GG & WP goes out to @flagrantangles and @charnel_mouse, the field narrows down to 6 players left.
Round 5, here we come. Please remember to post the rule changes at the beginning of each thread.
4 Likes
We have to say goodbye to @dwarddd GG & WP (don’t be too harsh with yourself)
Next comes round 6
[XCAPS22] Round6: P1 Moby_Dick [Finesse]/Ninjutsu/Discipline vs P2 zango [Finesse]/Ninjutsu/Feral- [XCAPS22] Round6: P1 Bomber678 [Future]/Peace/Blood
vs 
P2 FrozenStorm [Demonology/Necromancy]/Finesse -
[XCAPS22] Round6: P1 Dreamfire [Demonology]/Anarchy/Balance
vs P2 zango [Finesse]/Ninjutsu/Feral
Bye: Dreamfire
Edit: due to Moby dropping out, a slight change of plans has occurred.
3 Likes
I was curious why the algorithm came up with this pairing table being the one with the smallest penalty value. For all of you that are interested here is the reasoning behind it:
- A rematch is unavoidable, but the pairings can be restricted to only one rematch.
- Dreamfire and myself are the ones with the most losses, so one of us has to get the bye. I have already played 3 games as P1, all other four players left have played 3 games as P2. Hence, if I get the bye there will be two players with 4x P2 and 2x P1, which results in the unbalanced P1/P2 penalty kicking in twice. If dreamfire gets the bye and not me this penalty kicks in only once, hence he gets the bye.
- Swiss penalty will kick in exactly once, no matter what.
- Direct rematches have a high penalty, so I cannot play against bomber.
- If I play against Frozen (which would be a rematch), the left over pairing is also a rematch, hence I play against the last one: Moby.
- His P1/P2 ratio and mine dictate that I must be P2
- As bomber and frozen have already played against each other, the rematch checker dictates that bomber should be P1
Now here comes my question:
If in the current situation giving a bye to one of the three players with 1 loss instead of one of the two players with 2 losses would have resulted in a rematch being avoided, the algorithm might have preferred to avoid the rematch over the “bye should be given to a player with the most losses”, especially as then no swiss penalty might have been necessary. What is your take on what should be prioritized?
1 Like
I don’t know about “should”, but avoiding rematches has higher priority in our usual matching rules.
3 Likes
Seems like what I would have done yeah. Avoids double rematch, worst aspect is I get my fourth p2 and a kinda recent rematch but avoids double rematch and gives a bye to a two loss player. Pretty reasonable outcome
2 Likes
Guys, the tournament was a lot of fun, but I have to step out. The baby came tonight, and my codex proscription is now valid. I think the timing is good since dreamfire and zango can just go on.
@FrozenStorm: Since you have TS and might make the 21:00 CEST I would be very interested in a 2vs2 wirh zango. Maybe we find someone else. Just need some time to get in the new flow.
7 Likes
So I repeat my congratulations and best wishes here as well!
You should have gone for that mm+tech break move last turn against Frozen 
with this route of “choosing the slow death against nightmare” you could have left this tournament undefeated. But thanks still that you showed us that there are decks out there with Sets that are worth mentioning!
With this not unexpected (but I still hoped until the very last second that codex does not have to be reduced to zero) change of plans, I think it better to have a pairing dreamfire against me, instead of both of us more or less getting a bye in the same round. I assume that this is fine for you, @Dreamfire ?
If so please open up a game with this link:
Can please someone tell me how I can format text as striked out? That would be fine for the initial post of round 6 above.
3 Likes
You add [strike] before and [/strike] after the text you want struck through, just with < > instead of .
Or just copy+paste what you need from here:
We have to say goodbye to @dwarddd GG & WP (don’t be too harsh with yourself)
Next comes round 6
3 Likes
two more questions about “what should the algorithm do”:
- I just checked my assumption that no matter who wins the match between dreamfire and myself that then bomber will get the next bye (reasoning was: a bye has to be given to one of the two players with 2 losses, then the direct rematch needs to be avoided, alas bomber gets the bye). This is wrong. As bomber and dreamfire have not yet played against each other, the sum of the penalties is lower if frozen gets the bye in case of dreamfire winning
details for the penalties
frozen gets bye & dreamfire vs bomber (or P1/P2 swapped): penalty score = 5 as only the penalty for “bye goes to someone not in the group of least byes and most losses” kicks in
bomber gets the bye and P1 dreamfire vs P2 frozen: penalty score = 6 which consists of:
swiss penalty for players with a different number of losses playing against each other: 3
rematch penalty: 3
- A what if question: Currently the algorithm is configured such that if player a has already played as P1 against player b once, then for the rematch player a will be P2. What if this leads to a very unbalanced P1/P2 distribution among one or both of these players?
Currently I have a penalty score of 0.5 for each player that has a difference of 2 between the number of games as P1 and number of games as P2. If the difference becomes 3 or grater, the penalty becomes 1.5. So the overall unbalanced P1/P2 penalty is less or equal to three in this example. Only if player a would have 3 games more as P2 than as P1 and player b would have 3 games more as P1 than as P2 the unbalanced bye penalty would generate a penalty value of 3.
The penalty for a rematch being unbalanced, in this example the second game again being P1: a vs P2 b is equal to three.
The outcome of the algorithm therefore is: Unless the very extreme case of individual unbalanced P1/P2, the algorithm will always favor P1: b vs P2: a
If we have the absolute extreme situation as described above about the individual unbalanced P1/P2 situation, randomness will decide who gets P1.
What do you think: Should I reconfigure the penalty values or is it fine as it is?
Thanks a lot for this nail biter of a game, GG WP @Dreamfire
Here comes round 7
Bomber gets the expected and well deserved bye.
Also:
I’d appreciate opinions on the two questions above about the parameters for the algorithm. What would you prefer that the algorithm should do?
1 Like
I’ll start our game later tonight or tomorrow.
On the first point I don’t think a player with the fewest losses should be considered for a bye
On the second… I guess I would give stronger penalty to 3 differential of assignments and have the rematch not flipped, but I don’t think that’s as clear cut
1 Like
Somehow I managed to get @FrozenStorm to deviate from his A game and to steal a victory from him, so he now gets to sit and wait for his opponent to reach the finals.
Round 8
vs FYI: why did the algorithm choose this pairing?
- A direct rematch leads to the highest penalties, so bomber does not get the bye.
- Bomber and frozen already have played twice against each other, so me getting a bye has a large rematch penalty
- Bomber and myself have played only once, so it’s a given that frozen gets the bye
- The previous match of bomber and myself (while both of us have an almost even distributed number of P1/P2 games) dictates who gets to play as P1.
For the finals frozen will get to be P1 if bomber is his opponent and P2 in case it is me. This is due to it being the second rematch either way and the individual history of the number of games as P1/P2 is different for all three of us.
3 Likes
vs 
P2 FrozenStorm [Demonology/Necromancy]/FinesseNo matter how the finals will end, we now have concluded the pairing process. I’m really looking forward to feedback and opinions about the algorithm. Feel free to rant if you have the desire
Already planned changes:
- I’ll raise the penalty for “very unbalanced P1/P2 ratio” (if a player has to play his 5th match as P2, while he has only played 2 games as P1, for example) - see the what if discussion above about the why
- I’ll try to include a counter how many times a player has already played against an opponent with a different number of losses ( @charnel_mouse had to play twice against an opponent with fewer losses which is unfair )
1 Like
I thought the pairings were absolutely fine. I’d say the real questions are whether it’s actually easier for the organizer, and how easy will it be for other organizers to pick up?
2 Likes
Well it is far away from a one click only solution, but it still is less than 5 minutes overall including the post for the next round in the forum. I wrote a detailed step by step instruction with which I’m pretty certain that everybody who is capable of playing codex will easily be able to do the pairings.
I would be very happy if someone else would do the pairings for the next tournament, obviously I would be willing to help out in case there are questions or unforseen issues with the spreadsheet.
The only thing I’m not familiar with is what I have to do to get Google to the point where it trusts the code. When you want to use the macros from the spreadsheet Google asks you whether you really trust the creator of it, as it might get access to all your other Google spreadsheet data. I tried a lot but was unable to this better.
For reference:
And congratulations go out to @FrozenStorm for a very impressive way to win this tournament, well deserved!
I’m still curious about further feedback about the algorithm and also the question is out: is anyone else willing to do the organization of the next tourney? I’d love to help out with the pairing algorithm and to provide first level support in case it is required 
Edit: and what a crazy P2 tournament this was. If I did not miscount we only had 6 times a P1 victory out of 27 matches… Weird!
5 Likes
Wild how P2 heavy this tourney was. GG WP to everyone 
I can probably organize the next one but am happy for others to step up. If I don’t see a sign-up thread posted by end of next week I’ll post one
1 Like
Really looking forward to someone else using the new algorithm in order to safe time for the organizer!
In case it’s required I can always jump in for a future tournament again as it really isn’t that much work, but I think it’s better if the tournament organizer is a judge and for that I’m way too far away from the required understanding of the ruling.
I’m a little busy at the moment, but I’d be up for running XCAFS.
2 Likes