Pokémon TCG: Sword and Shield—Brilliant Stars

Serious Topic: Legitimate Proposal for Worlds 2005

Status
Not open for further replies.
okay POP a one shot simulation of 7 rounds of random pairings with 128 entrants.

After 7 rounds

18 players on 7-0
17 players on 6-1
14 players on 5-2

so only one space for a 5-2 player but which one ??



7-0: 1,2,3,4,5,6,7,8,9,10,13,14,19,20,21,22,23,39
6-1: 11,12,15,17,24,25,29,30,34,43,46,63,71
5-2: 16,27,28,32,33,35,40,41,44,47,49,50,55,93

Top 32 players that didn't make it because their record was 4-3!
18,26,31

None of the Top32 players at the begining of the tournament had a record worse than 4-3 in this single shot simulation which is I suspect is what you are anticipating.

However if you allow one player at random from the 5-2 group into the T32 elimination then probably 7 ranked players don't make it to the single elimination.

I bet you are going to ask what happens if I add another couple of rounds !
 
Last edited:
two more rounds...

After 9 rounds:
15 players on 9-0
10 players on 8-1
8 players on 7-2

this time one player on 7-2 has to be dropped from the cut.

9-0: 1,2,3,4,5,6,7,8,9,10,14,20,21,22,39
8-1: 11,12,13,15,17,23,24,25,34,46
7-2: 19,27,28,29,30,32,43,55


Four players definately wont make the cut this time : 16,18,26 amd 31.
with one 7-2 player to join them. So it looks like 5 of the ranked players don't make the T32 cut after 9 rounds.

and since you are interested in what happens on the cusp of making the cut here are the 6-3 results

6-3: 16,18,26,33,35,40,41,42,44,45,47,49,56,59,61,63,71,93

I know that it needs more trials to be statistically significant. However, using 9 rounds of completely random pairings did a reasonable job of selecting the bulk of the candidates for the T32 single elim, at least in this single trial. I'm slightly less happy with the performance of completely random pairings after 7 rounds, even though the performance after 7 rounds exceeded my personal expectations. [I was anticipating completely random pairings to be much worse than the swiss pairs and in reality they aren't massively different.]
 
Last edited:
Hmm.
Yep, that's about what I expected. My thought is that Swiss forces the better than average players to play against the top players, giving them a worse record than they would have in the random pairings.
Now, what would probably create even better results would be to pair the winners against the losers. This would accellerate the sifting of best players to the top and bad players to the bottom.
It gets slightly more complicated in later rounds, but basically you would always pair the people with the best records against those with the worst records. I imagine that you'd get the top players sorted out within.... 5 rounds is my prediction!
 
NoPoke said:
3 points for a full win (ie. all prizes taken)
2 points for a prize win
1 point for winning by benching your opponent when neither player has taken a prize.
Click to expand...
In that system, you neglected to factor in the overtime win, which I believe needs to be taken seperately from a prize win. I'd see something more like this...

3 points = full prize win OR bench win with more than 4 prizes taken by the winning player
2 points = timeout prize win OR bench win with between 2 and 4 prizes taken by the winning player
1 point = overtime win OR bench win with less than 2 prizes taken by the winning player

Has to be modified, but this is a rough estimate of what I think might be good.
 
POP, which is why in a straight swiss tournament the top players start by playing the bottom half. Not in reverse order like with a single elim but just a stright top half plays bottom half.
 
I guess starting pairings based on records wouldn't be a bad idea. It would speed up the matches, even though some people have high records due to the "big fish, little pond" phenomenon. It should even out, and it actually gives a reason to attend and do well at tournaments, since they will affect your worlds pairing, pending you make it in.
 
The rating system will need to have variable K values introduced before I'd suggest using it solely as the basis for ranking the entrants. (for the reasons that you outline)

I hadn't realised until I ran the idealised experiment just how important the rank ordering component is to the functioning of SWISS pairs. If I had run my simulation using chess software, that is software that uses rank ordering, then the standings would have followed the rank for the first 7 rounds! No instability and every single player in the top32 by rank making it through to the top32. This is quite a revelation for me.

So if we can come up with a valid rank ordering for the majority of the entrants then there seems to be every justification in cutting the number of rounds down from 7 to 5 and at the same time introducing match play. This will increase the amount of time allocated to the swiss but by less than an hour I expect. So rather than campaign for an increase in the number of rounds or a switch to 2/3 match play neither of which will happen because of time constraints lets fix a major source of randomness in the pairing procedure. This will free up that very time resource that will allow match play to be introduced.

So is there a valid and reasonable rating that can be used to rank the entrants??

Proposal: Use the results from last years worlds to determine the seeding for this years.
 
Last edited:
So what do we do about the people who are in worlds this year who didn't make it last year?

The problem now is that k values probably won't be implemented until next sesason even if there was a large discussion over it. Still, even if it were for next season it would still be better than not having it at all.
 
In chess unranked players are either placed at the bottom of the ratings or approximately 1/5 of the way up.

No rating system is perfect but using last years tournament seems like a good place to start. Of course if very few of the competitors this year played last year then the idea is a dead duck! However remember that the present system is that all entrants are given the same rating: which shows up in the need for several rounds just to eliminate the worst excesses of the pairings luck. A component in the SWISS algorithm is missing completely even its partial inclusion will be of benefit.
 
But remember, POP _decided_ to have ratings/rankings 'reset' at the end of the tourney year...surely there must have been a reason for that?

'mom
 
Venusaur: you can always ask for help when shuffling... or even ask your opponent to shuffle for you.

Mom: Since worlds takes place at the end of the tournament season using ratings that are reset each year wouldn't be so much of an issue. But I think that there is a bigger stumbling block to using the torunametn ratings. I expect we will always see a large number of Japanese players in future so using ratings to provide the missing ranking in the SWISS probably wont work untill such time as the Japanese join the rating system. Hence my suggestion of using the results from the previous years worlds.
 
Benlugia said:
however, people need energy to keep playing. i mean...12 hours and no eating? just because it works for you doesn't mean it works for everyone else.
Click to expand...


exactly. people with the skills like me to not eat all day and still keep going will be fine and have an advantage. but it's unfair to those who faint if they go that long....
 
Status
Not open for further replies.
Back
Top