My Idea for GAC Matchmaking - Please critique

Prev1
I know there will never be a perfect algorithm, but looking at the forums, there are obviously many complaints about the current Top 80 system currently employed. More so than the original GA matchmaking (it seems to me)
So, here's my idea
Week 1: You get matched with opponents with similar character GP (wouldn't say no to "weight" given to a select few characters; Revans, Malak, etc)
Week 2: Same thing as week 1 EXCEPT your opponents finished the same rank as you (1st vs 1st, etc).

So on and so forth. You might roll through the first three GAC battles, but next round you'll be in a bracket with other players who also won all their battles. Difficulty will increase/decrease depending how you do.
(Not sure if this currently happens, I didn't experience this personally last GAC).

Open to any and all potential flaws being exposed and fixed. Thanks!
#CloneHelmets4Life...VICTORY!!!! :smiley: "I don't like sand. It's coarse and rough and irritating and it gets everywhere."

Replies

  • SemiGod
    1445 posts Member
    edited September 5
    I thought that was how it previously was lmao.

    In the first GAC all my matches were a breeze besides the last one where I had to go tryhard to get Kyber.

    And I thought i was just being matched with other champs lol
  • Kyno
    21355 posts Moderator
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.
  • Your first match is done by GP of your top X characters (depends on your division, but about 30-80 I think), modified just a bit by a couple other things and restricted by your division and your league...but everyone starts out in the same league, so that's not a factor. However, in every round after the 1st, you're matched with people who scored the same or similar points.

    This is not the same as matching you only with people who finished with the same exact rank as you in the last round, but since every victory gives you 1000 and every set defense gives you the same number of points as others in your division, it becomes very hard to lose 3 matches by going all out on offense and winning 1000 points in squad victories but never getting the match-victory bonus of 1000 points because the other person went all out on offense and finished with one more toon at full protection than you. It's also very hard to win all 3 of your matches without gaining any points from squad victories.

    Theoretically, someone who isn't playing at all and is in a round where none of their opponents are playing either can win all the tie-breakers and get 3k in victory points. Theoretically someone can get 3k in squad points and territory-clear points without ever getting any match-victory points.

    Theoretically, then, you could match up one player who flounced through the first round on tie breakers against one player who played hard but barely lost all 3 times.

    But there are FIVE rounds. You're talking about a truly edge case happening at the beginning of round2.

    So, the point system functions a lot like the system you're talking about. It's not exactly the same, but it has many of the same virtues. If you're playing hard and losing, you won't be matched up against people who were playing hard and winning in the next round. By the time round 3 hits, you should be getting pretty decent matchups.

    Now, I don't know how my opponents felt during last GAC, and I know that I won more than I lost (took 1st in rounds 3 and 5), but I can tell you that even when I was winning, none of my victories from round3 on were smashing successes. When I won, it's because they couldn't clear either zCarth or zBroodAlpha (or both) that I tucked away in the same territory in back. When I lost, I usually cleared at least 2 territories, most of my losses I cleared 3. The only times I cleared only 2 territories were in round 1 or maybe 2 (I can't remember for sure) where someone had Malak on defense and I couldn't touch them. Maybe every other match I would mess up a fight and have to use 2 squads to take out one defending squad. Nothing was automatic. Even if I won by a few hundred points, I played with genuine concern b/c I knew their roster and didn't know for sure what was in the back ranks on defense and what I would have to save to get through them.

    Creating a "good match" sorting system is hard, but by stringing together 5 rounds in a row and keeping track of ongoing score, I think they do something similar to what you're suggesting...and while a single event isn't a lot of data, it appeared to me to work well from round3 on.
  • Vice_torn
    544 posts Member
    edited September 5
    -
  • Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.
  • Love the idea, except I wouldn't put any weighting whatsoever on any characters, at all, ever.

    Let roster construction, in-match performance, and interest serve as the sorting function. I get the argument that "leagues currently do that", but they don't because the matchmaking is so opaque. Matchmaking is putting its finger on the scale to determine "rosters more like yours" and obfuscating the criteria being used to define your "peer" group. This fixes that.

    If you're going to go this far, why not just go the extra mile, get rid of the Week long GAC session and just have a five week long series of single matches. No "brackets" or "pods" or whatever you want to call them. Day 1 - set defense against opponent. Day 2 - fight opponent - Day 3, set defense against new opponent. Day 4 - fight second opponent. Lather, rinse, repeat for five weeks.

  • I think starting the GAC with divisions roughly as they are now and having matchmaking for the following rounds based purely on number of round wins would have made for an easier to implement and better system then they developed. However, I think they are too far along in development to change course now.
  • They need a 'handicap' score, each time you lose you get a -1% (cumulative) to your gp score and each time you win you get a +1% added to your gp score; so the more you win the better your opponents will be and if you keep losing you just keep fighting weaker and weaker opponents.
    A lot of People don't lose because they play bad, they lose because they don't have key toons maxed to G13.
    Looking for 1 member with 3.5m+ GP roster to help in GEO TB; we are 185m guild who are 70-6 in TW.
  • Matchmaking should be with people in the same league cause when someone in chromium league faces someone in carbonate it’s obvious pretty much who’s gonna win,Or match up with someone that have x7 more gear13 that the system calculate the same gp for top80 chars.In the other hand gear 13 shouldn’t be adding way more gp cause than the total gp of the top guilds should be ok to go 33* in sep Tb in the first month which we should all agree isn’t ok.One zeta is around 10 chars with gear13 if the char is also a pilot it adds way more.My opinion is that it should calculate all the roster which means all the decisions good or bad that every player made.
  • People need to understand that they've found a matchmaking Algorithm that benefits whaling out at the highest level for the highest amounts of money. They aren't going to change it.
  • Matchmaking should be with people in the same league cause when someone in chromium league faces someone in carbonate it’s obvious pretty much who’s gonna win
    Matchmaking is constrained by League, however there are a lot of players in each League.

  • Rath_Tarr wrote: »
    Matchmaking is constrained by League, however there are a lot of players in each League.

    Than how I faced a carbonate player in first round and won easily which I don’t like and a chromium in today’s round when I am bronzium?
  • Gth
    277 posts Member
    I real liked ga.

    I hate gac.

    Matchmaking is terrible and they like it that way. Unbelievable. I’m giving up on gac. I’m simply sick of it and sick of worrying about it. Game over man, game over.

  • Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    No, not really. There are 4 groups - not 8 (1st, 2nd-4th, 5th-7th and 8th).So 2 of the 8 either went 3-0 or 0-3. Then three went 2-1 and three went 1-2. That's his point - 75% of players are in those middle 2 groups.
    In game name: Lucas Gregory - - - -"Whale blah grump poooop." - Ouchie

    In game guild: TNR Uprising
    I beat the REAL T7 Yoda (not the nerfed one) and did so before mods were there to help
    *This space left intentionally blank*
  • Number of wins/loses lifetime. Case closed. No need for this strange arbitrary numbers of GP.
    As simple as that. And the most fair.
    Thing is, CG simply does not want to. Fairness do not lead to panic spending...
  • TVF
    15164 posts Member
    Vinniarth wrote: »
    Number of wins/loses lifetime. Case closed. No need for this strange arbitrary numbers of GP.
    As simple as that. And the most fair.
    Thing is, CG simply does not want to. Fairness do not lead to panic spending...

    So a 5m GP player can play a 500k GP player now. I'm sure that will go over well.

    Case reopened.
    TVF's guild is recruiting. Say hi in our Discord! https://discord.gg/AmStGTH
  • Liath
    4381 posts Member
    Rath_Tarr wrote: »
    Matchmaking is constrained by League, however there are a lot of players in each League.

    Than how I faced a carbonate player in first round and won easily which I don’t like and a chromium in today’s round when I am bronzium?

    Matchmaking is done at the start of each GA when you are placed in a group of 8. Leagues can and do change during the week, and will change differently for different people in the same group, but you were most likely in the same league at the time of matchmaking.
  • Rath_Tarr wrote: »
    Matchmaking is constrained by League, however there are a lot of players in each League.

    Than how I faced a carbonate player in first round and won easily which I don’t like and a chromium in today’s round when I am bronzium?
    Matchmaking occurs at the start of each GA within the GAC and yes it is now constrained by League so at the start of each GA every player in your 8-player bracket should be the same League.

    During the course of each GA, players.may get promoted so that by the second of third match of the GA, your opponent may indeed be in a different League.
  • So the OP notes that there is no weighting system. What if we add a weighing system based on the meta reports in arena? So if a person has malak then that malak will carry more weight then a KRU because malak is meta, same can be done with JKA teams, and other meta teams
  • It's working just fine. When you move up leagues your matches get harder. Nothing to change here. Go complain about something else somewhere else.
  • TVF
    15164 posts Member
    So the OP notes that there is no weighting system. What if we add a weighing system based on the meta reports in arena? So if a person has malak then that malak will carry more weight then a KRU because malak is meta, same can be done with JKA teams, and other meta teams

    Nope.
    TVF's guild is recruiting. Say hi in our Discord! https://discord.gg/AmStGTH
  • Kyno
    21355 posts Moderator
    edited September 6
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    Yeah sorry, I meant the number of divisions are too small, only 4 divisions.

    This>>>
    Nikoms565 wrote: »
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    No, not really. There are 4 groups - not 8 (1st, 2nd-4th, 5th-7th and 8th).So 2 of the 8 either went 3-0 or 0-3. Then three went 2-1 and three went 1-2. That's his point - 75% of players are in those middle 2 groups.

  • TVF wrote: »
    Vinniarth wrote: »
    Number of wins/loses lifetime. Case closed. No need for this strange arbitrary numbers of GP.
    As simple as that. And the most fair.
    Thing is, CG simply does not want to. Fairness do not lead to panic spending...

    So a 5m GP player can play a 500k GP player now. I'm sure that will go over well.

    Case reopened.

    Never seen this system before? Yes, first time it could be as bad as it is now, but after several matches everything become very well balanced.
    No need to reinvent bicycle.
  • CaptainRex
    2753 posts Member
    edited September 7
    Kyno wrote: »
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    Yeah sorry, I meant the number of divisions are too small, only 4 divisions.

    This>>>
    Nikoms565 wrote: »
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    No, not really. There are 4 groups - not 8 (1st, 2nd-4th, 5th-7th and 8th).So 2 of the 8 either went 3-0 or 0-3. Then three went 2-1 and three went 1-2. That's his point - 75% of players are in those middle 2 groups.

    True, but the number of divisions would increase each week. By the 4th week, You could be facing someone who finished 2-4, last, then first, like yourself (hypothetically, just throwing out random placements).
    #CloneHelmets4Life...VICTORY!!!! :smiley: "I don't like sand. It's coarse and rough and irritating and it gets everywhere."
  • CaptainRex wrote: »
    Kyno wrote: »
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    Yeah sorry, I meant the number of divisions are too small, only 4 divisions.

    This>>>
    Nikoms565 wrote: »
    Kyno wrote: »
    I think the groups are too small and the pool of players is so large that in the end this wouldnt matter for the majority of players.

    I think it would all average out too quickly.

    This may help the top and bottom edge cases as they would get pulled from the main group, but the "middle group " would just blur together, too many players in the middle 2 groups of the 4 group system.

    This is similar to the league problem, they are too large and not enough of them so they dont create any stratification.

    The groups are not too small, your thinking is incorrect.

    Imagine 64 players, that's 8 first 8 second 8 third and so on week 1

    Week 2 there's still 64 people but all first places match vs first places, 2nd vs 2nds, and so on.

    No, not really. There are 4 groups - not 8 (1st, 2nd-4th, 5th-7th and 8th).So 2 of the 8 either went 3-0 or 0-3. Then three went 2-1 and three went 1-2. That's his point - 75% of players are in those middle 2 groups.

    True, but the number of divisions would increase each week. By the 4th week, You could be facing someone who finished 2-4, last, then first, like yourself (hypothetically, just throwing out random placements).

    First it would find someone who has finished with the same previous ranks as yourself, then it would group by character GP, as close as possible, then maybe current GAC score or something if they want other metrics beyond that.
    #CloneHelmets4Life...VICTORY!!!! :smiley: "I don't like sand. It's coarse and rough and irritating and it gets everywhere."
  • Kyno
    21355 posts Moderator
    I get what you are saying, and I agree we should have more round(action)based groupings, to help create more sub groups for matching during the whole event.

    But with such a large portion of the player base being caught in the 2 middle groups, generally speaking this will average out and have a large portion of the player base all in the same group.
    I think this could also lead to problems with the numbers in a group, you are limiting the numbers that can be placed in the very special groups, like the 2x W group for the last match.

    Just a thought, not sure how big of a problem it would actuall be without knowing the number of players and groups per round, ext...

  • They should match people based off of Jawas team GP.
  • My problem with all the matchmaking complaints is everyone wants a "fair" match where they have a 50-50 shot at winning. Meanwhile those who have worked to create a pvp focused roster have done so to try to create an advantage so they can win more than 50-50. So how do you create a system that strips the ability to create an advantage thru modding and careful gear selection? Would you want to? If every match was a sweat-fest I think many would lose their drive to play and those in the middle would lose interest because they all have a 50-50 shot.

    Already I'm noticing some loss of interest in GAC with a lot more players just setting heavy defense and letting their opponent bash themselves silly while they do a few fights and maybe win on banners. I think we need more content overall so GAC isnt having to hold so much of the community focus.
  • Daishi wrote: »
    My problem with all the matchmaking complaints is everyone wants a "fair" match where they have a 50-50 shot at winning. Meanwhile those who have worked to create a pvp focused roster have done so to try to create an advantage so they can win more than 50-50. So how do you create a system that strips the ability to create an advantage thru modding and careful gear selection? Would you want to? If every match was a sweat-fest I think many would lose their drive to play and those in the middle would lose interest because they all have a 50-50 shot.

    Already I'm noticing some loss of interest in GAC with a lot more players just setting heavy defense and letting their opponent bash themselves silly while they do a few fights and maybe win on banners. I think we need more content overall so GAC isnt having to hold so much of the community focus.

    Here's what happened, as I see it (let me know if I'm wrong).
    GA came out, a majority of the playerbase enjoyed it, so CG went ahead with GAC. However, as they mentioned, there weren't as many close matchups as they wanted, so they changed the algorithm to what we have now.
    Which seems off to me, trying to force equal results.
    #CloneHelmets4Life...VICTORY!!!! :smiley: "I don't like sand. It's coarse and rough and irritating and it gets everywhere."
  • Omeah
    932 posts Member
    edited September 10
    Why punish people for working to do their best?
Sign In or Register to comment.