Règlement 2013: points à améliorer/changer/conserver

[noctambule]

Clbrun il a pas ete question de mettre le TT en dernier (seul scoub s est senti mal pr proposer un truc pareil pr la cup le cdm).
T aimes bien le calendrier actuel, comme tout le monde.

Je propose simplement de switcher MR et GP
Je pense pas que quiconque soit perturbé de jouer le GP apres le TT. Le MR a la fin nous permettrait simplement de finir par une epreuve plus cool que le GP.

Mario >> ok cree donc un sondage.
Je voulais d abord exposer le pourquoi cette proposition et ce qu elle pouvait nous apporter.

[KVD]

One thing Drew and me absolutely agreed on (and it might even throw our attendance in the balance if it's not implimented) is the fact that we need a more elaborate KO phase (read double elimination).

Again my overall ranking suffered hugely from getting extreme draws in early KO rounds (Drew in BM 1/16 and Flo in MR quarter finals). One might argue that you can avoid such draws by performing better in the group stage, but that argument can very easily be countered by the lack of matches that matter to determine the result of the group stage. As an example, an important ranking deciding match of the MR group I was in, was ScouB versus me. ScouB got a lightning on the final lap of MC2, killing my chance to win that battle. One could argue that luck evens out, but obviously it doesn't even out when there are so few important matches in the poules (chickens).

The problem that follows later is: I get Neo in quarter finals. Due to the evident QF elimination, I get awarded a number of points equivalent to an MR final ranking of #5-8. Geo could have reached the final, reaping up a #2 MR ranking, without necessarily being a better MR player than me. He would have merely avoided Neo and Sami until the final round.
I'm not saying that I think that I am a better MR player than Geo (I actually don't think that I am), but it serves to illustrate how the current system fails to distinguish a total ranking due to the single elimination format, where harsh draws can really unfairly damage ones ranking. Drew facing ScouB in quarter finals of BM is another example. He was now effectively awarded a #5-8 equivalent BM ranking, eventhough I would bet solid money on him beating Neo for a #4 face-off (no offense Flo). He spent a trans Atlantic trip playing 3 hours of group stage (where only 3 matches of 4 maps really contributed to his final seed ranking...equivalent of about ~20 mins perhaps?). Then 30 minutes of playing later, he is sent packing due to an easy 1/16 draw (me  ) and an insanely tough draw in the quarters. No wonder he is strongly considering not to join us next year if this does not change.

If the tournaments would be isolated from each other and treat as seperate entities, that would be fine (like in a football world cup for example). But the MAIN ranking of the CDM is the overall ranking; therefore it should probably be the main incentive of the CDM to result in as accurate a reflection of overall SMK ability as possible within the alloted time. I strongly believe a double elimination system would be infinitely more accurate than the sudden KO system we have now. In conclusion, we need a system where more time is spent facing players of similar level and less time of facing players that are either much weaker or much stronger than you. I think a group stage of 4 groups and double elimination in the KOs would suit this purpose.

[noctambule]

what do you mean by double elimination in the KOs ? to play 2 matchs instead of just one ?

I agree we need to spent less time on the chickens (), but I'm not sure your proposal would be the right solution (don't say no as well). whatever, it is true this is something we need to talk about
division into 4 pools is a good start of thinking.

your problems in the pools are faced by everyone. I faced flo and sami lots of time (was the purpose sometimes ) in quarter because i got fucked in the pools in matchs I should have won.
moreover, the random switch we made this year on top16 and 1/4 was sometimes unfair (you in MR, drew in BM), but at least the system has been tested.

something to add btw : the ranking of the championship reflects your overall level not in life, but the d-day. you can over- or under-perform considering your level. exactly like in every sports/competitions.


otherwise, I think a winner of a 2players mode should gained 4 or 5pts bonus more because of playing under pressure for a long time

[KVD]

If you've ever watched a Street Fighter or Starcraft (or any other competitive gaming) tournament, you would know what double elimination is. All major gaming tournaments use it and it works brilliantly (well in my opinion).

http://en.wikipedia.org/wiki/Double-elimination_tournament

Clicking the scheme will illustrate it nicely also, if you're not interested in reading a wall of txt. Take into account that there is a winner and loser bracket. If you lose in winning bracket (where everyone starts), you move to losers bracket. If you lose in losers bracket, you're out. 

something to add btw : the ranking of the championship reflects your overall level not in life, but the d-day. you can over- or under-perform considering your level. exactly like in every sports/competitions.

Of course that is true and I think that is positive, it means surprises can happen. But on d-day, Drew could (would?) have ranked better than Flo in BM, it's just that the current system denied him this chance. I think we should fix that and double elimination does just that.

[djo]

I agree we need to spent less time on the chickens (), but I'm not sure your proposal would be the right solution (don't say no as well). whatever, it is true this is something we need to talk about
division into 4 pools is a good start of thinking.

Like CDM2012?

[noctambule]

If you've ever watched a Street Fighter or Starcraft (or any other competitive gaming) tournament, you would know what double elimination is. All major gaming tournaments use it and it works brilliantly (well in my opinion).

http://en.wikipedia.org/wiki/Double-elimination_tournament

Clicking the scheme will illustrate it nicely also, if you're not interested in reading a wall of txt. Take into account that there is a winner and loser bracket. If you lose in winning bracket (where everyone starts), you move to losers bracket. If you lose in losers bracket, you're out. 

something to add btw : the ranking of the championship reflects your overall level not in life, but the d-day. you can over- or under-perform considering your level. exactly like in every sports/competitions.

Of course that is true and I think that is positive, it means surprises can happen. But on d-day, Drew could (would?) have ranked better than Flo in BM, it's just that the current system denied him this chance. I think we should fix that and double elimination does just that.

it's ok I got it
just have to concentrate a lot not to read double pene...

definitely something we have to discuss.

[Anton]

Oui c'est le principe des loosers bracket comme dans les tournois de jeux de combat.
J'en avais déjà mentionné l'idée, mais pas sur le forum.
J'aimerais bien tester ce système en meeting dans un premier temps

[Neo, the Daenerys Targaryen of SMK]

C'est effectivement une idée à creuser. Et à tester (plusieurs fois).

[Alicia Kart]

Mais comment vous ferez pour la finale entre le meilleur de la Winner Bracket (qui n'a donc perdu aucun match) et celui de la Loser Bracket (qui en a donc perdu un), dans la mesure où le joueur de la Winner Bracket a toujours l'avantage de n'avoir à gagner qu'un seul match, là où celui de la Loser Bracket doit en gagner 2.

Or je ne vois pas vraiment nous faire une finale au premier à 2 matches "premier à 11" gagné (ou "premier à 5" pour le GP). Tout comme je ne vois pas ne donner aucun avantage au joueur de la Winner Bracket...

Une solution : ramener tous les matches des phases finales au premier à 5 (ou à 3 en GP).

[Anton]

le joueur de la Winner Bracket a toujours l'avantage de n'avoir à gagner qu'un seul match, là où celui de la Loser Bracket doit en gagner 2.

Oui sur le papier, après c'est sur il faut creuser l'idée et l'adapter sur le fait que le joueur qui revient des losers bracket doit d'abord gagner 11-x. A partir de là, la "vraie" finale commence, le 1er à 11.
C'est sur que wouah ça fait beaucoup d'heure de jeu

Mais j'insiste, à cogiter

[KVD]

Thanks for the open responses dudes, I would love you guys testing it during one of the many French SMK meetings!   

[djo]

Un système de ce type a déjà été testé en compétition, c'était à l'ODB et également lors des compétitions NTSC du CDM 2012 (http://www.championnat-smk.org/forum/index.php?topic=5455.0).
La présentation est différente mais on a bien un winner bracket et un loser bracket et même ensuite des niveaux inférieurs avec ceux à 2 défaites, 3 défaites, etc... puisque tous les tableaux ce juxtaposent "en 3 dimensions"!

Pour moi le test était déjà bien concluant (surtout pour le top8 ce qui est primordial) déjà avec seulement 8 rencontres. Avec 16 rencontres on aurait un top8 irréprochable pour constituer le tableau final.

Seulement un tel système fait un petit peu peur et est plus difficile à appréhender, il y a d'ailleurs un topic entier de discussion (où je démontrais pourtant que c'était le système le plus adéquat pour classer les joueurs selon leur niveau et cela sans devoir établir de tête de série)
http://www.championnat-smk.org/forum/index.php?topic=4786.0
topic qui a mené au système à 4 poules mis en place en 2012 mais pour moi le "système belge" est encore meilleur!

Bref je suis près à relancer la discussion

[clbrun]

i can t speak about THIS cdm since i didn't attend, but i guess the level is high enough in quaterfinals that there will always be some "too big matchs for quaterfinals", and the guys who will be eliminated there can't just blame the system.

i agree this year, the weird random in quaterfinals increases the "unfair feeling" and nullifies the following argument, but it s already known since many years that you have to be at the top in poules to ensure an easy path in the final phasis. and like harold said, YOU have to be at the top the d-day, instead of hoping the system will allow you several possibilities to compensate bad results. so on my opinion, you KVD may have been quite unlucky this year, but you can't blame the system for that.

i guess the matter is a bit "what is the aim of the cdm?" is it a "4 d-days competition", with an overall ranking that might be a bit unfair because some guy had to face another one in quaterfinal instead of the final ? for me, yes.

i'm aware of the double elim final phasis in fighting games, but for me the purpose is to compensate the shortness of the matchs : often, a duel consists in just one match (the first to win 2 or 3 rounds), and it s very very short. so the losers bracket is made to enable good players to stand one more chance.
in the smk cdm, i don t think a final phasis duel is so short/ random : maybe in 1/8th finals, but once you are in quarter finals, i don t think any duel is won just by luck, so there is no point creating a loser bracket (or at least i personnaly think it isn't as justified as in most fighting games tournaments). if the quarterfinals were "the first to win 2 or 3 rounds / 1 cup", i would agree that it s too random, but it's not the case. maybe you and drew are a bit disappointed by being eliminated prematurely in your favorite mode, but even i guess it's the rule d-day tournaments.

another problem of the double elim is the length increase of the poules : i guess we should shorten poules to make double elim, and this would mean less playtime for noobs/ average players. personnaly, i don t think it would be cool for them to divide their playing time by 2 just to ensure some strong players one more chance not to be eliminated. or maybe we should change the spirit of the cdm to make it an event for good players, where noobs are not as welcome as today (or at least they should be aware that they aren't supposed to play as much as today).

[KVD]

Well we need to win accuracy somewhere. If we're not gonna have a double elimination in KO, then the seeding resulting from group stage should be more accurate. As it is now you play 3 hours, but almost exclusively against players that are not your level, so scores can be determined with 99% accuracy beforehand. I would call such matches pointless (if im being blunt, i mean pointless scorewise, not in terms of fun). Right now the vast majority of CDM matches is of such pointless nature, almost making it not wortwhile to bother coming to France for. Then there are maybe 3 big matches between players that are your lvl (the result is hard to predict, they matter. Note that this is true for noobs and top players alike). Since these are very short, there's not always the opportunity to arrive at a proper difference in those.
 I still think double elim is the best way forward, but whichever system increases the number of meaningful matches would already be an improvement.

[djo]

Whichever system increases the number of meaningful matches would already be an improvement.

It's what the "belgian system" used last year for NTSC competition (you didn't played it) do and with a pretty good result at least for top 8 (most important for final title).
Here are an example of which opponents you could played (in green) with 15 rounds

After that you do a classic final phase!

For the result of this year, the problem was especially in BM where 3 of the 4 best players of the d-day were in the same group (Harold, Scoub and Drew) and with the random 1/4 we had a Drew-Scoub instead of Drew-Conor
Random 1/4 was really the badest idea...