Matchmaking based on tier does not appropriately match teams. New design idea.

GrendelGrendel Member Posts: 305
edited May 21 in Guild Wars Discussion
What is the point of matchmaking?

I believe it is to put 2 teams together that each have a chance of defeating the other.

The current system (based on # of players and guild tier) does not do that. The current system can work 100% as designed and still put 2 teams together that are completely unbalanced. Mediocre guilds are often facing the top-rated guilds, and the worst players from a guild are often facing the best players from another guild - simply because they are in the same tier.

The only way to fairly match teams is to base the matching on:
1) The number of players.
2) The average VP earned by the players on each team over the past 4 (or 6 or 8?) battles.

With this system, the vast majority of battles will be competitive.

This will cause top-ranked guilds to play each other a lot, which will tighten up the leaderboards. But that should not be a bad thing. Guilds should not be top-rated because 80% of their battles were against opponents that had no chance. (I do believe a mechanism should be put in place to prevent the same guilds being matched more than once per war.)

Uneven Teams
When teams are uneven, the team with fewer players should be given an extra 18 attacks for each player they are down. I realize this means some players will get a small boost to their personal RP, but I believe this is the simplest and most accurate way to even things out.
JLCATLAS-ZNanajjaQ20PasteSilverhawkcyclon98
«1

Comments

  • ATLAS-ZATLAS-Z Member Posts: 4,015
    I actually like the current system. There will be uneven matches at the beginning of a new season and as the season continues the matches will become more and more competitive. This seems reasonable to me



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
    KarajocaPasteGalexNanajjaQ20Silverhawk
  • GrendelGrendel Member Posts: 305
    ATLAS-Z said:

    I actually like the current system. There will be uneven matches at the beginning of a new season and as the season continues the matches will become more and more competitive. This seems reasonable to me

    I have to disagree. Two guilds being in the same tier does not in any way indicate that two particular sets of players from those guilds are fairly matched against each other. One team could consist of the best players of a guild, and the other team the worst. Also, one guild may have just broken into a tier while the other is on the verge of leaving it.

    I admit it does get somewhat better as the season progresses, but that is only because matching is so ridiculously bad at the beginning of the season. The best guild can be matched against the worst guild.

    Today we (a mediocre guild) are matched against a high-ranking guild just because we are both silver 3; and this is almost half-way through the season.

    Let me ask this: What percent of your guild's battles would you consider competitive (meaning that you had about a 50/50 chance of winning)?
    ADPaqNanajjaQ20Silverhawk
  • rogueDSrogueDS Member Posts: 573
    For gw though if have lower level survivors the points balance out. Should be equal to that in difficulty whether have level 15 or 27. @ATLAS-Z @avelardez @JayZ any input here. Maybe someone with 2 accounts can show this better. Try to clear same maps with an account at a lower level and see how much points differ? 4e for one with 27s and one with 16 should give the same guild rewards and rewards points correct. @Fluxxx.
  • BurmeliinisBurmeliinis Member Posts: 642
    The problem with @Grendel 's idea is that it deincentivizes you to win. For each time you win or play really well, you get hit by getting matched against stronger opponents. It's like playing in the playoffs, but with no prize for the ones that advance. Especially now that the guild shop does not have better rewards for the higher tiers it would only lead to the strange situation that you would need to try and optimize your victories so you win with the minimum amount of points AND make sure that top players also play a lot of PvP to minimize their personal scores.

    I think the current system, when working, is pretty good. Strong teams SHOULD have an easier time in GW.

    Personally I think the idea of a separate playoff tournament maybe at the end of the season would be good, and there your suggestion would make a lot of sense.
    Ingame username: Jubjab
    PasteSilverhawkcyclon98
  • GrendelGrendel Member Posts: 305

    The problem with @Grendel 's idea is that it deincentivizes you to win. For each time you win or play really well, you get hit by getting matched against stronger opponents. It's like playing in the playoffs, but with no prize for the ones that advance. Especially now that the guild shop does not have better rewards for the higher tiers it would only lead to the strange situation that you would need to try and optimize your victories so you win with the minimum amount of points AND make sure that top players also play a lot of PvP to minimize their personal scores.

    I think the current system, when working, is pretty good. Strong teams SHOULD have an easier time in GW.

    Personally I think the idea of a separate playoff tournament maybe at the end of the season would be good, and there your suggestion would make a lot of sense.

    I guess I oversimplify things. I picture teams playing as well as they can, getting the best scores they can, and winning whenever they can. The teams that score the most and win the most will rise to the top.

    If a team deliberately underplays in order to gain the advantage of playing weaker teams, in my humble opinion that team is a loser regardless of where they fall in the rankings.

    Think about it this way. With the current system, the guild ranking ultimately depends on how many times a top guild is matched against another top guild (and therefore faces the rare potential of losing.) So TopGuildA may beat TopGuildB in the rankings simply because TopGuildA was lucky enough to not face as many difficult battles as TopGuildB. So with the top 10 (or 20), you could very easily say that luck in matching has as much to do with the final rankings as player skill, tactics, or coordination.

    At least the system I am proposing takes most of the luck out of the leader boards.
    ATLAS-ZQ20
  • BurmeliinisBurmeliinis Member Posts: 642
    Perhaps, but I think that for the vast majority of players and guilds, the Guild Ranking is of low importance (I mean, you can only even see the top 30 guilds) and the main incentive is to win in order to get 1.5 x reward points. I don't think it is any more fair that a bad team would get 1.5 x points simply because they get matched against a even worse team, while a good team would get less reward points because they get matched against other top teams.
    Ingame username: Jubjab
    ATLAS-ZRiotZappaNanajja
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    edited May 22
    @Grendel have you ever heard the phrase "bring your A-game"?

    It's used frequently when to very strong opposing forces meet usually in some type of sporting event.

    If you had to bring your A game in every single match...? I can't see that being a good thing.

    Also if the Cleveland Browns only had to play against teams like the Detroit Lions or the Miami Dolphins then they probably would make the playoffs every year. And they're definitely not a playoff worthy team.



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
    Silverhawk
  • rogueDSrogueDS Member Posts: 573
    I dont think you realize how much goes into GW. Our guild has pictures of where to hit charts and sheets of what everyone's max level is. We talk on line through whole battle. Also taking on 39+ takes gold to heal and play. I take back my need more attacks statement. If I dont use gold I cant finish attacks of clearing 38+ for the guild I am in. It is by no means all sheer luck. And being your a game all the time what the hell. I do. We won in a 5 on 6 us being the 5 and still won. We planned everything and where smart about all attacks. I did all 40+ kill alls at end when we didn't have enough attacks to clear a sector.
  • GrendelGrendel Member Posts: 305
    edited May 23
    rogueDS said:

    I dont think you realize how much goes into GW. Our guild has pictures of where to hit charts and sheets of what everyone's max level is. We talk on line through whole battle. Also taking on 39+ takes gold to heal and play. I take back my need more attacks statement. If I dont use gold I cant finish attacks of clearing 38+ for the guild I am in. It is by no means all sheer luck. And being your a game all the time what the hell. I do. We won in a 5 on 6 us being the 5 and still won. We planned everything and where smart about all attacks. I did all 40+ kill alls at end when we didn't have enough attacks to clear a sector.

    The luck I am talking about has nothing to do with game-play. In fact, the luck I am talking about overshadows all of the game-play and tactics you mentioned.

    Suppose you have your guild and another guild that is completely equal to it. At the end of the war / season, the guild who has the most vp between the two of you is most likely to be the one that faced the most non-competitive opponents. The final results are based more on WHO you played instead of HOW you played because luck determines whether each match was against an opponent who could actually win.

    To me, the only answer is to try to make each and every match competitive. The only way to do that is to use better data than guild tier to match opponents. This would improve the experience for the 99% of the guilds that aren't at the top, and it would give some meaning to the leaderboards.

    I understand that top guilds will hate this idea; it means more time / gold / tools because they will have to fight to win more battles and not get easy wins as often.
    Q20ADPaqSilverhawk
  • Nova_TWDNova_TWD Member Posts: 22
    > @Grendel said:
    > (Quote)
    > The luck I am talking about has nothing to do with game-play. In fact, the luck I am talking about overshadows all of the game-play and tactics you mentioned.
    >
    > Suppose you have your guild and another guild that is completely equal to it. At the end of the war / season, the guild who has the most vp between the two of you is most likely to be the one that faced the most non-competitive opponents. The final results are based more on WHO you played instead of HOW you played because luck determines whether each match was against an opponent who could actually win.
    >
    > To me, the only answer is to try to make each and every match competitive. The only way to do that is to use better data than guild tier to match opponents. This would improve the experience for the 99% of the guilds that aren't at the top, and it would give some meaning to the leaderboards.
    >
    > I understand that top guilds will hate this idea; it means more time / gold / tools because they will have to fight to win more battles and not get easy wins as often.

    Mavericks OG faced 3 of the best teams in the world this war. The first two days they battled Red Machine, the third day they played against Team Poland and in the final battle, they faced WW:)
    TCBRITOATLAS-Z
  • Nova_TWDNova_TWD Member Posts: 22
    Very easy battles, i agree with you 100%😄😉
    NanajjaATLAS-ZSilverhawk
  • GrendelGrendel Member Posts: 305
    edited May 24
    Nova_TWD said:

    >
    Mavericks OG faced 3 of the best teams in the world this war. The first two days they battled Red Machine, the third day they played against Team Poland and in the final battle, they faced WW:)

    That's great. But I am not saying that top guilds never (or even rarely) face other top guilds. I am saying every match should be as competitive as possible - from the bottom guild to the top guild.

    Also, did the other top guilds that Mavericks OG are competing with on the leaderboards also face 3 of the best teams this war? If not, then that is the perfect illustration of the luck factor I am talking about.

    If some of the top teams played zero or 1 tough battle this war while other top teams played 3 or 4 tough battles this war, then the final rankings are largely based on luck.

    The discussion has shifted to top teams, but I am really more interested in creating a matching system that competitively matches up the 99% of guilds who are not on the leaderboards.
    Q20Silverhawk
  • GrendelGrendel Member Posts: 305
    JayZ said:

    Honestly, this should not be that hard. I’m not an expert on video game coding but I regularly code in statistical packages for work, and it should be simple to implement a reasonable set of rules like the following:



    1. Identify all teams with 10, 9, 8, etc. number of players. Group them all together so they can only play each other. This is THE most important step and removes 99% of people’s complaints.



    2. Within a group of teams that have 10 players, group teams based on their total points for the season. At the beginning, these groups can be broad. Say, 12 teams in each group. Start from the top and work your way down until every 10 player team is placed in a group.



    3. Within each of these subgroups, randomly match teams. So if you are a top 12 10-person team, you will play another top 12 10-person team. 13-24 rank teams will play 13-24 rank teams. And so on.



    4. As the season gets towards the end, subgroups can be narrower. So in the last week, maybe only top 4 teams play each other.



    5. Finally, no team should have to play the same team twice in a week.



    This method solves two problems:

    - First, no team will be at a disadvantage playing 5v6 or 8v10

    - Second, every team will play teams at roughly the same tier as them with a bit of RNG



    If there are uneven numbers of teams in a 10-player grouping, the lowest ranked team can play a bot.

    This is great. Thank you. It's almost exactly what I am asking for. Match teams based on points from the players participating in the battle - not based on the guild's points.

    But I would handle situations where one team has more players differently. If TeamA has one player more than TeamB, then TeamB should split an extra 18 attacks. If the difference is two players, the team with fewer players should split an extra 36 attacks to make up for the missing players. This isn't something that can be abused by players, because it would happen outside of their control (during matching). And I'll bet it would stop anyone from complaining about mismatched numbers of players.
    Q20
  • DaiikoDaiiko Member Posts: 102
    @Fluxxx hier j'ai fais ma dernière bataille dimanche 10 VS 10 nous avons gagné mais je n'ai pas ressus les points de victoires. Est-ce ce que c'est possible de regarder ça ?
    reekookr
  • ghost_pepperghost_pepper Member Posts: 327
    @Fluxxx To continue with the sports analogy, we have pro teams, college teams, high school teams, middle school teams, and herds of toddlers on the sidewalk all getting matched together. And this is fair as long as there are the same number of players? Who thought this up? Come on...

    Other than the first week of joke matching, the no-show issue brought up really is a problem. When your guild is down TWO players, and then you have a no-show, being down 3 is a horrible handicap. That happened to me. Luckily, the other group had a no show too. And had a 30 point player (toddler from the first analogy). We ended up winning, probably because the other team didn't understand area bonuses.

    Btw, area bonuses are not explained clearly. There is a bunch of stuff which needs to be explained IN GAME before any appropriate matching system can yield successful results. I get the sense that clueless teams are still clueless with gw. They are not learning it because it needs better in game explanation.
  • grzechExpgrzechExp Member Posts: 39
    ATLAS-Z said:

    I actually like the current system. There will be uneven matches at the beginning of a new season and as the season continues the matches will become more and more competitive. This seems reasonable to me

    Great Joke :smile:
    Usually the match between 2 equally skilled teams from the same tier ends up with a difference of up to 300 points.
    With additional player having 18 battles they can close 4C with 8 attacks and that is 312 points. And have still 10 left.
    1 player is a HUUUUUGE difference.

    @Fluxxx can you share a data on how many times teams in tiers from GOLD I and above had won with the team from the same tier having 1 player less?
    Percantage would be nice :smile:

    Honestly i think that there was mistake on the design phase. And honestly i even read it wrongly at the beggining.

    I interpretted it that in 2nd point search is expanded +/- tier. Not a participant.
    Is there a change that this will be changed?

    1. Matchmaking will prioritize finding you a match with the same number of participants in the same tier.
    2. If a guild with the same number of participants isn't found in your tier, the search will be expanded to look for a guild from ±1 tier..
    3. If the matchmaking still fails to find you an opponent, the search will look for an opponent in the tiers below. Since in this case your guild would be matched against a lower tier guild, the opponent can have up to 2 more participants than your team depending on the tier difference. 1-2 tier difference 1 player. 3-4 tier difference 2 players.

    Such a small change shall not increase a load much when running the match making script.
    Till that time playing less than 10 is to big of a risk.
    CrocodileFirekidignas
  • FirekidFirekid Member Posts: 2,527
    @grzechExp that is probably the most elegant and easiest way of sorting this out I’ve seen.
  • FluxxxFluxxx Community Manager Posts: 886
    Thanks for all your suggestions. We've been focusing on getting the matchmaking to work like it's supposed to, and it seems that today everyone got matched evenly, with a very small percentage of +-1 matchups.
    We'll discuss further on how to polish matchmaking, and I definitely see some good ideas here.
    ATLAS-ZPaste
  • ignasignas Member Posts: 369
    @Fluxxx Will those that have less people be compensated? I cannot see a way to beat guilds from the same tier having 18 attacks less. It’s not a war but execution.
    grzechExp
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    > @Fluxxx said:
    > Thanks for all your suggestions. We've been focusing on getting the matchmaking to work like it's supposed to, and it seems that today everyone got matched evenly, with a very small percentage of +-1 matchups.
    > We'll discuss further on how to polish matchmaking, and I definitely see some good ideas here.

    Thanks @Fluxxx this has been the experience of my whole guild family pretty much since you sorted the matchmaking (following day 1 cluster).

    I do think "polishing" is a good word for that, it is almost pretty, just needs a polish. The +/- 1 should be addressed in some be fashion. I still think a small handicap +/- VP is the simplest answer.

    Everything just about pretty except for cost $$$... That's the giant hole in the Titanic.



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
    romeo
  • GrendelGrendel Member Posts: 305
    As long as matchmaking is based on guild tier instead of the vp history of the players actually signed up, it will never consistently give competitive matches.
    romeo
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    Couldn't you cycle in a handful of PvP clearing level 15 accounts? @Grendel ?

    Their VP would be pretty low relative to the guild. And guilds could use this to force mismatch. Yes?



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
  • GrendelGrendel Member Posts: 305
    edited May 26
    ATLAS-Z said:

    Couldn't you cycle in a handful of PvP clearing level 15 accounts? @Grendel ?



    Their VP would be pretty low relative to the guild. And guilds could use this to force mismatch. Yes?

    No, because:
    1) Player level has nothing to do with the matching system I am proposing.
    2) A player with L27 survivors gets the same 20VP from clearing PvP maps that a player with L15 survivors does.

    The only way to try to manipulate the system would be to deliberately get fewer VP. But doing that is self-defeating because rankings are based on VP - not on wins.

    This reminds me of your earlier post where you talked about bringing your A-Game. Because the rankings are based on total VP (not on wins), I am surprised top guilds don't bring their A-Game to every battle. The extra 100-200vp (or more) per battle they may be able to get by really pushing every battle could make the difference at the very end.
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    edited May 26
    @Grendel you're missing the point.

    The cycled in troops would obviously have very low VP history yes? This could be done deliberately.

    Follow that progression in your system...



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    Let's say I made 10 Alt accounts. And just progressed them high enough to be able to participate in GWs...

    Then each match, cycle in a new one...

    A skilled player could easily handle the PvP even on this Junior alt account.

    Next match cycle in a new Alt.



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
  • GrendelGrendel Member Posts: 305
    ATLAS-Z said:

    @Grendel you're missing the point.



    The cycled in troops would obviously have very low VP history yes? This could be done deliberately.



    Follow that progression in your system...

    I don't understand how the cycled-in troops would have a low VP history? If the L15s have been playing battles and clearing PvP maps, they would have the same VP as L27s that have been playing battles and clearing PvP maps.

    If the matching uses average VP earned for the past X battles, then the only way to cycle in low-VP troops is to have those troops earn low VP scores in previous battles. So you would have to deliberately handicap battles so that certain players get low VP?

    If there is not enough VP history for a player, the matching system could use the average of the other team members.
  • ATLAS-ZATLAS-Z Member Posts: 4,015
    If you're using 1 Alt then the other 9 are getting zero VP because they're not participating.

    Then cycle them in and bam.



    Are you Lost? Alone? Looking for a killer team to have your back?
    Join ZOMBREX! We have a tiered guild structure so players of every level and ambition can find a home they fit in.
    Remember, search ZOMBREX SATISFIES. Check us out HERE
    EMAIL US --> [email protected]"
  • GrendelGrendel Member Posts: 305
    ATLAS-Z said:

    Let's say I made 10 Alt accounts. And just progressed them high enough to be able to participate in GWs...



    Then each match, cycle in a new one...



    A skilled player could easily handle the PvP even on this Junior alt account.



    Next match cycle in a new Alt.

    I get it. The system obviously couldn't treat new players has having a zero VP history. It could either not factor in that player at all in the VP average calculation or it could use the average VP history of the other players in the same battle. (Which, mathematically, should work out the same; so not really an "either".)
Sign In or Register to comment.