I am 100% opposed to fast tracking, I think it breaks the spirit of the league, which is: you must play the games to get ahead. Reputation should never influence group placement, only performance.

The problem with your super-promotion conditional on sweeping your group idea is that it's not mated with guaranteed relegations, which could make for unbalanced promotions and relegations. A group can't gain a player without losing one, and vice versa.

"Those who think the problem would get worse, what exactly would be worse?"

Think about it, if you increase the group size you are increasing the games per group just so you would have 1 or 2 less groups.
Do more groups stall or increase the speed of the tournament?
They make the tournament finish faster for the simple fact that each group has games running while you cannot have the same amount of games running with less groups.
Eg:
In a group of 6 players you can only have each player playing a max of 2 games, thus each player must play and finish 2 games to play the rest.(time of 3 games minimum to finish 5 games)

While if you have 2 groups of 3 players, the entire league can finish in the time of a single game.


3 times faster to finish just by having 2 smaller groups instead of 1 big one.


Also in a RR the smaller the groups the less games are played.
In a group of 10, each player has to play 9 games while in a group of 6 each player has to play only 5 games.(nearly half the time)

So not only you are reducing the groups= slower, but you are increasing the size of each group to make it even worse.(more games)

It is simple math.

More games are being played at one time the faster the league is.
The less games being played the faster the league is.

IT also helps with boots and people going on vacation.

Would you rather have a group of 10 being stalled by a player on vacation, or a group of 5 being stalled and another group of 5 finished?
+ when the guy comes back from vacation it be faster for him to finish 4 games rather then 9.

No matter from which angle you view it, if you do the math, small sized groups will always speed up the tournament much more then larger ones.


"Buns147 has an interesting idea of having qualification tournaments for lower groups but again, i think this is best implemented as a separate league."

Yes that is what i was proposing with my very first post here, where you have the bottom groups in another identical league.
Also things could be implemented to promote the guys who win groups to go to the first league.(like a qualification)
Though this would have to be done properly with a clear idea of how the relegated players will be moved back to the second league.
The advantage of it is that each league can start separately and winners do not need to be placed immediately for next season but be entitled to join the season which has not started yet.

The second league does not need to be RR either, this will maintain the classic PR league and also fix the overload of ppl joining an RR tournament.

Edited 7/25/2016 08:38:17

I want to be in,if you're not gonna ignore my messages as some other guys,add me.

I would be interested in joining but would like to know when the next season starts. It depends on the answer whether i'm able to join for the next season or just wait for 1 more season until i join.

I think Krzychu's idea (and others) is best.

There's no need to even explain why if you look around you. Every major promotion-relegation system in sports works like this. Soccer leagues, basketball leagues, volleyball leagues, they all have a structure where at the lower level, the number of groups increases quickly, close to doubling per level even. Typically, the first doubling starts at the 3rd tier (sometimes already 2nd) and very rarely will you find a competition with more than 10 tiers with most leagues that I know having 5 to 9 tiers. In most competitions, the number of teams per tier decreases slightly for lower tiers.

For this format, this would be exactly as Krzychu proposed
Tier A: 1 group of 7, 2 relegate
Tier B: 1 group of 7, 2 promote, 2 relegate
Tier C: 2 groups of 6, 1 promote, 2 relegate
Tier D: 4 groups of 6, 1 promote, 1 relegate
Tier E,F,G,H: as tier D, but with likely more promotions due to dropout

C or higher: 26 players (28 if 7 instead of 6)
D or higher: 50 players (56 if 7 instead of 6)
E or higher: 74 players (84 if 7 instead of 6)
F or higher: 98 players (108 if 7 instead of 6)
G or higher: 112 players (136 if 7 instead of 6)
(I used 6 instead of 7 for lower tiers because (A) this is somewhat typical in sports leagues and (B) with fewer promotions and relegations, you don't want more than 70% (5/7) to not move every season in the lower tiers).

So 7 tiers (up to G) should be enough. At the lowest level (where newbies are added), more groups than 4 could be created as enough promotion spots would be all but guaranteed due to dropouts. In this case, even 6 tiers could work. Good players effectively promote faster in this system because there are fewer tiers to jump. Even if you miss promotion once or twice because it is harder, you'll still get to the top tier quicker. For above average players, their speed rising up should be about the same.

The only point of discussion could be how to handle promotions with dropouts. If someone from group B drops out, which of the numbers 2 of group C promote?
I think a simple rule, decided in advance should work, something like: (1) head to head (or in case of 4 players single-elimination tournament) if there is enough time (2) better record in current season (3) better finish in previous season (higher tier or higher rank in same tier) (4) repeat (3) for earlier seasons. If always equal, first to join.

Note that (much) larger groups would basically offset this idea. If you want to keep the average promotion-relegation speed as it is while still allowing players to be sorted at their desired level quicker, then groups of 6-7, max 8 seems to be the best way. More than that would make it a very stable system with only a small fraction of players promoting/relegating every season (unless you increase the number of promotion/relegation spots again of course)



Postscript:

Just for fun, a large group promotion-relegation system could look like this (much longer seasons, fewer groups)
Tier A: 1 group of 16, 3 relegate
Tier B: 1 group of 16, 3 promote, 4 relegate
Tier C: 2 groups of 14, 2 promote, 4 relegate
Tier D: 4 groups of 12, 2 promote, 3 relegate
Tier E: 6 groups of 12, 2 promote, 5 relegate
Tier F: 10 groups of xx, 3 promote, no relegate
I don't think that's a good idea with a round robin though.

A 8/16-player double-elimination promotion-relegation system could make sense though.
Tier A: 1 group of 16, 4 relegate (first round loser brackets losers)
Tier B: 1 group of 16, 4 promote (2 finalists, 2 from last rounds loser bracket), 4 relegate (as Tier A)
Tier C: 2 groups of 16, 2 promote (finalists), 4 relegate (as Tier A)
Tier D: 4 groups of 16, 2 promote (as Tier C), 4 relegate (as Tier A)
Tier E: 8 groups of 8, 2 promote (finalists), 2 relegate (first round loser bracket losers)
Tier F: 16 RR's of xx players, winners promote

My main please, I was inactive for about a month :/ - also glad to be back dear :)

When is the next season starting roughly?

Bayern, don't be an ass. You are not in the league either because you never requested it, you booted too much, you didn't join a tournament, or some other reason involving your non-participation. I'll add you to the new players list, see that you stay active this time.

Edge/Maximo, I'm still waiting on a tiebreaker game to finish, and I still need to flesh out the group changes for season 20. It will start no sooner than 2 weeks after i create tournaments and invite people. Sorry I can't give you a more concrete start date.

Glam, your main is added.

Season 20 players:
kevin#1 https://www.warlight.net/Profile?p=186244509
MarkusBM https://www.warlight.net/Profile?p=725103081
Motoki https://www.warlight.net/Profile?p=6862295334
Doc https://www.warlight.net/Profile?p=5627835089
Doron Hoh https://www.warlight.net/Profile?p=8847518898
KKND https://www.warlight.net/Profile?p=2921525624
MisterT https://www.warlight.net/Profile?p=7714693552
[FCC] Aura Guardian https://www.warlight.net/Profile?p=4127329486
Electric_Kitten https://www.warlight.net/Profile?p=2614926786
BYG Jacob https://www.warlight.net/Profile?p=8957617662
MIFRAN https://www.warlight.net/Profile?p=4825272432
Njord https://www.warlight.net/Profile?p=7139233655
Izazovnog https://www.warlight.net/Profile?p=5016951331
Semicedevine https://www.warlight.net/Profile?p=4328709837
Dogberry https://www.warlight.net/Profile?p=7267714858
kroete https://www.warlight.net/Profile?p=7137002334
Naram-Sin https://www.warlight.net/Profile?p=3940013870
Coronel Gavilan https://www.warlight.net/Profile?p=5842796420
carlos12.0 https://www.warlight.net/Profile?p=7218885952
dry-clean-only https://www.warlight.net/Profile?p=4419488318
Glamorous❤ https://www.warlight.net/Profile?p=2731867411
Bayern München is the best! https://www.warlight.net/Profile?p=2132686256

Krzychu, Math Wolf, and MoTD (via skype) all like the parallel group idea, so I am tempted to adopt it. I would prefer to keep group size at 7 players, 8 if needed due to a tie or some such like I've already been doing. I'm tempted to have the following scheme for season 20, which would utilize all players:

A
B
C1 C2
D1 D2
E1 E2
F1 F2 F3 F4
G1 G2 G3 G4? <--- newbies

I notice that krzychu & math wolf suggest combining the D & E groups into a single tier, but I am tempted to make people work just a bit harder to reach the top. What do the rest of you think?

I propose the following constraints for group organization going forward:
* To make promotions easier, tiers will have 1, 2, or 4 groups
* Tiers A and B have no more than 1 group
* Tiers C and D have no more than 2 groups
* All other tiers have no more than 4 groups
* A two group tier promotes the top player from each group into a 1 group tier
* A four group tier promotes the top player from each group into a 2 group tier
* tiers with multiple groups are randomly determined, keeping approximately even (a) the number of players relegating into the tier, (b) the number of players remaining in the tier, and (c) the number of players promoting into the tier
* there will be no super-promotions; all league dropouts will be replaced by additional players from the tier immediately below
* all groups will always demote 2 players, no matter the group size, unless the group is in the bottom two tiers. this is to promote more movement between groups.
* When deciding which identically ranked players in different groups in a tier should promote, the following tiebreakers are used: (a) head-to-head tiebreaker game or tournament, (b) win-loss record (omitting games played against lowest ranked player when comparing groups of uneven size), (c) win-loss record against players ranked higher, (d) ratios of fizzer's metric in all games
* However, in the league admin's discretion, a group may swell to 8 players to avoid using a hacky promotion tiebreaker.

I am noticing one potential downside, and that is waiting for additional games to complete at the end of the season when we notice people leaving the league. But I guess this isn't too big a deal, won't inactive players show they are inactive before the end of the season?

Please let me know your feedback.

Edited 7/26/2016 03:29:00

Thx Pushover. Then i will just wait for Season 21.

I like this idea, I think parallel groups is the way to go.

* A two group tier promotes the top player from each group into a 1 group tier
* A four group tier promotes the top player from each group into a 2 group tier

Not sure what you mean by this though, if there are no super promotions? I think nobody should promote more than one tier in the absence of dropouts. Do you mean that 1 person promotes whenever moving to a tier with a smaller number of groups, and 2 otherwise? Because that would be my suggestion.

Edge, when the season 20 thread comes up, please mention then that you want to join. I'm not gonna be maintaining multiple lists.

awesomeusername, typically when we have dropouts the gaps are filled using super promotions, for instance 3 spots in C are filled by 2 in D and 1 in E. Under the new system, 3 spots in C will be filled by 3 spots in D.

Since each tier may have multiple groups, you have to take half the promotions per group in order to promote to a tier with fewer groups.

Edited 7/26/2016 04:28:45

so, in this case, would top 2 promote from E1/E2 into D1/D2?

I think it is a better system then the current one and I do like the more groups used in Krzychu's idea.
However this is only 1/2 of what needs to be done and that is why it won't be effective enough.
This won't solve the problem the league has.

The main problem is the time it takes to finish and get ready for the next one.(3-4 months)

More groups solves only the issue of having more players at the bottom of hell in the last groups that will never see group A.(assuming the league is fast enough which it is not)

It does not change the fact that for a player starting from the last group it will take him years to get to group A (assuming he wins every promotion) simply because it takes 3-4 months for the season to finish.
I myself wanted to play the top group when I joined pr league, and after some 2 years the top group have all but retired before I even got the chance to fight them.
This was when pr league was still at most up to Group E.
This means that unless you make a big change you will not change anything.

That is your main issue here.

The problem is starting from the fact that in an RR if you have vacations and too many players it takes too long to progress.

1)Vacations will stall to a deadly stop the progress of the league.
There are only few solutions to this.
- Either remove vacations
- or make fewer players in every group to speed this up.

2)Too many players/groups will result in having the usual problems multiplied by a factor depending on how many players/groups you have.
This RR has too many players, adding more groups solves a bit this issue but it adds more administrative work.
More players = more games in an RR system.

You must find a way to increase the number of games being played at once which Krzychu's solution does cater for with more groups.
However it just leaves the system as it is, which is a bad state, it does not improve much the problem.

There are some solutions to this, some mentioned before:

-Like reducing group size and thus increasing the number of groups to fit all the players.
-Change the RR system for the bottom of the groups(as mentioned before).
-Divide the league in 2 leagues so you can have twice as many games playing at the same time.(bringing 3- 4 months to just 1.5-2 months)

If it was me I would do all of those(including Krzychu's solution) to ensure the league is as efficient as possible.

After all the faster it is the more fun we get from it.

Edited 7/26/2016 05:10:47

Yeah i will sign up then.

The RR system is not changing. Group size will not be reduced. I am not breaking up the league into multiple leagues. These are non-negotiable.

Vacations will not be disabled. They are a crucial part of the league and ensure that everyone can participate and still step away from warlight every now and then. Worst case scenario is that we give a delaying player lots of losses. This hasn't been an issue in the past.

I will not change boot time. Curious how that was never mentioned as a time saving device...

3 games at a time is in play, if the community seems to want it. But we are all in lots of games and I'd prefer to not increase participation requirements. So it'd have to be a lot of support for me to change my mind about this.

Increased number of promotions is also in play, but if we go this route I'd prefer to also increase group size so we aren't constantly shuffling players to different levels. I like the ratios the way they are now (2/7 promoting, 2/7 relegating, 3/7 staying put). I would be okay with 3/8 promoting, 3/8 relegating, and 2/8 staying put. Don't lecture me about how increasing group size makes things worse. I disagree and you won't be successful in convincing me.

I think reducing tiers by increasing number of groups is the change that will have most impact on the problem of it being too difficult for a new player to have a shot at the top group.

Hey,I didn't mean this. I never requested,but for some reason, some of the „top“ players ignore my posts on any threads... from much time before. So I just thought you could be as them,but looks like you're nice. Thanks.

Edited 7/26/2016 06:55:19

i prefer 2 games at a time. no need to rush things :P

I think your current proposal is nice.

Something different that I forgot to mention but that may be important:
In RR, there's this weird assignment of the first games. It seems that in groups of 7 often A plays B, B plays C, C plays A, D plays E, E plays F, F plays D, G play nobody. This is a little annoying as G has to wait for a game to finish to get their first game. Not sure if the same happens in groups of 6 or 8.
Note that you could speed up the league by creating a lot more work for yourself/organisers. Basically: don't make an RR tourney, but make all games yourself at fixed times, like in the seasonal. You start with 2, a week later the third, etc until after 4 weeks you started all 6 games. (But I guess this really is too much work.)

I think that's because the algorithm goes player-by-player/team-by-team and tries to fill their maximum if possible, hurting the lower letters.

So it's:

A: paired with B, paired with C

B: already with A, paired with C

C: already with A, already with B

D: paired with E, paired with F

E: already with D, paired with F

F: already with D, already with E

G: no one left to pair with

So if my hypothesis is correct, you want group sizes to be multiples of (game limit + 1). Notice that A, B, and C are off by themselves, as are D, E, and F. If you added an H, both G and H would get one game- but add an I, and G, H, and I form a similar complete group. I'm ignoring games as the tournament proceeds, though, so someone who takes that into account might find a different set of optimal group sizes.