161 replies to this topic

[Wil McCullough]

FRAGTAST1C, on 14 March 2019 - 06:56 PM, said:

Just got back into MWO last night for the events and played about 11 QP matches. I noticed that people in Units were more in number on one side than the other. Now, before you say, "NO!! THAT DOESN'T MEAN THEY WERE ANY GOOD", I agree and I don't mean to claim anything by it but the thing is, I won a few matches out of those 11 and they were all when my team had more people in Units and vice versa.

What's your point though? People in units can't drop in qp at the same time?

[FRAGTAST1C]

My point is that people in Units have more experience in MWO than those who aren't (generally speaking as I know people choose not to join Units as well) and the MM clearly doesn't work i.e., people with experience tend to perform better at MWO and MM's current mode of putting players in teams is nonsensical, which is what is being said.

[Wil McCullough]

FRAGTAST1C, on 14 March 2019 - 09:34 PM, said:

My point is that people in Units have more experience in MWO than those who aren't (generally speaking as I know people choose not to join Units as well) and the MM clearly doesn't work i.e., people with experience tend to perform better at MWO and MM's current mode of putting players in teams is nonsensical, which is what is being said.

You can't judge whether the mm is working based on the distribution of unit tags in your qp game. Wtf.

If i enter a game and see 3 )mm( members on one side and one on the other, it's quitw a stretch to come to the conclusion that mm.is borked. Mm weighs teams based on pilot tier, not unit tag.

[PhoenixFire55]

UMPA Viper, on 14 March 2019 - 01:38 PM, said:

You mean you guys dont find it fun to either win or lose by matches of 12-1 or 12-0 in under 5 mins ??? Thats lunacy ! Matchmaker is so horrible for group drops anymore that its pathetic. There also seems to be a mechanic in group drops of damage given versus damage received based on tier which doesnt seem to be in place in pug drops. My friends and i tested this theory and it seems to be true.

Anyway , doesnt seem to matter , PGI wont do anything about it. If ya want better drops with your friends , theres always the ole sync drop in pug play. Much better matches.

Coming from a guy that usually is the last one alive and usually the one with least damage on his team at the same time I think I kinda know the reason why group queue matchmaking is so horrible for you.

FRAGTAST1C, on 14 March 2019 - 09:34 PM, said:

My point is that people in Units have more experience in MWO than those who aren't (generally speaking as I know people choose not to join Units as well) and the MM clearly doesn't work i.e., people with experience tend to perform better at MWO and MM's current mode of putting players in teams is nonsensical, which is what is being said.

The only difference between a player with a unit tag and a player without a unit tag is that the latter has by far more chances to happen to be a complete newb. Thats why it seems that people with tags are more 'experienced' overall.

[FRAGTAST1C]

Wil McCullough, on 14 March 2019 - 10:05 PM, said:

You can't judge whether the mm is working based on the distribution of unit tags in your qp game. Wtf.

If i enter a game and see 3 )mm( members on one side and one on the other, it's quitw a stretch to come to the conclusion that mm.is borked. Mm weighs teams based on pilot tier, not unit tag.

That is literally not what I meant but the wording was poor that it came by that way. What I was trying to say was that MM will put a team together based on Tier system and it is quite possible that people in Units have been playing this game for longer than those without (not always the case as I mentioned in my previous post that people choose to not join a Unit) and MM is bad enough to simply rely on Tier system as the only means of splitting the team and grouping people up.

PhoenixFire55, on 15 March 2019 - 12:21 AM, said:

The only difference between a player with a unit tag and a player without a unit tag is that the latter has by far more chances to happen to be a complete newb. Thats why it seems that people with tags are more 'experienced' overall.

Yes, I know. But the fact remains that "experienced" players often end up on one team despite being the same Tier as a total newb in Tier 3 who won't torso twist.

[Shadowomega1]

Currently MM doesn't balance the team, it just grabs that are in the waiting list within the preset tier values. T1 should only be matched with T1 and T2, T2 can be matched up with T1, T2, and T3, T3 can be matched up with T2, T3, and T4, T4 can be matched up with T3, and T5, T5 can be matched up with T4; as of the last change back in september 2018. However, with less players the matchmaker likely opens this up by 1 Tier and maybe by 2 tiers at a lower player count. What makes matters worse is that there are some pilots in T1 that barely qualify as T1 pilots with an over all match score of 150 while those with more skill are 300 or more average match score.

To sum it up MM does nothing to balance the teams, it just randomly tosses people into a match based on tier; which leads to the current state of the solo queue. If there is any attempt of balancing it is based on tier not on an individual's skill level.

[justcallme A S H]

Shadowomega1, on 15 March 2019 - 03:53 AM, said:

Currently MM doesn't balance the team, it just grabs that are in the waiting list within the preset tier values. T1 should only be matched with T1 and T2, T2 can be matched up with T1, T2, and T3, T3 can be matched up with T2, T3, and T4, T4 can be matched up with T3, and T5, T5 can be matched up with T4; as of the last change back in september 2018. However, with less players the matchmaker likely opens this up by 1 Tier and maybe by 2 tiers at a lower player count. What makes matters worse is that there are some pilots in T1 that barely qualify as T1 pilots with an over all match score of 150 while those with more skill are 300 or more average match score.

To sum it up MM does nothing to balance the teams, it just randomly tosses people into a match based on tier; which leads to the current state of the solo queue. If there is any attempt of balancing it is based on tier not on an individual's skill level.

Balancing teams has nothing to do with matching tiers.

The two are totally different discussions.

[Curccu]

Wil McCullough, on 14 March 2019 - 10:05 PM, said:

You can't judge whether the mm is working based on the distribution of unit tags in your qp game. Wtf.

If i enter a game and see 3 )mm( members on one side and one on the other, it's quitw a stretch to come to the conclusion that mm.is borked. Mm weighs teams based on pilot tier, not unit tag.

Had a solo game Yesterday with 3 or 4 (can't remember was it 3 or 4) people from same unit in opposite team and one in mine, we won that game 12-2

edit: read what you have written before clicking send...

Edited by Curccu, 15 March 2019 - 05:09 AM.

[Shadowomega1]

justcallme A S H, on 15 March 2019 - 04:30 AM, said:

Balancing teams has nothing to do with matching tiers.

The two are totally different discussions.

Not when the current system treats all players within a tier as equal. Player A is T1, Player B is T1, Player C is T2,

Player A = Player B

Player A != Player C

Player C != Player B

However, if we look at real skill.

Player C > Player A

Player A < Player B

Player B ~= Player C

Edited by Shadowomega1, 15 March 2019 - 04:53 AM.

[Knight Captain Morgan]

Can they at least tweak matchmaker so that it puts at least some of the AFKers and disconnects on red team instead of dumping them all on blue team? Pretty please with sugar on top?

[Ssamout]

Andres Gomez, on 15 March 2019 - 05:30 AM, said:

Can they at least tweak matchmaker so that it puts at least some of the AFKers and disconnects on red team instead of dumping them all on blue team? Pretty please with sugar on top?

You could pay pgi for that flag to get solid teams. What was it, ~15$ a month or so...

[Variant1]

Vxheous, on 21 May 2018 - 06:49 PM, said:

Stomps should happen, even with two perfectly balanced teams. Its due to the snowballing mechanics of this game, once one side loses 1-2 pilots, the numerical/tonnage advantage will create a snowball effect. You see stomps happen in comp play as well, same reason.

Stomps shouldnt happen if the matchmaker pairs up even teams, thats going to be hard but manageable. it is a known fact that mm does not distribute player skill and average match score evenly which results in stomps. I would link you to the thread that has the mathematics but id have to search for it a long time.......

[Tarl Cabot]

Variant1, on 15 March 2019 - 07:31 AM, said:

Stomps shouldnt happen if the matchmaker pairs up even teams, thats going to be hard but manageable. it is a known fact that mm does not distribute player skill and average match score evenly which results in stomps. I would link you to the thread that has the mathematics but id have to search for it a long time.......

Say that to the teams in MWOWC.. Stomps happen. In my experience, most of the time stomps did not happen is when BOTH sides made mistakes and the other side was too timid to capitalize on it, allowing each side to either get out of that predicament by either retreating or re-enforcing to prevent a roll over.

Edited by Tarl Cabot, 15 March 2019 - 08:00 AM.

[K O Z A K]

Variant1, on 15 March 2019 - 07:31 AM, said:

Stomps shouldnt happen if the matchmaker pairs up even teams, thats going to be hard but manageable. it is a known fact that mm does not distribute player skill and average match score evenly which results in stomps. I would link you to the thread that has the mathematics but id have to search for it a long time.......

after all these years you guys haven't figured out this is a pipe dream. There will always be stomps as we can clearly see in comp, and in qp it will always be worse, even with perfectly skill matched teams, someone will be skilling up a mech, someone will be playing with joke loadouts, someone will be drunk, someone will be stoned, someone will be tired, someone will just flat out f*** up (oh yeah, even the best players f*** up sometimes). Avoiding stomps is impossible

it would have been nice to have a skill (not grind) based tier rating a few years ago, it would help make games a little more interesting, but never eliminate stomps, at this point it doesn't really matter

[Ssamout]

Actually if you put together two teams with avg matchscore around 100-150 it could be that the taters are just so inefficient that snowballing is rarer. Lets be honest, a lot of those guys resemble that escort atlas in their ability to shoot.

* edit* individual player matchscores in that 100-150 area, cos an average 250 player would be a wolf in a hen house in there.

Edited by Ssamout, 15 March 2019 - 08:13 AM.

[BWS2K]

Unless things have changed significantly in the last year when I stopped playing, there's simply no way to measure player skill accurately. That's the first factor. Secondly, the Tier System doesn't represent anything of value - certainly not skill alone - that would be of much use to match players by... and that's what the MM uses anyway because it has to use something.

Ex)

Player A is a stellar pilot. Works with the team, communicates, pilots well, strategizes, etc. Makes it to Tier 1 - rightfully so - and then gets a Twitch partnership. Now it's about doing things that viewers want to see, like silly loadouts and challenges. There's no reason to start a new account, so now there's a Tier 1 pilot in Tier 1 doing Tier 5 things for fun. This is allowed.

Player B is a terrible pilot. Never uses chat, goes off solo, etc. BUT only pilots big stompy 'mechs like Assaults with Meta Builds and does tons of damage (but doesn't bother to focus fire and secure kills) and and plays lots of hours a day. Eventually Player B Tiers up to Tier 2 because the Tier System heavily favors damage dealt over anything else. This is allowed.

Player C is a decent pilot and loves to play Lights - one in particular, in which this pilot excels. Communicates with the team and so on but also doesn't mind self-sacrifice via squirreling or distraction either. This 'mech isn't capable of putting out more than, say 225 damage reliably because of ammo constraints and armor but that's a pretty consistent number. This pilot eventually Tiers up to Tier 3 where s/he plateaus. This is allowed.

Player D only plays events because it's actually an alternate account for a Tier 1 pro whose name everyone knows. Being an alternate account infrequently played, however, this account is only Tier 3, so we have a Tier 1 pilot in Tier 3. This is allowed.

With global player population low, or during off-peak hours, all four pilots could be on the same team in a QuickPlay match. Things may have changed (I know they did a re-balance of score factors but it still favored Damage Dealt over anything else) and I might be completely wrong. Here's a video I made a while back:



Thank you for coming to my TED Talk.

[Weeny Machine]

Andres Gomez, on 15 March 2019 - 05:30 AM, said:

Can they at least tweak matchmaker so that it puts at least some of the AFKers and disconnects on red team instead of dumping them all on blue team? Pretty please with sugar on top?

Afking/Intentional DCing is really a plague. It seem every other game you have 2-3 mechs either afking or being "DC" and never come back even if they are the last ones alive

[K O Z A K]

Ssamout, on 15 March 2019 - 08:12 AM, said:

Actually if you put together two teams with avg matchscore around 100-150 it could be that the taters are just so inefficient that snowballing is rarer. Lets be honest, a lot of those guys resemble that escort atlas in their ability to shoot.

* edit* individual player matchscores in that 100-150 area, cos an average 250 player would be a wolf in a hen house in there.

yes, when everyone is in different corners of zimbabwe humping terrain, snowballing is rarer

[adamts01]

Tarl Cabot, on 15 March 2019 - 07:57 AM, said:

Say that to the teams in MWOWC.. Stomps happen.

Hazeclaw, on 15 March 2019 - 08:06 AM, said:

after all these years you guys haven't figured out this is a pipe dream. There will always be stomps as we can clearly see in comp,

I can't seem to find it, but a player tallied the match scores of the teams in stomps over 100 matches in QP and posted the spreadsheet to Reddit. I want to say his classification of a stomp was no more than 2 deaths on the winning team. As far as QP is concerned, almost the entirety of those stomps had a ridiculous average match score imbalance. I've always been on the fence about snowballing because of how comp works, but that data was enough to make me believe that a poor match maker is the primary reason for stomps in a mode with next to no teamwork. Once again, just like with LRMs, the comp scene has little to do with the other 99% of the game.

[K O Z A K]

adamts01, on 15 March 2019 - 09:05 AM, said:

I can't seem to find it, but a player tallied the match scores of the teams in stomps over 100 matches in QP and posted the spreadsheet to Reddit. I want to say his classification of a stomp was no more than 2 deaths on the winning team. As far as QP is concerned, almost the entirety of those stomps had a ridiculous average match score imbalance. I've always been on the fence about snowballing because of how comp works, but that data was enough to make me believe that a poor match maker is the primary reason for stomps in a mode with next to no teamwork. Once again, just like with LRMs, the comp scene has little to do with the other 99% of the game.

you're right comp scene does have little to do with the other 99% of the game, but it does provide the environment where the closest matched teams fight each other, most likely everyone is playing the best possible builds, pretty sure mostly sober and ready and nobody is derping around, so it does reduce the variables to a minimum. And even there within the environment of a few dozen players in division, the skill gap between the top team in the division and the bottom team tends to be pretty significant, which is why you see stomps. And that's within a few dozen best matched players. Now take that into an environment of thousands of players with skill gaps the size of grand canyon, and add a whole bunch of other variables. You guys are dreaming if you think there will ever be consistent 12:11 drops in QP