Jump to content

Matchmaker Still Curb Stomp After Curb Stomp

Gameplay

52 replies to this topic

#41 DaZur

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 7,511 posts
  • LocationWisconsin

Posted 03 July 2014 - 07:50 PM

View PostSaxie, on 03 July 2014 - 06:28 PM, said:

Well we better find another boogey man now that the "SYNC Dropping groups" are done... lol

Oh, believe me... They will. :ph34r:

View PostWillard Phule, on 03 July 2014 - 03:29 PM, said:

Chaos Theory...when a drooling mouthbreather stands still and zooms in, it effects the weather patterns of Terra Therma and guarantees a loss for your team.

Yep, I get it.

Not to be picayune... But you're actually describing the butterfly effect not chaos theory. That said, IMHO both are equally present on our battlefields so you get a point for abstract thinking... ;)

Edited by DaZur, 03 July 2014 - 09:25 PM.


#42 Simbacca

    Member

  • PipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 797 posts
  • LocationCanada

Posted 04 July 2014 - 06:58 AM

View PostWintersdark, on 03 July 2014 - 06:23 PM, said:

Far more likely answers:
1) Just bad luck.
2) You're not as good with that mech as you though
3) Your Medium mech Elo rating is much higher than it ought to be, for whatever reason, and is matching you against much better opponents.

Update Hunchback-4P:
39 Wins, 63 Losses [18 wins, 47 losses since measurement start]

From what I can tell the issues seem to be chassis specific:
Shadowhawk 5M == 11 Wins, 20 Losses
Grid Iron == 21 Wins, 19 Losses
Cicada 3C == 25 Wins, 23 Losses
Banshee-3S == 43 Wins, 31 Losses
Battlemaster-3M == 29 Wins, 34 Losses
Spider-5K == 37 Wins, 29 Losses
Adder-D == 14 Wins, 21 Losses
etc.. (I am not going to list the scores of all 93 mechs in my collection)

This really makes me wonder how Matchmaker does its choices....

View PostBe Rough With Me Plz, on 03 July 2014 - 06:28 PM, said:

:/ We're having opposite experiences it seems... I'm roughly at a 3-4:1 win to loss ratio since the patch and increased my overall kdr by ~0.2. Guess you're just unlucky with the people you get matched with/the people you're being matched with enjoy turtling and being surrounded.

I don't know enough about the whole ELO MM criteria to comment on that, but from my rough understanding the more losses you accrue the lower your ELO which leads to being matched with players who are less likely to achieve a group win? Maybe switching up mechs chassis will turn the loss-tide.

Either way, I hope your experience improves.

The problem I describe seems for whatever reason contained to specific mechs (see above) - and that is what makes it frustrating. For the time being, no more Hunchback-4P....

#43 Lykaon

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,815 posts

Posted 04 July 2014 - 07:58 AM

View PostWillard Phule, on 03 July 2014 - 02:07 PM, said:

Working as intended.

3/3/3/3 is merely a guideline...not written in stone.

Elo (Extremely laughable outcome) is working EXACTLY as intended. Every team is SUPPOSED to have 8 brand new mouthbreathers stumbling around fully zoomed in and shooting their teammates. The other 4 are supposed to carry them against the premades.



You mean the other four ARE the premade.

May want to read up on how the new match maker works.

#44 ztac

    Member

  • PipPipPipPipPipPipPip
  • 624 posts

Posted 04 July 2014 - 07:59 AM

I would certainly blame the matchmaker , you don't need science but you do need common sense. The worse a players performance is the lower his ELO will go , thus the bad players should all end up at the bottom playing each other and so on and so forth all the way to the top.

The way the matchmaker is behaving would suggest that there simply is not the player base to put those kinds of matches together. Or that the matchmaker is simply next to useless or does not take ELO into account. Take your pick!

#45 Wintersdark

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • 13,375 posts
  • Google+: Link
  • Twitter: Link
  • LocationCalgary, AB

Posted 04 July 2014 - 12:47 PM

View PostSimbacca, on 04 July 2014 - 06:58 AM, said:

Update Hunchback-4P:
39 Wins, 63 Losses [18 wins, 47 losses since measurement start]

From what I can tell the issues seem to be chassis specific:
Shadowhawk 5M == 11 Wins, 20 Losses
Grid Iron == 21 Wins, 19 Losses
Cicada 3C == 25 Wins, 23 Losses
Banshee-3S == 43 Wins, 31 Losses
Battlemaster-3M == 29 Wins, 34 Losses
Spider-5K == 37 Wins, 29 Losses
Adder-D == 14 Wins, 21 Losses
etc.. (I am not going to list the scores of all 93 mechs in my collection)

This really makes me wonder how Matchmaker does its choices....


The problem I describe seems for whatever reason contained to specific mechs (see above) - and that is what makes it frustrating. For the time being, no more Hunchback-4P....


First, understand that the matchmaker is only concerned with weight class, not variant. You have an Elo score for each class. So, the only relevant stats to this discussion are your other medium mech chassis.

Now, because the Matchmaker doesn't care about what particular mech you're in, outside of weight class, your differing WLR for different variants has no bearing on the matchmaker whatsoever. Those losses must be either or both due to:


You suck in a HBK-4P (running 9ML and triggering Ghost Heat?) Note that you do worse in a SHD-5M (0.55:1) and closer to 1:1 with other chassis, so the 4P isn't your worst/only bad medium.

This has nothing whatsoever to do with the matchmaker. Now, I'm not saying it's all on you either - there are 11 other people in every match adding to individual match randomness - but doing worse in a particular chassis is highly indicative of that chassis just not being good for you personally.

But it absolutely has nothing at all to do with the Matchmaker.

#46 Wintersdark

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • 13,375 posts
  • Google+: Link
  • Twitter: Link
  • LocationCalgary, AB

Posted 04 July 2014 - 12:53 PM

View Postztac, on 04 July 2014 - 07:59 AM, said:

I would certainly blame the matchmaker , you don't need science but you do need common sense. The worse a players performance is the lower his ELO will go , thus the bad players should all end up at the bottom playing each other and so on and so forth all the way to the top.

The way the matchmaker is behaving would suggest that there simply is not the player base to put those kinds of matches together. Or that the matchmaker is simply next to useless or does not take ELO into account. Take your pick!

Or you have no idea what you're talking about.

If he's just not good in a HBK-4P (it's the most powerful of the hunchbacks, but requires the most skill to fully utilize) that will result in exactly his situation. It's very, very normal for people to be much better in some chassis than in others.

Also, keep in mind the Matchmaker requires a substantial amount of matches (50+ to get you where you should be), and that still doesn't garauntee a 1:1 WLR in any given chassis, but rather that in aggregate your matches will approach 1:1 averaged over the entire weight class. Buy a new chassis, it'll bounce around a bit because without pilot skills or personal experience, your performance for several matches will be below what it used to be. Climb into another of your skilled mechs (pilot efficiency and personal skill) and you'll be playing against somewhat lower ranked players, so this will make the disparity seem greater.

Elo is all about balancing it out over the long term.

#47 Haipyng

    Member

  • PipPipPipPipPipPipPip
  • The Grizzly
  • The Grizzly
  • 595 posts
  • LocationIn Transit

Posted 04 July 2014 - 12:54 PM

Solo, MM is working good for me and certainly better than before. Group MM is borked as far as sometimes matching multiple small groups up against 8 to 12 man groups.

#48 xe N on

    Member

  • PipPipPipPipPipPipPipPip
  • 1,335 posts
  • LocationGermany

Posted 04 July 2014 - 01:06 PM

View PostWintersdark, on 04 July 2014 - 12:47 PM, said:


You suck in a HBK-4P (running 9ML and triggering Ghost Heat?) Note that you do worse in a SHD-5M (0.55:1) and closer to 1:1 with other chassis, so the 4P isn't your worst/only bad medium.


As far as I know the ELO is not calculated from your kill/death or average damage ratio per match, but from win/loss of games. You can perform exeptionally in your team like 500+ damage and 3+ kills, however, if your team is just teribad, you will, nevertheless, lose and decrease your ELO.

For example: My Victor DS has a kill ration of 2.11 and scored average 443 damage per match at a total umber of 129 matches by pure PUG-play (and I have done several matches with some experimental layout that even decreased my efficiency in this chassis) . However, my win/loss ratio with the DS is only 0.9. Apperently, both values are contradictory.

Edited by xe N on, 04 July 2014 - 01:09 PM.


#49 Wintersdark

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • 13,375 posts
  • Google+: Link
  • Twitter: Link
  • LocationCalgary, AB

Posted 04 July 2014 - 01:25 PM

View Postxe N on, on 04 July 2014 - 01:06 PM, said:

As far as I know the ELO is not calculated from your kill/death or average damage ratio per match, but from win/loss of games. You can perform exeptionally in your team like 500+ damage and 3+ kills, however, if your team is just teribad, you will, nevertheless, lose and decrease your ELO.

For example: My Victor DS has a kill ration of 2.11 and scored average 443 damage per match at a total umber of 129 matches by pure PUG-play (and I have done several matches with some experimental layout that even decreased my efficiency in this chassis) . However, my win/loss ratio with the DS is only 0.9. Apperently, both values are contradictory.


How Elo works is spelled out very well in the Command Chair section of this forum, but I'll explain here.

You have 4 Elo scores, one per weight class.

Elo is based solely on wins and losses considering the two team's relative Elo scores.

What happens is this: the matchmaker takes the difference in the team Elo scores (average of all the players), and calculates the likelihood of each team winning.

The math is publicly available, but I'm not going to dig it out here.

If your team was rated higher, the MM expects you to win. If you do win, there is little to no change to your score. If you lose, your score is reduced, by up to 50 points (max) depending on how far apart the teams scores where.
If your team was rated lower, the MM expected you to lose. If you win, your score increases a lot, whereas if you do indeed lose it doesn't decrease so much or not at all.

Like rated teams will cause small Elo adjustments on win/loss.

So, its possible to win more games than you lose, but have your score _decrease_ or vise versa (though that's unlikely).

Your score has NOTHING to do with in match performance. It can't, because there is no way to actually score that properly. Elo isn't a ranking of how good you are at killing robots, its a measure of how much your actions contribute to victory, no matter what those actions are.

#50 Simbacca

    Member

  • PipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 797 posts
  • LocationCanada

Posted 04 July 2014 - 01:56 PM

View PostWintersdark, on 04 July 2014 - 12:47 PM, said:


First, understand that the matchmaker is only concerned with weight class, not variant. You have an Elo score for each class. So, the only relevant stats to this discussion are your other medium mech chassis.

Now, because the Matchmaker doesn't care about what particular mech you're in, outside of weight class, your differing WLR for different variants has no bearing on the matchmaker whatsoever. Those losses must be either or both due to:


You suck in a HBK-4P (running 9ML and triggering Ghost Heat?) Note that you do worse in a SHD-5M (0.55:1) and closer to 1:1 with other chassis, so the 4P isn't your worst/only bad medium.

This has nothing whatsoever to do with the matchmaker. Now, I'm not saying it's all on you either - there are 11 other people in every match adding to individual match randomness - but doing worse in a particular chassis is highly indicative of that chassis just not being good for you personally.

But it absolutely has nothing at all to do with the Matchmaker.

My Hunchback-4P = 80 Kills. 65 Deaths. Many Many Many Assists
6 Medium Lasers
1 ER PPC
1 TAG

With that said, it still does not explain loss, after loss, after loss - with 2/3 of those losses being curbstomps. One mech (especially in the 50 ton range) is not the reason for a string of losses.

#51 Wintersdark

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • 13,375 posts
  • Google+: Link
  • Twitter: Link
  • LocationCalgary, AB

Posted 04 July 2014 - 02:50 PM

View PostSimbacca, on 04 July 2014 - 01:56 PM, said:

My Hunchback-4P = 80 Kills. 65 Deaths. Many Many Many Assists
6 Medium Lasers
1 ER PPC
1 TAG

With that said, it still does not explain loss, after loss, after loss - with 2/3 of those losses being curbstomps. One mech (especially in the 50 ton range) is not the reason for a string of losses.


Correlation is not causation. You are one of 24 players. Blaming the matchmaker for those losses is absolutely ridiculous, because the matchmaker doesn't care which of your mediums you're in.

#52 Khobai

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 23,969 posts

Posted 04 July 2014 - 02:53 PM

3/3/3/3 is far from perfect but its a definite improvement over what we had before which was absolutely nothing.

#53 ollo

    Member

  • PipPipPipPipPipPipPipPip
  • The Merciless
  • The Merciless
  • 1,035 posts

Posted 04 July 2014 - 03:13 PM

View PostDaZur, on 03 July 2014 - 02:51 PM, said:

This just in: MM cannot control engagement outcomes only set the parameters in which the participants engage.

Or in other words... the MM can't ensure terms of outcome.

It's called "chaos theory".. might want to look it up. :)

:D


likeliness of winning 100%, actual outcome: 30% wins. Blame the team i think...





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users