Trying to find the logic behind this

[detlefchef11]

I joined a tournament and didn't realize the settings were this way, but it's set up with groups of 8 and we play 7 games to determine which 3 move on. Straight forward enough, as 7 games is enough to play everyone once. Only, you don't play everyone once. The first game I played the other player with an ELO over 400. Fine enough, just luck of the draw. Then, my 2nd game, I played one of the other players who won his first game (who happened to be the next highest ELO (still not strange). But, as I kept winning, I noticed that I always got whoever was in 2nd place at the time. Also, the last place would always get the 2nd to last and the people in the middle would play each other. Of the 7 possible opponents, I've played 3 of them. Two of them twice, and if the pattern holds, will likely play the other one 3x.

What is trying to be accomplished by doing this? I don't get what purpose setting things up like this serves.

https://boardgamearena.com/tournament?i ... ge_display

[ErikLevin]

The tournament is Swiss and set to "Don't try to prevent same players to be opponents twice". In Swiss you get paired with players as close as possible to you in ranking, while (normally) never playing the same person more than once. But the second criteria is disabled in this tournament. So everybody gets paired with the next person in standings. While 1st, you will keep getting paired with the 2nd player.

[detlefchef11]

The tournament is Swiss and set to "Don't try to prevent same players to be opponents twice". In Swiss you get paired with players as close as possible to you in ranking, while (normally) never playing the same person more than once. But the second criteria is disabled in this tournament. So everybody gets paired with the next person in standings. While 1st, you will keep getting paired with the 2nd player.

I understand what is happening. I'm asking what purpose that serves. Why one would want a tournament set up that way? You're basically penalizing players for doing well by giving them the hardest path possible.

[ErikLevin]

Idon't know, in my opinion, Swiss with "Don't try to prevent same players to be opponents twice" and an excessive number of rounds is not a good tournament format. Round-robin probably makes more sense for 8 players and 7 rounds.
I guess it's for people who don't mind playing the same opponent many times and who want many rounds but not quite as many as required for round-robin.

[detlefchef11]

My issue with it is this: It's not going to affect me because I've clinched 1st in the group. However, I could easily envision a log-jam from 2-5 where it's very much in question as to who will advance (3 advance to the knockout stage). The person in 2nd place going into the last game has to play the player who is likely the best player. Meanwhile 5 gets to play the person in 6th place. If the standings are close enough between 2 and 5, the 2nd place person could drop to 4th with a loss coupled by a win by the 5th place person and he'd have this system to thank for having a harder match-up.

[Mathew5000]

My issue with it is this: It's not going to affect me because I've clinched 1st in the group. However, I could easily envision a log-jam from 2-5 where it's very much in question as to who will advance (3 advance to the knockout stage). The person in 2nd place going into the last game has to play the player who is likely the best player. Meanwhile 5 gets to play the person in 6th place. If the standings are close enough between 2 and 5, the 2nd place person could drop to 4th with a loss coupled by a win by the 5th place person and he'd have this system to thank for having a harder match-up.

In my view, for an online Swiss-system tournament it's absolutely preferable to organize it like the one you describe, where a player will face the same opponent more than once if they happen to be "neighbours" in the ranking.

Let's suppose that each competitor plays seven games, and after game 6 there is one player still undefeated and four players who are tied for second, each with a record of 4 wins and 2 losses. The way it ought to work is, the tied player who has faced the toughest opponents so far (based on the win-loss record of those opponents) should now be "rewarded" by facing the player ranked #6 (whose win-loss record is presumably either 3-3 or 2-4). The tied player who has faced the softest opponents so far should now face the player ranked #1. Then the remaining two of the tied players should face each other.

That type of pairing algorithm would be most in keeping with the goal of Swiss system, which is to rank the competitors from best to worst.

I'm not sure whether BGA's algorithm does it that way, however.

[ErikLevin]

I think you, Mathew5000, are basically describing the Dutch system.

I think BGA uses something closer to a Monrad system, but it also seems to be to simplistic even for that, creating wonky pairings, based on other threads about pairing issues that I'm too lazy to find and link right now.

IMHO, people who organise tournaments on BGA seem to excessively prefer the really complicated structures and an excessive number of matches overall. In the example tournament here, it's a 64 player tournament with, apparently, a groups stage of 8 groups with 8 players playing 7 rounds of Swiss (!?) of which then X number of top players per group apparently continue to a single-elimination playoff?? Did I get that remotely right? That's a mind-boggling tourney structure.

Swiss is good for many players and few rounds. 8 players and 7 rounds of Swiss is not suitable.

Swiss would not be my first choice for groups stage elimination at all. Swiss with cutoff (say, top-4 continues, or get prizes or whatever) can always feel unfair because it can come down to tiebreakers, and it can often happen that that in the last round, someone who is playing for top 4 faces someone with no stakes to play for.

With only 64 players, just run 6 rounds of Swiss and be done with it? Or single elim for the whole thing.

[detlefchef11]

The tournament is Swiss and set to "Don't try to prevent same players to be opponents twice". In Swiss you get paired with players as close as possible to you in ranking, while (normally) never playing the same person more than once. But the second criteria is disabled in this tournament. So everybody gets paired with the next person in standings. While 1st, you will keep getting paired with the 2nd player.

What is strange is that I also noticed that 3 and 4 didn't play each other in the last matchup. 3 played 5 and 4 played 6.

But, back to the overall point. I can understand the desire for evenly matched games, in spectator sports. If all or even most of the games are between evenly matched competitors, the fans get a better experience. Of course, even THEY opt for a system that leans towards top players facing easier opponents in the early rounnds because, what we really want, is to see the best players or teams play in the semis or finals.

The whole thing seems needlessly complicated and ineffective. Based on the last responses, you need to take into account strength of schedule (which wouldn't be an issue if everyone just played everyone). Further, the whole system seems geared towards creating a log jam in the middle. After all, players that are a bit better than average keep getting pushed back to the middle because they beat an average player, then, by virtue of having done that, have to play agaisnt a better player and likely get pushed back to the middle. Meanwhile, the player they beat gets an easier matchup and also likely goes back to the middle. Rinse, repeat.

So now you have to rely on more tiebreakers than you likely would have otherwise.

[Mathew5000]

But, back to the overall point. I can understand the desire for evenly matched games, in spectator sports. If all or even most of the games are between evenly matched competitors, the fans get a better experience.

It's not only in spectator sports, though. Many recreational players prefer to face an opponent of approximately the same skill level, rather than someone considerably worse or better. For those players, Swiss System is a great format because after the first couple of rounds you are likely to face an opponent (or opponents, if it's a multiplayer game) whose skill level is close to your own. (Especially with "Don't try to prevent same players to be opponents twice".)

Of course, even THEY opt for a system that leans towards top players facing easier opponents in the early rounnds because, what we really want, is to see the best players or teams play in the semis or finals.

Swiss system generally does better than other formats in the goal of having the best players face each other in the last round of the tournament, if the tournament has been properly set up.

The whole thing seems needlessly complicated and ineffective. Based on the last responses, you need to take into account strength of schedule (which wouldn't be an issue if everyone just played everyone).

Well, strength of schedule is used as a tiebreaker in Swiss system. It's true that in a Single Elimination tournament you generally wouldn't need tiebreakers like that. But if you're comparing Swiss system to Round robin, in both cases there is a tiebreaker issue, and the way it's resolved in Swiss system is much more satisfactory than how it's resolved in round robin. See viewtopic.php?t=27700 and the other threads linked to therein. You say Swiss system is "needlessly complex" but a computer can figure it out easily.

Further, the whole system seems geared towards creating a log jam in the middle. After all, players that are a bit better than average keep getting pushed back to the middle because they beat an average player, then, by virtue of having done that, have to play agaisnt a better player and likely get pushed back to the middle. Meanwhile, the player they beat gets an easier matchup and also likely goes back to the middle. Rinse, repeat.

Well, yes, because taking the group of tournament entrants, their skill levels are usually distributed like a bell curve. Most of the players will be close to average in skill level, so a good tournament format should mostly push them to the middle of the standings.

As ErikLevin said, the particular tournament you are in has a structure that is not suitable. For one thing, stage 2 is going to be played as single-elimination, but it will have 24 players, which is not a power of 2. That makes no sense.

Where I disagree with ErikLevin is this: I believe that on BGA Swiss is good in a variety of different cases, not just where you have many players and few rounds. Swiss can be quite good for a set of players who all want to keep playing matches even after they have no chance of winning the tournament. For example, my friends and I play Swiss tournaments of 4-player games like Catan and Living Forest, where there's only 4 of us in the tournament and the number of matches is 10 or even 20. We like playing against each other, game after game. Similarly, if you have a group of 12 people who all want to play 8 games of Catan one after the other, Swiss works fine (as long as people aren't going to be inclined to drop out if they haven't won after a few matches).

For some two-player games it's boring to play someone who's considerably worse than you. If you have, say, 15 entrants, and everyone is eager to play a lot of matches (say 20), then Swiss system is ideal, with the option selected to permit repeat matchups. Then the second-best player in the tournament is probably going to play the best player five or six times, and probably also will play the third-best player three or four times. That's a good thing, if it's a game where it's most enjoyable to play against someone with similar skill level. For players in the middle, their schedule will be more varied than those at the top or bottom.

[detlefchef11]

As ErikLevin said, the particular tournament you are in has a structure that is not suitable. For one thing, stage 2 is going to be played as single-elimination, but it will have 24 players, which is not a power of 2. That makes no sense.

My guess, at least I hope this is the case, is that each group winner is given a bye in the first round of the knockout stage. That would make the math work.

I appreciate the rest of your comments and will just chalk this up as a poor implementation of the system rather than an indictment of the system itself.