COMMUNAUTÉ > Tournois & Manifestations
[ATP]Classement des Joueurs Français de Warmachine / Hordes
SHWTD:
Parce que j'aime remuer la merde que je dénonce (et pour montrer à David que je ne me contente pas de mettre des coups de shotgun dans les idées des gros bills assoiffés de gloâre ! ;D), voici la transcription d'une idée venue "de l'autre côté". Qu'en pensez-vous ?
--- Citer ---A while ago there was a thread on an ELO rating system, where I mentioned a tournament system alternative we’ve come up with that gets around some of the organisational difficulties we have in our events. I promised some details, but didn’t get the time to write it up until now. I’d love to hear what the forum, with all its experience as tournament organisers, thinks of this alternative.
The problem:
We experience several issues when running tournaments:
- There is a wide range of skill and experience among the players, from very fresh to hardened veterans.
- There are different mentalities, from very competitive to very casual.
- We often have time constraints that prevent us from running the appropriate number of rounds for a given amount of players.
This last point means there can be several people with all wins at the end of the tournament, and you need to use tie breakers to name the final winner. The problem there is that there are no truly good candidates for tie-breakers in Warmachine/Hordes. SoS is very random with a small number of rounds, and scenario or army scoring disadvantages certain list builds. Also, it's very anticlimactic and unpopular to declare a winner based on the performance of other players. People feel they had no control over their loss.
The first two problems are mostly only a problem if these types of players are paired against other types. Two competitive players can give each other a fun game, as can two inexperienced players. A competitive player (in a tournament setting) who is looking to win the whole thing stomping on a newer player is not an interesting match. The new player is demoralized and the veteran is bored and has his SoS ruined.
(An attempt at) a solution:
I wanted a system that would let players predominantly face other players of a similar skill and temperament, that would always produce a 'final table', and was to a minimal extent subject to random or unknown factors out of the players' control. The inspiration was the McMahon tournament system, which is a variant of Swiss that's popular in the Go community. This introduces a number of bands that players can be sorted in, based in that case on Go rank.
The players beginning in each of these bands are accorded a number of 'virtual tournament points' or McMahon points, and these are added to your tournament points to calculate pairings. This has the effect of creating several strata of players that mostly play each other, but where you can drift to the upper/lower strata if you are/aren't successful in your games. McMahon isn't very suited for Warmachine though, so I wanted a variant that met our requirements a little better. Obviously, this does require some kind of way to assess the players before the tournament. You can use any kind of ranking, but a stated preference of the individual players also works perfectly fine.
So, the modified McMahon (McPirate? McMulg? Bad with names.) that we tried (using an ELO ranking in our case) is as follows:
- Sort the participating players according to skill or preference.
- Starting from the top, introduce a band per every 2^(number of rounds) players. So, for a 3 round tournament you would have a number of 8-player bands.
- A player can choose to play 1 band lower than he qualifies for, if he prefers. Another player is then promoted from a lower band.
- From the bottom band up, allocate players in each band a increasing amount of McMahon points, starting with 0 points. So, a 20 player tournament with 3 rounds would have 8/8/4 players with 2/1/0 McMahon points.
- Pair players at random with other players within their own band.
- Play!
- After each round, pair on Total Points (TP + McM), then TP, then CP, then AP. This accomplishes the stratum effect.
- At the end of the tournament, sort players first on TP, then on McM points, then CP, then AP.
Note: If you DO have an appropriate number of rounds, you can still choose bands to get the stratum effect and start veterans off apart from beginners. The 2^(number of rounds) is an upper limit, not a lower one.
Note 2: As mentioned, this will work just as well if you have players mark their preferred level of competition on their registration sheet, you don't strictly need a ranking. As long as the bands do not exceed the 2^n number, the size does not matter much.
This accomplishes several of our goals:
- Foremost, it keeps everyone in the same tournament, and everyone can have fun and challenging matches.
- If you have an appropriate number of rounds, the player with the most wins will obviously still win, regardless of the band he started in.
- If you have fewer rounds, in the final round 'n', at the most 2 players from the top band will have n-1 wins and will face each other.
- If the results go to tie-break for the final standings, the winner of this ‘final table’ game is guaranteed to have the best tie-breaker, but also to have played all of the hardest opponents.
This amounts to non-random Strength of Schedule. A rating that puts you in the top band affords you a definite opportunity for the win, but you will have to earn it by beating all the other opponents with a high rating.
- Players who prefer more casual games will have mostly played other casual players, competitive players are thrown in with the other psycho-crazy killers. Casual players who start low but are on a killing streak can still finish in the top prizes.
- The second-most skilled player that meets the most-skilled player early on still has a very good shot at second place, which is rarely true in straight Swiss.
- It's much easier to calculate on the fly than SoS, and drops and byes matter less.
One unfortunate side-effect when you're using a ranking is that people who participate in their first ranked tournament might not be well-represented by the default ELO rating of 1500. If (and only if) there is not an appropriate number of rounds, it is a fact they will not get the same shot at the #1 spot that players from the top band get for winning all their games (although the games were also easier).
In normal Swiss, however, that shot at the #1 spot would basically have been a lottery anyway.
In this system, if they do well against opponents from their starting band, their rating will increase sharply and they will be in the higher bands the next tournament and balance is restored to the universe.
Still, to recognize this, or if you have a special band for beginning players for example, you could have an award for the best-scoring player per starting band.
Any thoughts? I apologize if it all seems very abstract.
--- Fin de citation ---
Adrezo:
Je vais être un peu à contre courant, mais je trouve ca intéressant cette idée de classement. Et je suis sur le c** devant le boulot effectué ^^
J'ai lu que Warmachine était un jeu de dés et que tout pouvait aussi se jouer sur un jet de dé et que donc c'était du hasard. Je ne suis pas d'accord avec ça pour 2 raisons :
- Si Eric joue encore à Warmachine depuis tout ce temps c'est qu'il n'y a pas que de la moule aux dés.
- La connaissance de son armée et de celle de l'adversaire, le positionnement sur le champ de bataille, l'utilisation du Feat, du focus, etc... font que la stratégie et le sens tactic sont de mises !
A partir du moment où le hasard n'est pas pré-dominant, un classement peut être intéressant.
Je me rappel d'un cas similaire de dialogue lors de la création de la Fédération frnçaise de Mémoire 44. ::)
Un des soucis du classement vient du fait que tu prends en compte le nombre de participant dans le tournois et non le nombre de parties jouées pour gagner. Ce qui fait que le mec qui va à tout les tournois est mieux classé que le mec qui gagne deux tournois. Tu pourrais donner simplement créer un classement participation indépendant du reste. Bref.
Pour rester sur l'idée d'établir un classement il faudrait, je pense, instaurer trop de paramètres (règles communes à tout les tournois, scénarios conventionnoés, etc...). Ce qui semble amener trop de contrainte pour l'apport d'un classement.
Après il reste toujours un moyen simple pour établir un classement du genre :
- Victoire = 5 points
- Défaite = 2 points
- Participation
- Points pour classement = (PTS Victoire + PTS défaites + PTS participation) / par nombre de bataille
Ca permet de voir quel joueur gagne globalement ses batailles. Car le plus important n'est pas de savoir si c'était un tournois à plein de personnes ou pas. Mais de savoir si le joueur remporte ses affrontements.
Mon avis vaut ce qu'il vaut, étant nouveau parmis vous, mais bon je le donne :P
Portal:
Je dis juste que c'était une bonne idée de remettre ça sur la table à 5 jours de l'Open :)
Ça mets une vieille pression à HudsonH et Q².
SHWTD:
--- Citation de: Portal le 21 mai 2012 à 15:45:27 ---Je dis juste que c'était une bonne idée de remettre ça sur la table à 5 jours de l'Open :)
Ça mets une vieille pression à HudsonH et Q².
--- Fin de citation ---
Et ça n'a pas de prix ! :)
HudsonH:
--- Citation de: Portal le 21 mai 2012 à 15:45:27 ---Je dis juste que c'était une bonne idée de remettre ça sur la table à 5 jours de l'Open :)
Ça mets une vieille pression à HudsonH et Q².
--- Fin de citation ---
La vache, je vais pas en dormir cette nuit moi ^^
Navigation
[#] Page suivante
[*] Page précédente
Utiliser la version classique