Implemented RE: Round Robin (3 player) Finals

[Finchinator]

Round robin finals are lame for both players and spectators. I believe we can do away with round robin (3 player) finals without much consequence, but a community discussion is needed to see what is best. There is a prior thread on this for live tournaments here which I will link as a reference point, too. Let's get into why it is worse for both players and spectators.

Why round robin finals are bad for players?

Less is more when it comes to scheduling and playing series per round. More Pokemon may seem exciting, but it creates a minimum of two times as much schedule, preparation, and playing with the ugly chance of being far more with bracket resets. All of this after a long road to get to the finals is far from ideal. Having a larger first round may mean more byes, which lead to clunky 0-or-2 point R2 games, but this dwarfs in importance when compared to the finals.

Why round robin finals are bad for spectators?

The finals of any tournament is meant to be a spectacle where enthusiastic viewers can celebrate two of the strongest competitors duking it out with everything at stake. Competitive Pokemon has evolved into a more spectator-friendly product in so many ways, ranging from the adoption of Pokemon Showdown, replay threads, and SmogTours server years back to more modern conventions like tournament-specific matches tags, mandated replay rules, and the evolution of tournament coverage through site articles. The hype surrounding a 3-way final, which is oftentimes disconnected with series at different times or even resets across multiple days/weeks, is a huge downgrade. A conventional 2-way final is far more appealing and more linear to follow. A round robin final has less at stake in each individual game and lacks potential permenance until the second or third series in.

---

Ok, so we have established the quality-of-life hits from round robin finals. What is the issue of doing away with them? Logistics -- you are going to either exclude people from tournaments or end up with much more R1 byes in order to assure overall bracket sizes work for a clean, 2 player final rather than a 3-way round-robin. This discussion boils down to if we value the improvement in quality for players and spectators more or if we want to keep the closest sized bracket more.

I believe that the finals are far more important towards our community than the first round. I believe we can have some flexibility when starting tournaments to have more byes when a round robin final is otherwise needed. I believe that the best tournaments have a clean, 2 player final and prioritizing this over all else is for the best. I propose that we shift towards avoiding round robin finals at all cost moving forward. Please use this thread to discuss it as this topic; feel free to use recent examples from tournaments like Smogon Classic when appropriate -- thanks and have a nice day!

 

[QT]

I support this change. I think it makes sense in general, for all the reasons Finch outlined, but it's especially bad in cases where for instance you have a tournament that qualifies for other tournaments, like opens. If the last open gets to a round robin and has a bracket reset or two, you have to either ignore the impact on seeds (unfair to players who would have qualified with the points from finals) or delay the entire tournament (unfair to every other player, especially if it means playoffs last long enough to interfere with prep for important games in the next tour). And what do we give up for this? Round 1 becomes easier because there are more byes. Maybe this is just me, but I'm not exactly quaking in my boots about that - unless you get really unlucky in who you get matched with, is round 1 not going to be really easy anyway? When was the last time you were sweating up a storm in your round 1 games?

 

[BeeOrSomething]

100% agree. 3-way finals are completely terrible. There's an awful feeling as a spectator trying to watch hype games in finals just knowing that at any moment the 3 games u waited a week for end up going even and u gotta wait another week to see the conclusion, potentially even longer if multiple resets happen. Additional byes in the first round are barely a meaningful consequence. Quality and enjoyability of finals is by far much more important.

 

[Star]

I’ve generally been pretty indifferent to round robin finals as a player. Maybe mildly annoying but not that impactful and I don’t have a strong opinion on them staying/going.

I do expect that community sentiment will lean towards abolishing them, and if we go that way I think adding byes is pretty clearly better than cutting people out at signups. Round 1 is a little less competitive obviously but there’s always a shitload of byes anyway and I don’t think adding more is really a huge concern. Far better to me than the alternative of not letting a ton of people play.

As for the exact numbers I like the cap at 20% excluded that D4 outlined here: https://www.smogon.com/forums/threads/abolish-round-robin-finals-from-live-tournaments.3739593/

 

[goldmason]

Agreed with Star above. Really do not like the idea of a bunch of people not being able to play for what is part of a much larger tournament because we only accomodate 2^n sizings. I think that the first round (which is already a bit meme anyways; lotsa dropouts and inactives etc) being less slightly less competitive is a fine way to go. I like the 20% cap proposed in the above linked thread a lot.

 

[ABR]

I support the general sentiments above but would up the max bye threshold past 20%

To me there is no real functional difference between "some people get byes" and "slightly more people get byes" - like theoretically the limit is 50%...

byes > cutting ppl > rr

 

[Kingler]

I’m just gonna add my two cents because i played my first lc open win as an RR finals and while it was resolved in the first set of games, we already have week long tiebreakers so its understood tournaments can extend and players know to account for that. If the option like star and abr said is either cut a bunch of people or have an RR finals i lean towards the latter. No idea about the cutoff cap or whatever I dont really give a shit and i agree with abr about byes.

 

[Vileman]

I’m just gonna add my two cents because i played my first lc open win as an RR finals and while it was resolved in the first set of games, we already have week long tiebreakers so its understood tournaments can extend and players know to account for that. If the option like star and abr said is either cut a bunch of people or have an RR finals i lean towards the latter. No idea about the cutoff cap or whatever I dont really give a shit and i agree with abr about byes.

Thanks grandpa

 

[teal6]

Round Robin finals are awful. Getting rid of them is good for everyone involved.

 

[Eledyr]

Hi, this thread has not been active at all in recent times, but I also support getting ENTIRELY rid of 3 ways-finals. It adds unnecessary awkward possibilities for a tournament to simply not end in a reasonable amount of time. This can easily be fixed; simply never allow a tournament to have a number of slots equals to 3*2^n. The issue comes entirely in: how many subs should we allow in tournaments? Ruling right now is 25% of subs, which I believe is not fine. To get some numbers, it means that instead of getting a tournament with 48 sign-ups (making it a round robin), we can go as high as 64, with 16 subs, which means there'd be... 48 players! Yes, with 25% subs, we can avoid getting a round robin finals. However, you can point out that with no in-between between a 2^n and a 2^n+1 tour, there'll be a lot of players left unable to play. Well, it can go as high as half the size of the tournament in the setup of 25% subs, so for example with 95 sign-ups there'd be a 64-men tours, with 31 subs left. But if there was just a single additional sign-ups, the tournament would become a 128 men tournament with 32 byes. This can be fixed by increasing the number of subs. A number that I like is 37,5% of subs (which is 3/8): let's take the example of a 128 tournaments players. With 3/8 of subs, it means up to 48 byes could be assigned, meaning that if 95 people signs up for the tournament, we easily get the good number of people to make a 128 men tournament. This means that instead of getting a cut-off (here, I mean jumping from a 2^n tournament to a 2^n+1 tournament) at 1/2 of tournament size, we get a cut-off of 25% of the tournament size, meaning that, for example, a tournament of 64 men would get up as is with up to 79 sign-ups, which means only 15 people who signed up would be subs. This way, we always avoid to get a 3-way finals, while also avoiding awkward numbers of ins because there isn't enough players to make a bigger tour, leading to dozens of people forced to sub.

I hope I made my point clear. Tl;dr: let's get rid of RR, and increase the subs to up to 37,5% (3/8) of the tournament size.

 

[Boat]

I support the general sentiments above but would up the max bye threshold past 20%

To me there is no real functional difference between "some people get byes" and "slightly more people get byes" - like theoretically the limit is 50%...

byes > cutting ppl > rr

Personally, I've always disliked First Come First Serve. I'd rather have a round one with only one real game (everything else byes) than have to exclude someone from a tournament they registered for. Byes are ultimately not that impactful on the outcome of a tournament, and getting hoed by the 5/8ths Rule is a shitty tournament experience. Having a lot of byes is especially not impactful on Smogon because people can sign up as a substitute during the week, an advantage not afforded to live events.

Killing First Come First Serve and RR finals at the same time would be good.

The ranges would be the following :

Size 8 : 5-8
Size 16 : 9 - 16
Size 32 : 17 - 32
Size 64 : 33 - 64
Size 128 : 65 - 128
Size 256 : 129 - 256
Size 512 : 513 - 1028
etc

At a certain point, it does become a bit silly to have a Round 1 with a handful of matches. At that point, you could treat it like an extension, and have (A vs B) vs C in Round 1 (technically this would be starting in Round 2). You'd have to be pretty flexible with extensions since some people can only play weekends, but if that's undesirable, just run the very, very small Round 1 and accept it for what it is; a bit silly and not that big of a deal.

 

[LBN]

Not much to contribute from me but anything that sees round robins get nuked away will get my full support

 

[Merritt]

Because I ran the numbers recently, here's a look at how removing round robin finals tour sizes from Slam and Classic would impact things. It depends on where a new cutoff between sub list vs add byes would be, but one of these two scenarios has to occur. These are the numbers from last year, and tours which already went to a 512 bracket size were excluded.

Ubers Open - 375 signups, options are 137 byes (119 pairings in round 1 / ~63% of signups play r1) or add 119 subs (~46% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 9 byes in round 1 / ~97% of signups play r1.

NU Open - 368 signups, options are 144 byes (112 pairings in round 1 / ~61% of signups play r1) or 112 subs (~44% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 16 byes in round 1 / ~95% of signups play r1.

RU Open - 363 signups, options are 149 byes (107 pairings in round 1 / ~59% of signups play r1) or 107 subs (~41% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 21 byes in round 1 / ~94% of signups play r1.

GSC Cup - 326 signups, options are 186 byes (70 pairings in round 1 / ~43% of signups play r1) or 70 subs (~27% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 58 byes in round 1 / ~82% of signups play r1.

PU Open - 312 signups, options are 200 byes (56 pairings in round 1 / ~36% of signups play r1) or 56 subs (~22% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 72 byes in round 1 / ~77% of signups play r1.

RBY Cup - 308 signups, options are 204 byes (52 pairings in round 1 / ~33% of signups play r1) or 52 subs (~20% of initial bracket would need to drop out or be inactive for everybody to be subbed in). 384 bracket implementation has 76 byes in round 1 / ~75% of signups play r1.

To take it to the extremes for the 256/384/512 bracket sizes, currently the cutoff point for a 384 bracket is 385 signups going to a 512 bracket with 127 byes (129 pairings in round 1 / ~67% of signups play r1). Inversely, there can be up to 287 signups and still be a 256 sized bracket, resulting in 31 subs (~12% of initial bracket would need to drop out or be inactive for everybody to be subbed in).

I'm not a massive fan of round robin finals, they're a bit of a pain logistically and the hype of a reset only goes so far. That said, I do think that implementations which give more than 50% of round 1 a bye are somewhat unfair to the people who are essentially dealing with a play in, and means we probably need to rework our points system due to the increased likelihood of people getting to round 3+ on a slew of free rounds (bye in round 1 of course, but this also will increase the number of completely inactive players who make it to round 2).

We could potentially use the current double elimination numbers. This would mean tours would all fall somewhere between having a maximum of ~60% of players having a round 1 bye or up to 25% of bracket size being on the sub list (1 in 4 players would need to be inactive or drop out). To me both of these numbers, but particularly the maximum sub list size, aren't ideal. Quick illustration of these numbers below.

128 Bracket:
- low end 81 signups with 47 byes (17 pairings in round 1 / ~42% of signups play r1)
- high end 160 signups with 32 subs (25% of initial bracket)
256 Bracket:
- low end 161 signups with 95 byes (33 pairings in round 1 / ~41% of signups play r1)
- high end 320 signups with 64 subs (25% of initial bracket)
512 Bracket:
- low end 321 signups with 191 byes (65 pairings in round 1 / ~40.5% of signups play r1)
- high end 640 signups with 128 subs (25% of initial bracket)


I'm not saying that removing round robin is impossible, but I want to make sure that people who are asking for it understand exactly what changes here.

 

[Finchinator]

I think this boils down to picking between the following two options:

• Compromising the first round with the potential for far more byes than we have been accustom to
• Compromising the finals with some potentially awkward (disjointed and/or repeating) round robins

Merritt's numbers do a great job illustraiting how debilitating the former would be to the start of the tournament. Seeing a large fraction of people at random getting a pass to r2 is not ideal, but I personally think this is worth the trade-off.

A clean, 2 player final is a huge win for the community, in my opinion. Competitive Pokemon has evolved into a more spectator-friendly product in so many ways. The hype surrounding a 3 player final, which is oftentimes disconnected with series at different times or even resets across multiple days/weeks, is a huge downgrade. A conventional 2 player final is far more appealing and more linear to follow. A round robin final has less at stake in each individual game and lacks potential permenance until the second or third series in. This has a lot of value to me.

In short: is it worth sacrificing a level first round to assure a clean, 2 player final? I think so.

 

[teal6]

In short: is it worth sacrificing a level first round to assure a clean, 2 player final? I think so.

Correct, this is the ultimate consideration (per your post and the few above you). I also believe the answer is yes. We should eliminate RR finals.

 

[Merritt]

We are removing Round Robin finals from tournaments in the future. Tournaments generated will now automatically scale up to the next 2^n sized bracket, filling in all missing slots with byes. This will also remove all cases of an initial substitute list from tournaments. This applies to both single elimination as well as double elimination tournaments.

This change is now reflected in the bracketmaker. Thank you all for your feedback. Potential changes to scoring for Round 1 in tournaments like Classic, Slam, and Smogon Tour will be discussed in the immediate future.

If you are hosting an ongoing tournament you MUST read the following.

For ongoing tournaments that will have a round robin finals, you will have to generate your bracket slightly differently. After inputting the list of players for the following round, you will need to go to the bottom of the page where it says "Pairing mode", select "Advanced" and then select the "Include 3*2^n (legacy)" option so that the correct bracket size can generate. It should look like the image below before you click "Create".

If you generate your bracket and notice a large number of byes in Round 2 or later, contact a TD immediately so we can help correct things.


On Double Elimination Tournaments:

This change will result in many double elimination tournaments not needing to have a losers bracket in Round 2. If at least 25% of your Round 1 bracket is made up of byes, then your Round 2 losers bracket does not need to play since every real user in it should be paired against a bye. If you're hosting a double elimination tournament and your Round 2 winners bracket and Round 2 losers bracket have a different number of pairings, then something has gone wrong and your Round 2 losers bracket probably does not need to play. If you're unsure, please contact a TD for assistance.

More comprehensive details will be provided in a planned bracketmaker guide in the near future.
The bracketmaker guide is out.