20 replies to this topic

[nullBot]

Well the title says it all, I just played around 10 matches on team deathmatch and we either lost completely horrible, or we won big time, like 40-5 was the score.

 

I cannot suggest much since I'm not sure how your balancing works, but I'd suggest a way to rate player skills based on team fight, mobility and experience playing the game, a vet player shouldn't fall in a match with only noobs for example, as I've seen this several times, or whole noob teams against experienced players.

[Trigary]

There's a balance, but as players join, leave, it is breaking, I think. Before the round starts, you are seperated into two teams, by your MMR number. I think the match would stay balanced, if noone would leave or join. If a good player leaves and a bad player joins or if a bad leaves and a good joins and it's repeated, it breaks the balance, in my opinion. Of course, I can imagine players who aren't seperated well, but I don't think that that's the matter.

[opicr0n]

On high mmr servers the balancing could be a bit more random too. I just finished a streak of maps where iDim and myself were always set against n3on and Greedy. It would be nice to swap out some of those combinations every now and then.

 

It was pretty balanced but I think it would be too if iDim and n3on were joined and me and Greedy for example.

[CrimsonKaim]

On high mmr servers the balancing could be a bit more random too. I just finished a streak of maps where iDim and myself were always set against n3on and Greedy. It would be nice to swap out some of those combinations every now and then.

 

It was pretty balanced but I think it would be too if iDim and n3on were joined and me and Greedy for example.

 

Just like I never will get with Rozer on the same team. And if we try to party up, we won't even find any single match

[OmegaNull]

I suggest a shuffle feature after each match. Not sure how it would work, but make it so the same people are not always on the same team. Ultimiately comes down to how the system uses MMR to split players into teams. MMR is not a perfect way to judge player skill. 

[TheFrostnessMonster]

I suggest a shuffle feature after each match. Not sure how it would work, but make it so the same people are not always on the same team. Ultimiately comes down to how the system uses MMR to split players into teams. MMR is not a perfect way to judge player skill. 

A shuffle would be a good thing to put in.  Players could vote for a shuffle, and when a majority is reached, it shuffles the teams.  But it could make it worse, but that's why this is in the Suggestions and Ideas sub-forum

[Trigary]

I forgot, the parties break balance, too.

[Gookywun]

I don't really know what you propose to fix something like this.

 

You're wanting complete balance in your pub games which is balanced around the current MMR system.

 

Arguably, the MMR system is completely broken at the moment. In the last 12 months that the game has been inactive the handful of players that grinded this game got incredibly high MMR's by just playing Deathmatches.

In a game where you're balancing public games on MMR where you have <100 players each region is going to be unbalanced nearly 99% of the time unless you made a organised game. Also this game doesn't punish players heavily enough for leaving games at any time they want, leaving people to drop in and out constantly.

 

OPINION (SPECIAL WORD HERE)

There's honestly no way of fixing this IMO unless you restart the MMR system completely, change the way it works and have a huge wave of new people join the game to help balance overall.

[nokari]

Have we not discussed the causes and opinions on what could be done to fix them to death already 1000 times?

[Sylhiri]

Have we not discussed the causes and opinions on what could be done to fix them to death already 1000 times?

 

New devs and new forums. It must be done.

[Draigun]

The autobalance system isn't exactly what's the issue here. The problem lies within the amount of players available, and the effect of leaving and joining mid-game.

 

OmegaNull pointed out that measurable MMR isn't the golden number to determine the skill of the pilot. In retrospect, MMR is (still?) calculated based on match performance. Defter, developer from the now defunct Adhesive, mentioned the way MMR was calculated per match. In essence, the formula was:

• If personal match score exceeded at least 50% of the enemy's (and/or team?), increase MMR
• If personal match score falls behind at least 50% of the enemy's (and/or team?), decrease MMR
• Comparison of server average MMR to personal, apply *insert MMR increase/decrease equation here*?

Feel free to correct me, as I'm not so sure if these details align with what Defter had mentioned. Also, I'm not so sure what the system looks for when in fact decreasing or increasing MMR. Is it team performance averaged out, or is it the comparison of each individual MMR value to your own?

 

The act of leaving and joining mid-game, usually always degrades the match quality, since it doesn't take this into account. Obviously, there are slim chances of it still being perfect, whether or not people leave the game and or join.

 

The other problem that we face in HAWKEN, is the low player base. I'm sure EM10 mentioned this before, but what it boils down to is: a larger player base increases the effectiveness of the autobalance system. In contrast, as what one would presumably guess, a lower player base decreases the effectiveness of the autobalance system. It's simple math and statistics. The larger player base allows a more accurate match-making system, since the chance of people leaving mid-game decreases significantly; the chance of finding a quality match increases significantly. It was probably designed with this fact in mind. Believe it or not, early to mid February 2014, there were almost 10,000 HAWKEN pilots in the game.

Edited by Draigun, 21 March 2015 - 03:22 PM.

[Amidatelion]

K/D apparently factors in heavily, which is the only thing that makes sense for how MMR gains and losses work these days. Especially when I'm in a match with ThirdEye and them, outscoring the entire enemy team, but because I'm 6/10/22, I lose MMR.

[Grollourdo]

Damn talking about all this technical stuff ... Why not just be able to switch teams like some kind of apply button to switch teams if another opposite team player presses the button too XD if this makes sense ...


This might not solve the problem but it might help ... XD

[crockrocket]

Well matchmaking will get a lot better as the playerbase increases. Some incentive to not change lobbies after each match could help a lot as well.

 

 

K/D apparently factors in heavily, which is the only thing that makes sense for how MMR gains and losses work these days. Especially when I'm in a match with ThirdEye and them, outscoring the entire enemy team, but because I'm 6/10/22, I lose MMR.

 

If I did that in the same match my mmr would almost certainly go up, or at the very least remain essentially unchanged. So I think that there is definetly a component that hinges on the mmr of the people you're facing compared to your own mmr. If I do well as a 1900 mmr player in a 2300 mmr server, I gain more than a 2400 mmr player in the same server. I also have less to lose, if I get stomped by people that are 500+ mmr higher than me I generally lose very little mmr. That's what I've noticed anyway.

[MomOw]

I find boring to let new player come when a 4 vs 4 (or more) game is balanced, and I'd like a "room" to wait for TDM to come up and a system that mix players between sever to create balanced games. Maybe add some botsTDM to have some fun while queuing for a real TDM server.

[nullBot]

The autobalance system isn't exactly what's the issue here. The problem lies within the amount of players available, and the effect of leaving and joining mid-game.

 

OmegaNull pointed out that measurable MMR isn't the golden number to determine the skill of the pilot. In retrospect, MMR is (still?) calculated based on match performance. Defter, developer from the now defunct Adhesive, mentioned the way MMR was calculated per match. In essence, the formula was:

• If personal match score exceeded at least 50% of the enemy's (and/or team?), increase MMR
• If personal match score falls behind at least 50% of the enemy's (and/or team?), decrease MMR
• Comparison of server average MMR to personal, apply *insert MMR increase/decrease equation here*?

Feel free to correct me, as I'm not so sure if these details align with what Defter had mentioned. Also, I'm not so sure what the system looks for when in fact decreasing or increasing MMR. Is it team performance averaged out, or is it the comparison of each individual MMR value to your own?

 

The act of leaving and joining mid-game, usually always degrades the match quality, since it doesn't take this into account. Obviously, there are slim chances of it still being perfect, whether or not people leave the game and or join.

 

The other problem that we face in HAWKEN, is the low player base. I'm sure EM10 mentioned this before, but what it boils down to is: a larger player base increases the effectiveness of the autobalance system. In contrast, as what one would presumably guess, a lower player base decreases the effectiveness of the autobalance system. It's simple math and statistics. The larger player base allows a more accurate match-making system, since the chance of people leaving mid-game decreases significantly; the chance of finding a quality match increases significantly. It was probably designed with this fact in mind. Believe it or not, early to mid February 2014, there were almost 10,000 HAWKEN pilots in the game.

 

I agree, a larger player base could make the game more playable. I also think that there has to be some kind of campaign mode to let new players learn how to play, the tutorial helps only to learn the controls, but fighting those bots is no help against 5 hardcore elite players, team fight and understanding your mech's role are important things to learn for newcomers.

Edited by nullBot, 26 March 2015 - 01:35 AM.

[LaurenEmily]

Is it just me or does the mmr system seem a little broken to anyone else ? i don't know how it works but sometimes it seems that luck is too big of a factor in how it's calculated. For example you can be placed in good teams and that way dominate players that are in reality much better than you but constantly die just because their team is failing. Also it works the other way too, someone who wouldn't last 10 seconds in a straight-up 1v1 with you can end up fragging all the kills and wrecking your rating, with the help of a good team. I personally have a rating of around 2100 but i'm starting to suspect this is way higher than my actual skill level is, and in a balanced match i don't do that well.

[Anichkov3]

A few words about matchmaking. It was suggested earlier repeatedly by different people (including me).

 

MMR system is quite accurate thing. The problem is that only a small number of players. Just do not forget that a team game, and a lot depends on the coordinated team play. I played matches where our team was much stronger players (or vice versa in the team of the enemy), and we lost. All due to the fact that people do not play as a team. 

[opicr0n]

In Enemy Territory the previous scores on maps were used to divide the teams in the next map. If I remember correctly.

 

This will, after a map or two, balance things out quite nicely based on points thoughout the previous matches.

 

The good point is that the balance is created as a snapshot of the last match. Not an vague MMR count.

[LaurenEmily]

In Enemy Territory the previous scores on maps were used to divide the teams in the next map. If I remember correctly.
 
This will, after a map or two, balance things out quite nicely based on points thoughout the previous matches.
 
The good point is that the balance is created as a snapshot of the last match. Not an vague MMR count.

This sounds pretty good except hawken might be too 'small' for it to work at the moment. Some people's performance varies a lot from map to map (i'm looking at open maps with hellfires and sharpshooters) and so a player who was dominating with a rocketeer in an open map might get high rank despite being garbage in more closed maps, messing up the balance.
I hope this made sense.

[opicr0n]

This sounds pretty good except hawken might be too 'small' for it to work at the moment. Some people's performance varies a lot from map to map (i'm looking at open maps with hellfires and sharpshooters) and so a player who was dominating with a rocketeer in an open map might get high rank despite being garbage in more closed maps, messing up the balance.
I hope this made sense.

 

It does. And there is no perfect solution.