is now competing with vonraider for a place in the Asgarthr group; but Helfrich is in Heithabyr group (whit 7 members) whereas vonraider is in Tunsberg group (with 8 members). It follows that Helfrich's points are out of a maximum of 12 points, whereas vonraider's are out of a maximum of 14 points. Another way of putting it is that vonraider had two more opportunities of gaining a point than Helfrich had. This wouldn't matter, as Tuireann rightly said, when Helfrich is competing against players in his own group, but now there is competition between players from 2 different (and different-sized) groups for a place in the Asgarthr group. Now, let's say that vonraider loses his remaining 2 games, and remains on 7 points. That's 7 points out of a possible 14, whereas Helfrich has 7 points out of a possible 12. 7/12 is more than 7/14, so it seems to me, in that case, that Helfrich should be favoured. However, if Vonraider wins one of his remaining games, then Helfrich's 7/12 STILL beats vonraider's 8/14, since 7/12 = 1/2 + 1/12, and 8/14 = 1/2 + 1/14. So again, Helfrich wins, even though he has one more point than vonraider! Now, if vonraider wins both his remaining games, then his 9/14 does finally beat Helfrich's 7/12, since 9/14 = 1/2 + 1/7, which beats Helfrich's 1/2 + 1/12. So if Vonraider wins both, he goes through to Asgarthr; if he wins only one, or neither, then Helfrich goes through to Asgarthr. Somebody please check my maths, it's not my speciality! :)

2016-10-05 23:58:25 crust: also where I said "even though he has one more point than vonraider" I should have said one LESS. Just to be clear, if Vonraider wins one of his remaining games and has 8 points compared with Helfrich's 7 points, then vonraider has 8 out of 14, which is still LESS than Helfrich's 7 out of 12. So Helfrich wins, even though he has one LESS point than vonraider. Anyone agree or disagree?

2016-10-06 00:04:55 crust: another way of putting it: Helfrich might have 7 points and vonraider have 8 points, but Helfrich would still have won a greater share of the available points in his group than vonraider had. 7 out of 12 beats 8 out of 14.

2016-10-06 00:14:29 HELFRICH: crust -- thanks for the explanation. I understand your math logic. However, my contention was and still is that the player (who was removed) points should be awarded, thus I would be at nine points not seven. The player who stopped his games was very new at the game. My standing in the group does not change but my point value in tie breakers does improve. Tuireann -- respectively submit that you misunderstood completely as I was not talking about points to get into the top group or the standing I had within the group I was placed in during the first round. It did/does clearly penalize me for tiebreakers. Historically it does count for future reference for the instance of players joining who repeatedly do not finish their games. Showing what actually happened in the tourney is historical and a record should be preserved.

2016-10-06 00:18:04 crust: The problem with that analysis is that we're really not comparing Vonraider's score with Helfrich's, to decide who goes through; we're comparing it to the other players in vonraider's "Tunsberg" group, in fact, to be precise, it's with Steiger that his points must be compared, and since Steiger and Vonraider are in the same group, we don't need to express their scores as fractions. Allocation to second-round groups is determined by position within first-round groups, not by comparing scores across different groups. So I can see that all of my calculations were in vain. But vonraider still needs to win both games against Steiger, to get to Asgarthr! So I was right, but for the wrong reasons....

2016-10-06 00:41:11 crust: Helfrich - you are worried that your loss of points because of the disqualification of a player in the first round might affect your tie-break status in the second round? My understanding is that tie-breaks are based only on what happens in that round, not on what happened in previous rounds, so you should be completely unaffected.

2016-10-06 01:01:32 HELFRICH: crust -- interesting revelation. So if the points (wins) don't matter, I am in 4th place in my group regardless and that qualifies me for placement in the next round -- so why then am I not placed. The tie breakers I have ALWAYS been talking about are those used for advancement to and placement in the next round (tiebreakers between placement from different groups). Vonraider's and Steiger's games outcome should then only affect which group they are placed in and has nothing to do with my not being placed. If so then make it so and place me! My original question was why wasn't I placed ....

2016-10-06 01:11:29 HELFRICH: Crust -- I just reread your last statement and it appears we are not communicating. I am referring to "tie breakers" in the FIRST Round that are used for player PLACEMENT in the SECOND round groupings.

2016-10-06 02:06:22 Tuireann: Yeah, I thought Helfrich meant Valholl advancement and the tournament as a whole in general. I've written a few replies and then refreshed read and reread and at this point I've lost the plot. I assume the reason you have not been placed yet is because the outcome of the remaining games influences into which second round group you end up in for the sake of balancing the groups based on the results of the first round. And that it's because Vonraider will end up with a score of 0.50,0.57, or 0.64 to your 0.58 so presumably one ends up in Asgarthr and one ends up in Mithgarthr. But I have no idea how the groups are being distributed because its not stated.

2016-10-06 02:38:54 HELFRICH: Tuireann -- I was hoping the tourney organizers could provide a simple straightforward answer. I am confused. I have been politely provided with three different answers. Your last remark in your post appears to be very accurate, in short it is not defined. I also can 'assume' reasons, but I was looking for clarification.

Re: Hnefatafl Internet Championship

Posted: Thu Oct 06, 2016 6:34 am
by Hagbard
Helfrich wrote:Your removal of two games wins does not seem 'cricket' to me. It is unfair that I or others in my first round grouping should be penalized because some non-serious entry didn't finish any of his games. That player should not have been 'wiped' from the tourney records either: historical records count for future tournaments.
A player is only removed from the table, if he times out in more than half of his games.
An example: You lose two games against player X who then times out in all other games, thereby giving all other players of your group two gratis points. That would not be fair.
Helfrich wrote:As I noted in an earlier posting, some sort of qualification should occur for entry into this championship cycle (like perhaps having played at least 25 games).
"New and unknown players" are only new and unknown to us. They could be very seasoned tafl players elsewhere. Or perhaps not so seasoned, which is ok, because all players no matter their strength, are distributed to a suitable Round 2 group.

Being the World Championship Tournament, it's important that the tournament is open to all tafl players.

To minimize the impact of players who could possibly drop out of the tournament (that is "new and unknown" players plus players who have done so at earlier occasions), these are distributed equally among the Round 1 groups.
This year four players dropped out of Round 1: RBerenguel from Fornebei, TheBloodyJarl from Heithbyr, and Teondrae and Chalhub from Vinland.

Re: Hnefatafl Internet Championship

Posted: Thu Oct 06, 2016 6:47 am
by Hagbard
crust wrote:let's say that vonraider loses his remaining 2 games, and remains on 7 points. That's 7 points out of a possible 14, whereas Helfrich has 7 points out of a possible 12. 7/12 is more than 7/14, so it seems to me, in that case, that Helfrich should be favoured. However, if Vonraider wins one of his remaining games, then Helfrich's 7/12 STILL beats vonraider's 8/14, since 7/12 = 1/2 + 1/12, and 8/14 = 1/2 + 1/14. So again, Helfrich wins, even though he has one more point than vonraider! Now, if vonraider wins both his remaining games, then his 9/14 does finally beat Helfrich's 7/12
Tuireann wrote:Vonraider will end up with a score of 0.50,0.57, or 0.64 to your 0.58 so presumably one ends up in Asgarthr and one ends up in Mithgarthr.
That is exactly right.


The calculation method for distributing the players to the Round 2 groups is sort of a First-Order Approximation, which was also used last year (First-Order = Linear, meaning not 100% precise but very precise):
  • As the Round 1 groups are equal, a point earned in one group is equal to a point earned in another.
  • But, as some Round 1 groups have fewer players than others, points are scaled up/down before comparing. (Ex.: in group 1 you can earn max. 14 points and in group 2 max. 10 points. Then all group 2 results are multiplied by 14/10 before comparing).

Using this approximation, the players' Round 1 results are:

Tunsberg max 14 pt
Steiger 9 + 2
Arevidar 6
Fellhuhn 5
Vonraider 7 + 2
Mykle 1

Fornebei max 14 pt
odinhimself 10
gavinbegbie 0
fraech 7
agmundr 8
fairland 5
waegn 2

Heithabyr max 12 pt (x 14/12)
Animals 9 -> 10.5
Helfrich 7 -> 8.2
Gjallarhorn 3 -> 3.5
Sybil 3 -> 3.5
Patterson 0

Vinland max 10 pt (x 14/10)
Altti 6 -> 8.4
Tonythebook 6 -> 8.4
Kratzer 1 + 1 -> min. 1.4 max. 2.8
Aluric 0 + 1 -> min. 0 max. 1.4


Now sorting by points and dividing into Round 2 groups:

Asgarthr
Animals 9 -> 10.5
odinhimself 10
Steiger 9 + 2
Altti 6 -> 8.4
Tonythebook 6 -> 8.4
Helfrich 7 -> 8.2

Mithgarthr
agmundr 8
Vonraider 7 + 2 (Vonraider could theoretically go to Asgarthr instead of Helfrich)
fraech 7
Arevidar 6
fairland 5
Fellhuhn 5

Utgarthr
Sybil 3 -> 3.5
Gjallarhorn 3 -> 3.5
waegn 2
Kratzer 1 + 1 -> min. 1.4 max. 2.8
Mykle 1
Aluric 0 + 1 -> min. 0 max. 1.4
Patterson 0
gavinbegbie 0

(More players in Utgarthr, because if anybody drops out of Round 2, he will often be from Utgarthr.)
As a check, it's clear that the distribution looks reasonable.

By October 1st the distribution of everybody to Round 2 groups was clear, except for two, Helfrich and Vonraider, Helfrich going to Asgarthr and Vonraider to Mithgarthr or reverse.

So it was decided to start Round 2 for the certain 26 players, and add Helfrich and Vonraider later as soon as their positions are clear.

Re: Hnefatafl Internet Championship

Posted: Sun Oct 09, 2016 1:50 pm
by Hagbard
A very accurate, alternative calculation of Round 2 groups.

Accurate method for distributing the players to Round 2 groups:
  • One or more "ghostplayers" are added to smaller Round 1 groups, so that all groups will have the same number of players.
  • The "ghostplayers" are given ratings so that all groups are equally balanced.
  • Additional points from missing games of ghostplayers are calculated from the probability of wins against the "ghostplayers".
In [] brackets is shown the results from the simple method above for comparison.


Using this method, the players' Round 1 results are:

Tunsberg 8 players
1957 Adam 12
1571 Tuireann 12
1677 Steiger 10
1432 Arevidar 6
1380 Fellhuhn 5
1383 Vonraider 8
(1474 Ded Fomich 2)
1211 Mykle 1

Fornebei 8 players
1917 Crust 13
1591 Xerxes 11
1826 odinhimself 10
1227 gavinbegbie 0
1384 fraech 7
1363 agmundr 8
1298 fairland 5
1338 waegn 2

Heithabyr 7 players
1899 Herjan 11
1635 Hagbard 9
1738 Animals 9 -> 10.72 [10.5]
1534 Helfrich 7 -> 8.31 [8.2]
1354 Gjallarhorn 3 -> 3,80 [3.5]
1340 Sybil 3 -> 3.76 [3.5]
1091 Patterson 0 -> 0.13 [0]
1424 Added ghostplayer -> 5.28

Vinland 6 players
1927 Sigurd 9
1615 Bigwurm91 7
1831 Altti 6 -> 9.87 [8.4]
1583 Tonythebook 6 -> 9.43 [8.4]
1435 Kratzer 2 -> 4.98 [2.8]
1129 Aluric 0 -> 1.34 [0]
1248 Added ghostplayer -> 2.19
1248 Added ghostplayer -> 2.19


Now sorting by points and dividing into Round 2 groups:

Asgarthr
Animals 9 -> 10.72 [10.5]
Steiger 10
odinhimself 10
Altti 6 -> 9.87 [8.4]
Tonythebook 6 -> 9.43 [8.4]
Helfrich 7 -> 8.31 [8.2]

Mithgarthr
agmundr 8
Vonraider 8
fraech 7
Arevidar 6
Fellhuhn 5
fairland 5

Utgarthr
Kratzer 2 -> 4.98 [2.8]
Gjallarhorn 3 -> 3,80 [3.5]
Sybil 3 -> 3.76 [3.5]
waegn 2
Aluric 0 -> 1.34 [0]
Mykle 1
Patterson 0 -> 0.13 [0]
gavinbegbie 0


So, this very complicated method results in precisely the same distribution as did the simple method!!

Re: Hnefatafl Internet Championship

Posted: Mon Oct 10, 2016 2:02 pm
by Hagbard
Championship Tournament 2016 Round 1:
RBerenguel and Chalhub did not move at all, never showed up, and were removed from the tournament tables.
Teondrae did one move in some games and zero moves in the rest, before timing out in all games, and was also removed from the tournament tables.

TheBloodyJarl completed one game and is reinserted in the tables.

Hereafter the players' Round 1 results are:

Tunsberg max 14 pt
Steiger 10
Vonraider 8
Arevidar 6
Fellhuhn 5 Sonneborn-Berger score 22
(Ded Fomich 2)
Mykle 1

Fornebei max 14 pt
odinhimself 10
agmundr 8
fraech 7
fairland 5 Sonneborn-Berger score 11
waegn 2
gavinbegbie 0

Heithabyr max 14 pt
Animals 11
Helfrich 9
Sybil 5 Sonneborn-Berger score 9
Gjallarhorn 5 Sonneborn-Berger score 9
Patterson 1
(TheBloodyJarl 1)

Vinland max 10 pt (x 14/10)
Altti 6 -> 8.4
Tonythebook 6 -> 8.4
Kratzer 2 -> 2.8
Aluric 0


Now sorting by points and dividing into Round 2 groups:

Asgarthr
Animals 11
odinhimself 10
Steiger 10
Helfrich 9
Altti 6 -> 8.4
Tonythebook 6 -> 8.4

Mithgarthr
agmundr 8
Vonraider 8
fraech 7
Arevidar 6
Fellhuhn 5 Sonneborn-Berger score 22
fairland 5 Sonneborn-Berger score 11

Utgarthr
Sybil 5 Sonneborn-Berger score 9
Gjallarhorn 5 Sonneborn-Berger score 9
Kratzer 2 -> 2.8
waegn 2
Mykle 1
Patterson 1
Aluric 0
gavinbegbie 0


Distribution to Round 2 groups is unchanged.

Re: Hnefatafl Internet Championship

Posted: Wed Oct 19, 2016 10:39 pm
by Adam
All thanks to Hagbard for this exhaustive testing of methods. It looks like this is wrapped up now. Some final thoughts from me:

The hypothetical question here is 'how does my (for example) fourth place in my group compare to the other fourth places from the other groups? Am I 1st 4th, 2nd 4th, or 4th 4th? Or will I get demoted to a 1st 5th? Or better yet promoted to a 4th 3rd?

The concern is that players dropping out will skew this comparison creating the above demotions.

The alternative to players' round 1 table ranking getting pushed down by players dropping out, is their ranking getting pushed up, which seems just as unfair. Perhaps more unfair.

You can count an unplayed game as nothing at all, like it never happened, or as a win or loss regardless of the drop out players rank, which didn't happen.

An extreme example: A group has eight players, the six highest ranking drop out, leaving two weaker players. If we counted the unplayed games as wins against the stronger players, it would create an unrealistic and skewed result. Either way the two will go through to the top level finals as winners of their group, and most likely come in last (unless everybody else drops out). If they were 3rd and 4th in their group, and we count their unplayed games as wins, they are catapulted to the top of the 2nd and 3rd divisions. Where they will find themselves up against opponents who have actually beaten players of that level...

So the only advantage closing this loophole might confer, is being moved up from a group where you are ranked top, to play in a group where you are ranked bottom. With the loophole as it is, you might get pushed down a group, where you will find yourself hot favourite.

Re: Hnefatafl Internet Championship

Posted: Wed Dec 21, 2016 12:24 pm
by Hagbard
This is my bid for an evaluation of the Championship Tournament 2016:

My bid will be very similar to last year.

What worked well.
  • A month of signing in before the tournament. An Are-You-Sure step in the sign in.
  • That everybody can join. We had again new names, who turned out to be strong players, one new name even made it to the top final!
  • Tournament start at September 1st, after summer holidays.
  • Using the Thue-Morse sequence to distribute players in even groups for Round 1.
  • Splitting the tournament into two rounds, and splitting both Round 1 and Round 2 into four subgroups. The resulting 6-8 players for each group was adequate.
  • In my opinion, it was a great improvement to use the Historical Hnefatafl 11x11 (Welsh Tawlbwrdd 1) for Round 1, before the usual Copenhagen Hnefatafl 11x11 for Round 2. This made Round 1 easier and quicker for the players to get through, before the more demanding Round 2. And it made the whole tournament more varied and interesting.
  • The timeout method worked well. 1 day per move and Round one 7 days buffer, Round two 14 days buffer. The winner was found after 80 days (13 days faster than last year), and the whole tournament was finished after 91 days (19 days faster than last year).
  • If a player didn't move at all during the first eight days of Round 1, he was removed, and the same with Round 2 and 15 days. Also a player who caused more than half of his Round games to time out, was disqualified. 8 players were disqualified by these rules.
  • Use of the Sonneborn-Berger principle a.o. in case of a tie.
  • Valhalla Final with the 8 strongest players, and that all players had their Round 2 Final.
What could be improved next year.
  • Being this many players, 33 from the start, makes many players even in strength, and therefore a very accurate method is needed to distinguish their results and distribute the players to Round 2 groups. An accurate method like this one:
    http://aagenielsen.dk/hnefataflforum/vi ... p=726#p722

33 players registered for the tournament, and 25 players saw it through.
In total 358 games, 16107 moves and 2935 killed pieces. 198 white wins and 160 black wins.
Longest game was 170 moves.
Shortest game was 7 moves.

Congratulations to World Tafl Federation Champion 2016, Leo Kolassa ("Herjan"), Formby, UK!
Many thanks to Tim Millar and Adam Bartley for being umpires, and to the 25 players who finished the tournament!

Re: Hnefatafl Internet Championship

Posted: Tue Jun 20, 2017 10:42 am
by Hagbard
Time to prepare for the World Championship 2017!
Here's my proposition for the tournament rules this year.
Notice that Copenhagen rules are proposed for both Round 1 and Round 2. This is to avoid that potential finalists don't make it to the final just because of different rules for Round 1.
Round 1 results are not transferred to Round 2, that is, even if Player 1 and Player 2 already met each other in Round 1, they'll still play new games in Round 2.
Round 1 should be a rather quick round with the time buffer halved.


Proposition for tournament rules for the World Tafl Federation Championship 2017:

Sign up: all July and August. There will be a button for registering for the tournament.
Start: September 1st.

Tournament rules.
The Copenhagen rules, described here:
http://aagenielsen.dk/copenhagen_rules.php

Points are 1 for won, 0 for lost and 0.5 for a draw.
Umpire: xxx. For xxx's games, umpire is yyy.

The Thue-Morse sequence is used to distribute players in groups for Round 1 based on ratings.
One (!!) player per preliminary group qualify for the final.

In case of a tie:
1. direct comparison of the games of the tied players (for this round).
2. Sonneborn-Berger score calculated for this round.
in case of still a tie:
3. preliminary round: both players proceed. Final: new rounds until one wins.

Preliminary round: time buffer 5 days (!!)
Final round: time buffer 10 days.
But even if a player has no time buffer left, he has always 1.25 day to do his current move.
(Please remember that the time buffer is meant for use when away on weekend or vacation. Could be unwise to use it up in everyday moves.)

If at any time you want to leave the tournament, just say so.
If a player does not move at all during the first 6 days of Round 1, he is removed from the tournament. The same with Round 2 and 11 days.
A player who causes more than half of his Round games to time out, is disqualified from the tournament.

Round 1:
The preliminary groups are named Tunsberg, Fornebei, Vinland, Heithabyr etc. (Viking sites).

Round 2:
The final group (with the future Hnefatafl World Champion) is named Valhallah.
All other players will also have their own finals, divided into final groups based on the Round 1 results. The groups are named Asgarthr, Mithgarthr, Utgarthr etc. (mythical sites).

Distribution of players in Round 2:
The top winner from each Round 1 group make up the top group in the final. The remaining players are distributed into further groups according to their Round 1 points taken as a percentage (6/12=50% 5/10=50%), such that drop outs and uneven group sizes are accounted for.
Or calculated even more accurately if necessary.

To shorten the tournament time and avoid that everybody wait for a few slow games:
When the placement of 80% of the players for Round 2 is known, Round 2 can start with these, and the last players added when ready.


Example:
36 players.
Round 1: 6 groups of 6 players each.
Round 2: Top final with 6 players. Rest of the players also split into groups of 6 players each.
(Number of groups = SQRT(Number of players))