Loading...
  OR  Zero-K Name:    Password:   

Team matchmaking improvement suggestions

38 posts, 2370 views
Post comment
Filter:    Player:  
Page of 2 (38 records)
sort
In my opinion, having played several matches, the team matchmaking is in dire need of improvement. Or at least a big red warning to new players saying "do not click, lol".

Reasoning:

1. It puts premades vs random queuers (even when there is a huge elo difference).
2. It puts players with a huge elo difference on the same team, yet still shares mass equally. So 50% of your eco is going to some guy's turret swarm or bertha at 5 min.

For example: http://zero-k.info/Battles/Detail/559238 . Me + random vs premade of two players both much higher elo and level than me. Random player makes 4 stardusts to begin, then leaves. Not a good experience at all.

Suggestion:

1. Ban premade, or make it clear and allow people to dodge. The community is too small and this is just a tool to stomp people.
2. Distribute mass by 1v1 elo predict%. So for example if a blue player is with a yellow one, yellow player gets whatever his win% would be in a 1v1 (or something similar), so 20% or whatever the implied probability is.


mod-edit: fix broken link
+1 / -0

5 years ago
I understand the general issue here. However some people would not even touch the match making pvp without their friend they came to the game with or their buddy they met in some battle. why? well it comes down to Comfort, Cooperation and most of all Trust. Yea sometimes its best to know you have someone with you thats not going to resign when 1 of his wind generators gets got.

However that doesnt stop those players who only care abut that sweet rank to piggy back on the backs of players whom have worked to get the awesome color they have. Which mostly leads to lobsters getting over cooked when young and not playing the game or pvp anymore cuz of the horrible taste that leaves in your mouth.

However Matchmaking having too much or too little freedom is bad in both ways.perhaps stop the awesome colors from queuing with the not so awesome colors to make them play, learn and work for it. Balance is a funny thing to achieve sometimes.
+1 / -0

5 years ago
I have remarked on the second issue in another thread. More recent experience reinforces my impression that the underlying problem is where a team with a high rated players gets drawn against the general pool. Some more recent games in my personal experience:

Multiplayer B558937 4 on Adansonia v4.1
Multiplayer B558767 4 on Onyx Cauldron 1.9

Both lopsided battle from the outset resulting in curb stompings, and my frustration at being drawn against the same higher rated team several times in one day was apparent.

However, when the matchmaker has a higher rated person to put in the pool, the game unsurprisingly becomes properly competitive ( Multiplayer B558950 4 on Shimmershore v1.0 ). If the team's average rating is higher than the matchmaking pool's average, fun stops happening for everyone else. While noting Odlaner's remarks, I still think that removing the ability of premade teams to join the matchmaker queue is one situation where the needs of the many (the general matchmaker pool) outweigh the needs of the few (that subset of players who absolutely insist on playing with their buddy).

The underlying problem remains the small player base. If you're not playing the same people over and over then there is less of an issue (being cooked by a high rated team isn't too big a problem if it only happens once in a session instead of 3 or 4 times), but I do believe allowing premade teams into matchmaker has an overall negative effect with the current game population.

With regards to team quality, ultimately that is the risk you take playing teams. Sometimes you'll have proper team players (unrated novice players can still be good team players if you engage with them, the game provides ample means to do so), sometimes you don't. If you don't want to run the risk of having a weak team-mate, play 1v1 (though I find the 1v1 pool feels like it's usually higher average rated than team pool, no doubt due to higher confidence of the players there).
+1 / -0


5 years ago
We can set a search range width multiplier for battles that include parties. I've reduced it to 65% from 80%. This is a dynamic configuration so the change is live instantly.
+1 / -0
5 years ago
Thanks for that change, the team games I had recently have been generally better (aside from trolling etc, not much to be done about that tho).
+0 / -0

5 years ago
I think the current system of team matchmaking does not work properly. In fact, we need to wait a long time to have a game and not because there are no players but to find those with the same Elo. The excellent system devised by Dein Stein unintentionally creates situations of disparity in the matchmaking team.
The problem of the evaluation of teams according to the system 95% of the games are team 2 vs 2 and this is not due to lack of players but due to the fragmentation in micro games with players of the same elo. In the evaluation of the formation of the teams, the sum of the players involved is the sum,
so in team 1 players A and B have for example an elo 1500 + 1500, while team 2 players C and D have an elo of 2000 + 1000.
Is not the same thing. I lost several games against 2 good players because my partner started the game by ddm or by defending himself or building bertha.
Therefore it would be more natural to make the evaluation of the Elo more elastic because:
1) there would most likely be team matches 3 vs 3 or 4 vs 4, much more fun
2) you would avoid playing room games of 16 vs 16 players, who suffer terribly from lag (it seems sometimes to play in replay mode)
3) the nerf player who wants to face a matchmaking team match, (as it already does ruining the game of others), would be balanced
by the presence of more players in the same team
4) also do not understand why there must be protection for the weakest players, to the detriment of others. In the case of a tournament, everyone plays against everyone therefore those who participate do so at their own risk. If the nerf wants to play 1vs1 with a friend he can easily do it in the room.
+2 / -0
5 years ago
The balance has been alot better with 65%. It prob could be reduced to 60%. More then a 100 elo diff in teams starts getting unbalanced.
+0 / -0
After a bit of reflexion on the subject, I feel like premades on (ranked) matchmaking should only face other premades.

Many premades I've faced have been very good players teaming themselves for what ends up to be an easy win against average players and to be honest I wonder if it's not related to the rank insecurity "issue" stated here: https://zero-k.info/Forum/Thread/25690 , after all you wont lose your rank if once in a while you stack the odds in your favor.

It kinda breaks the matchmaker: I don't think I'm mistaken if I say its point is to have random balanced games. And Instead with this we have situations where it would have mixed the players for a fair and balanced game but we end-up with too much unbalancing which might be frustrating for some.

It's fine for people to play with their friends, it really is. But don't forget you can still host regular games without having to rely on the matchmaker for that and this way people know what they are facing.
And if you really want the thrills of ranked matchmaking with your buddy, let it be that it's only against another team of premades because to be honest even if we achieve ELO balance, there is always the advantage of knowing how your friend plays and reacts for a much better coordination (and I'm not even speaking of mic issues).

EDIT: maybe we can find some kind of compromise where the party is actually a wish: the matchmaker will take it into account that you want to play with your friend but it will break it and mix the teams if the odds of winning for one team is greater than 60%? This way everyone should be pleased?

My 2 cents, thank you for reading.
+1 / -0
5 years ago
There has been quite alot of comments in general agreeing with my OP.

However right now for most of the day any "Teams available" game is randoms vs a premade of 2k+ players. I'm pretty sure that other people don't like this (except the 4-5 same people who are involved).

Team games with <20% chance to win are horribly imbalanced imo because they are always something like this:

Bronze + Gold vs Gold + Blue. If you separate this it is Gold v Gold (~50% chance to win) and Bronze v Blue (~0% chance to win). The real win percent here is about 0% for the Bronze + Gold team, but the algorithm thinks its the average. Its more like the minimum (because in the above example Blue v Bronze will be over very very quickly and the game will turn into Gold v 2xeco Gold + Blue, which is unwinnable).

I would propose to either ban premades altogether (as really this is exploiting the above argument for free wins) or narrow the allowed range alot more.
+1 / -0

5 years ago
I agree that something needs to be done because it comes to a point where I stop queueing matchmaking if I see the same (highly) unbalanced premade several times a row, even if I still want to play.
Last round I tried, the team I was in decided to resign even before the game started.

Loving the game, I keep coming back but I can only imagine what it does to many newbies.
Having played a lot of dwarf fortress, I fully embrace the old saying that losing is fun -but- getting crushed so hard you don't learn a thing and don't make any progress is certainly not, especially if you know it could have been mixed and balanced.

I need to stress out that this also works the opposite way: I played 2 rounds in a row today where the opposite team was the same premade and they were no match at all (understatement). I don't think that was fun for them, it certainly wasn't for me from the total absence of challenge (in that way I guess I'm different from many top-players! <- easy troll, sorry :P).
+1 / -0


5 years ago
Can you post some links? Try to post whether one or both of the teams were are party. I ask because there may be technical issues. Perhaps the width multiplier for parties does not work. Perhaps the balancer is broken and parties are not actually involved.
+0 / -0

5 years ago
If you're looking for lopsided fights involving teams, I posted a couple earlier. In all examples, I believe the other side were a premade team and can't supply any personal examples which I know had teams on each side.

More recently:

Multiplayer B564345 6 on Fairyland 1.31

Of course, when you do get a high rated player in the general pool, things are better.

Multiplayer B564307 6 on Calamity 1.1
Multiplayer B562399 6 on Titan v3.1

When the team is in line with the pool, the game is pretty balanced (and in fact I'm not sure teaming up by itself offers more than marginal advantages).

Multiplayer B563269 4 on Comet Catcher Redux v3.1 (my ally and I conceded early losses that we couldn't come back from against similarly rated players exactly as could have happened if we were each 1v1ing)

Sometimes of course, the team isn't that formidable ( Multiplayer B563485 4 on AlienDesert was very hard work, but I was able to grind out a victory 1v2) but as I said earlier, the problem happens where the team's average rating is higher than the pool- which one would expect. Higher rated players generally beat lower rated players, and so do teams. Team games just mean 2-4 people get upset at once about being stomped instead of 1.

I think the matchmaker works well without premades in the pool.

For me, getting good enough at the game to turn the tables on high rated teams is a spur to improve my own game but like FRrankEarlGrey, there are limits to my endurance for being pummelled.
+0 / -0


5 years ago
I'll try further reducing the width multiplier for games involving parties.
+0 / -0
5 years ago
If you still need more examples:

http://zero-k.info/Battles/Detail/564813 -- probably the best recent example I have. My ally lost his fac almost instantly, never recovered or expanded. This one was vs a classic teamstack premade. Please note that in these extremely imbalanced games generally speaking both 2k+ players will do some kind of early game cheeze and try to just eliminate one player instantly thus guaranteeing them victory.

http://zero-k.info/Battles/Detail/563183 -- another game vs the same stack, lost entire map basically instantly. My random ally was also quite upset with this game.

http://zero-k.info/Battles/Detail/561721 -- Won this one, shows how nonlinear it is between 15% chance to win (which I conjecture is zero actually) and 30% chance to win (which is probably accurate). If you don't want to watch the reason we won was basically Astran is similar skill to me and knows also that they will do some kind of cheeze. We helped each other to not die at start, from then it was a pretty easy game.

http://zero-k.info/Battles/Detail/564795 -- paired with 1200 ally as ~1900. Had to permanently devote most of my units to defending him. Didn't win despite bombing an enemy singularity reactor and destroying half their eco.

I have more replays but they are much older -- I havent played team MM much recently.

+1 / -0
I am a little worried that the "rush" for making games "fair" will just result in no ability to play the game at all, during some less-than-heavy hours - or to wait for 10+ minutes, despite having enough people waiting (had it, more often than not).

I, for one, prefer unbalanced game (and yes, it is mostly unbalanced NOT in my favor) than no game at all, during the time window I can actually play. It is kind of stupid that in scenario of party + few random people (but not enough to pair randoms vs themselves), BOTH party and the random people can't play - just because someone (tip: OP) takes losing to more skilled opponents very hard.

---
+4 / -1
Adjusting the matching width in response to current online player count may help a bit on both issues of too long wait time and too unbalanced games.
+6 / -0

5 years ago
quote:
I am a little worried that the "rush" for making games "fair" will just result in no ability to play the game at all, during some less-than-heavy hours - or to wait for 10+ minutes, despite having enough people waiting (had it, more often than not).

I, for one, prefer unbalanced game (and yes, it is mostly unbalanced NOT in my favor) than no game at all, during the time window I can actually play. It is kind of stupid that in scenario of party + few random people (but not enough to pair randoms vs themselves), BOTH party and the random people can't play - just because someone (tip: OP) takes losing to more skilled opponents very hard.


That's why I think tweaking the range is only at best a partial solution and removing premades from the pool would be better for the matchmaker. Without premades at all, there is little chance of a 2K+ team being drawn against a pool average team (and why should the lone high rated player not teamed up have to try to carry a lobster against two or more players at least as good as he is?) Remove premades, and the 2K+ players will probably get split between the teams, raising the quality of play on both sides, creating more competitive games. Alternatively, they were just in it to troll everyone else and stop playing matchmaker which wouldn't be a bad thing if it meant more people stuck with it.

You can still play together in custom matches and I am unconvinced despite your protestations that you or most of the playerbase would join a custom game knowing for a fact you would face high rated players. I know I wouldn't unless I knew the other guy on my team was at least comparably skilled to the best on the other side.
+3 / -0
quote:
I am a little worried that the "rush" for making games "fair" will just result in no ability to play the game at all, during some less-than-heavy hours - or to wait for 10+ minutes, despite having enough people waiting (had it, more often than not).

I, for one, prefer unbalanced game (and yes, it is mostly unbalanced NOT in my favor) than no game at all, during the time window I can actually play. It is kind of stupid that in scenario of party + few random people (but not enough to pair randoms vs themselves), BOTH party and the random people can't play - just because someone (tip: OP) takes losing to more skilled opponents very hard.

Except the point of the matchmaking is to have balanced games. It's not like matchmaking is the only way of playing multiplayers matches: no one stops people from playing regular casual games with traditional hosting. If I see that matchmaking is taking too much time, I look for regular hosted games and it's fine (unless there truly is no one around but then there is nothing to be done). The issue is, in many cases matchmaking could have been balanced but it is not because one side wanted to brute-stack for an easy win.
And with this, GBrankehtomlol is right with:

quote:
Please note that in these extremely imbalanced games generally speaking both 2k+ players will do some kind of early game cheeze and try to just eliminate one player instantly thus guaranteeing them victory.

which might substantiate the idea of this issue linked to "rank insecurity" in some way.

If more examples are needed:
Multiplayer B564837 6 on Onyx Cauldron 1.9
Right after this one, we faced the exact same team again and mine decided to resign even before starting the game.

More examples of another traditional stack that plays mostly together:

Multiplayer B564328 6 on Comet Catcher Redux v3.1
Multiplayer B564369 6 on Eye of Horus v13
After facing this two times in a row, I stopped queueing matchmaking for a bit because I knew it was bound to happen again, and I wasn't wrong because this match followed:
Multiplayer B564345 6 on Fairyland 1.31

More of this stack:
Multiplayer B565109 4 on Akilon Wastelands ZK v1
Multiplayer B565057 4 on Wanderlust v03
Multiplayer B564285 6 on Red Comet v1.3
+1 / -0

5 years ago
I agree with FRrankEarlGrey . But I point out another element that comes out from what FRrankEarlGrey said: the current system for team matchmaking is circumvented when a clan presents itself.
The clan will be placed in the same team and as the example reported by FRrankEarlGrey on 7 matches mentioned 6 the team players won them, in this case the Vox Populi. This is not correct. In fact the players of the same Clan play in a much more synergistic way than randomly matched players in the same team, especially for the harmony and training to play together.
+0 / -0


5 years ago
Clans are weaker than parties.

While there's some extra length the system is prepared to go to keep the clan together, it will also split it without too much issue if the gap is still too wide.

Parties are entirely unbreakable.
+0 / -0
Page of 2 (38 records)