[Swiss] Framework repeated a matchup instead of pairing two "active" winners

[Meeplelowda]

In this tournament (which inexplicable only had 5 entries when previous iterations had 16+, but I digress), in the second round the player who had a bye in the first round was paired with one of the players who won their first round match. One player who lost their first round match was given a bye, meaning the only two players left were the other winner of the first round match and the other loser of the first round match (yours truly). So I ended up paired with the same person again despite the option to "try to prevent same players to be opponents twice."

I get that the same person can't have two byes in a row. But shouldn't there be a preference for the two active winners being paired in this situation? This is probably an edge case that is an artifact of the low number of entries, so I'm not sure it's even worth a suggestion/bug report.

Or given that this can happen with 5 players, what is the fewest players where this *can't* happen? That could become the minimum number of entries required. Of course, we'd have to know exactly how it is deciding the pairs to know how many people are required to avoid it.

[Mathew5000]

I think it's because, the setting "points attribution for skipped players" is "maximum points a player would gain in a full game". (That's compulsory where number of players in a match is 2.)

So I understand what you mean by "two active winners", but I suspect the algorithm just sees "three players tied with 1 point each". In round 2, one of the non-winners gets a bye (I think the algorithm picks the lowest-ranked player, who has not had a bye yet), and then two of the three players with one point are randomly chosen to play each other. (Or maybe that part is not random? I don't know.)

I take your point that since "try to prevent repeat matchups" is checked, the algorithm should have avoided the pairing where you go against Mario a second time. But I believe we've established in various previous threads, that the algorithm starts at the top of the rankings, and does the matchups from there, without regard to what will happen at the bottom. Is that a bug? Well it's not ideal but it's the kind of thing, in my opinion, where fixing that could easily introduce other problems.

Also, consider this: Radix had a bye in round 1, and obtained 1 point. So in round 2, Radix should go up against the (presumptively) best player in the tournament, i.e. somebody who won their first match. Because otherwise, Radix has too easy a schedule, getting a bye in round 1 and then playing a "loser" (no offence intended!) in round 2.

[Meeplelowda]

But I believe we've established in various previous threads, that the algorithm starts at the top of the rankings, and does the matchups from there, without regard to what will happen at the bottom.

I'd forgotten about this discussion, but I remember it now.

Also, consider this: Radix had a bye in round 1, and obtained 1 point. So in round 2, Radix should go up against the (presumptively) best player in the tournament, i.e. somebody who won their first match. Because otherwise, Radix has too easy a schedule, getting a bye in round 1 and then playing a "loser" (no offence intended!) in round 2.

This also makes a certain amount of sense. So to answer my other question, there needed to be a third first round match so that the person with a bye could be matched against one of the "active" winners, and the other two active winners could be matched against each other. So we need a minimum of 6 players to avoid this scenario. Obviously with 6 there wouldn't be a bye in the first round, and with odd numbers 7 or higher we have enough players to both match the bye with a first-round winner and not repeat a first-round matchup.

No offense taken. I'll try not to blunder my position in this rematch.

[BananaHarambe]

I like the way the tournament handled the pairings in round 2 by pairing the player with a Bye against a winner. This is the best way to avoid misrating a player in this size of tournament.

The tournament needed to assign a bye to one of the bottom scored players. Maybe it give it to the lowest Elo player, maybe it was random, not sure but it was certainly prioritized first before pairings.

Did it give mario the pair down because he was the highest ranked player without a bye? Maybe it did and that would be because Meeple is higher rated than the other player that lost Round 1, hence mario appears "better". Not sure if order within each scoring bracket even matters for pairing.

It is interesting that mario didn't get paired against the player with the bye, which resulted in a duplicate pairing.

Looks like 5 person 2p swiss tournaments cannot really function properly, which makes sense since they shouldn't function anyways. Round 2 is guaranteed to have a "double pair down" for the winners, one plays a previous loser and one plays a player with a bye (minor pair down, performance unknown).

I agree to make it 6 person minimum and if attendance fails you could rehost it as a small round robin tournament maximum 5 players.