138 replies to this topic

[Wispsy]

If 80tons is horrendously bad then the only acceptable thing for you would literally be every mech variant had to be mirrored on the enemy team. Would suck and would increase queue times. What I do notice though that Elo seems to be doing alright and giving you seemingly fair matches despite the weight difference.

[Asakara]

Wispsy, on 02 June 2013 - 10:09 AM, said:

If 80tons is horrendously bad then the only acceptable thing for you would literally be every mech variant had to be mirrored on the enemy team. Would suck and would increase queue times. What I do notice though that Elo seems to be doing alright and giving you seemingly fair matches despite the weight difference.

Read the OP again. I am just rating the matches according to PGI's standards. PGI stated 7% of the matches were what they call "Horrendously Bad" match-ups (i.e. the difference between one and two assault mechs). To PGI a "Tolerable" match-up is up to 40 tons difference.

All I am asking for is better weight balancing to make matches more fair for everyone. Do you think that is a bad thing?

Edit: Please ignore the victories and defeats as the point is the seemingly common (to me) "horrendously bad" match weight imbalances when PGI states it only happens 7% of the time.

Edited by Asakara, 02 June 2013 - 10:27 AM.

[Chemie]

Asakara, on 02 June 2013 - 10:15 AM, said:

Read the OP again. I am just rating the matches according to PGI's standards. PGI stated 7% of the matches were what they call "Horrendously Bad" match-ups (i.e. the difference between one and two assault mechs). To PGI a "Tolerable" match-up is up to 40 tons difference.

All I am asking for is better weight balancing to make matches more fair for everyone. Do you think that is a bad thing?

Edit: Please ignore the victories and defeats as the point is the seemingly common (to me) "horrendously bad" match weight imbalances when PGI states it only happens 7% of the time.

There is nothing MM can do if you form up with 3 lights and there are no lights in the queue!

80% of the mechs in the queue are assault/heavy. Many 4-man are dropping with 4 assaults.

If they imposed weight limits, then those assaults would have to sit there and wait a long time for a game...which would give incentive for ppl to drop in lighter mechs.

Until there is a real role warfare, MM is screwed because there are no mechs to match you against.

Edited by Chemie, 02 June 2013 - 10:49 AM.

[BlackBeltJones]

We already know the MM is messed up and produced undesirable results too often but to me OPs post demonstrates how out of favor lights are currently. A lance of assaults and heavies would have likely matched against the same teams but not displayed a weight discrepancy - to me this just shows that lights are under utilized and this will give light pilots the impression of horribly bad match making more often.

Edited by BlackBeltJones, 02 June 2013 - 10:54 AM.

[CancR]

It;s about time people are starting to realize that match up should be by tonnage limits and not this BS 'weight matching' or 'class matching'

[Asakara]

Chemie, on 02 June 2013 - 10:48 AM, said:

There is nothing MM can do if you form up with 3 lights and there are no lights in the queue!

80% of the mechs in the queue are assault/heavy. Many 4-man are dropping with 4 assaults.

If they imposed weight limits, then those assaults would have to sit there and wait a long time for a game...which would give incentive for ppl to drop in lighter mechs.

Until there is a real role warfare, MM is screwed because there are no mechs to match you against.

Yet even when there are other lights it still fails:


Difference: +115 Tons - "Horrendously Bad"

Spoiler

[Deathlike]

If the distribution of the players in the queue were even in weight classes and in premade construction and ELO itself (it assumes that everyone's ELO is the same, which is clearly never the case), then it would work.

However, given that the current meta favors heavies/assaults AND using examples while in some sort of Jenner-based premade, you're pretty much skewing the results.The DCs caused by the netcode don't help either.

If you had some sort of evenly distributed premade (one in each weight class), you might get some sort of credible evidence, but right now, PEOPLE for the most part are screwing with the MM, because of the meta.

The MM on its own should be able to handle minor skewing due to player AND mech distribution that is available at the CURRENT moment. However, there's too many outside factors that is skewing it, thus increasing the likelihood that imbalanced matches are occurring.

Edited by Deathlike, 02 June 2013 - 11:10 AM.

[CancR]

Mechassault of duty: Modern powerups

[Chemie]

Asakara, on 02 June 2013 - 11:02 AM, said:

Yet even when there are other lights it still fails:


Difference: +115 Tons - "Horrendously Bad"

Spoiler

combination of not enough lights in the queue AND ELO.

I revise what I said: Nothing MM can do if you drop with lights and there are no lights of equal ELO in the queue.

Combine low # of players playing and the meta geared to assaults/heavies, and you are screwed.

Edited by Chemie, 02 June 2013 - 11:23 AM.

[Fate 6]

I just had a drop where my team was 5 mediums (4x50 tons, 1 BJ), 1 light (JR7), 1 assault (HGN), 1 heavy (DRG). Enemy team has 2 lights (SDKs), 1 medium (HBK), 3 heavies (2xCTF, 1xJM6), 2 assaults(HGN and AS7). How is this balanced? It EASILY could have been balanced with that setup, but instead we get a team that's well over 100 tons less.

The matchmaking system needs to account for weight more than Elo. Sure we could have tried to cap, but the Spiders are way faster than our mediums. We started off winning the fight but the weight disparity caught up with us in the end when we couldn't take down the Atlas, Hunchback, and CTF with just 4 heavily damaged mediums.

[Nicholas Carlyle]

If a 4 man deciding to play 3 lights is "skewing" the results, and "messing up" the matchmaker...then the matchmaker sucks.

Or the population is very low, and the mech meta is all sorts of fubar.

Take your picks, but they are all bad.

[MasterErrant]

looking at the samples it appears that having higer weight doesn't guantee wins skill is a big factor. and the exact maech do as well. BJ and commandos are less desireable than jenners and Ravens.

[Hammertrial]

Nicholas Carlyle, on 02 June 2013 - 12:33 PM, said:

If a 4 man deciding to play 3 lights is "skewing" the results, and "messing up" the matchmaker...then the matchmaker sucks.

Or the population is very low, and the mech meta is all sorts of fubar.

Take your picks, but they are all bad.

Yes to both? He was playing at 2am EST time. How many people do you think are on at that time of night? And everyone knows the mech meta is fubar. 2/3 of all mechs nearly 2 months ago were heavies/assaults, and I'd aruge that population has only increased with time.

Fate 6, on 02 June 2013 - 12:10 PM, said:

I just had a drop where my team was 5 mediums (4x50 tons, 1 BJ), 1 light (JR7), 1 assault (HGN), 1 heavy (DRG). Enemy team has 2 lights (SDKs), 1 medium (HBK), 3 heavies (2xCTF, 1xJM6), 2 assaults(HGN and AS7). How is this balanced? It EASILY could have been balanced with that setup, but instead we get a team that's well over 100 tons less.

The matchmaking system needs to account for weight more than Elo. Sure we could have tried to cap, but the Spiders are way faster than our mediums. We started off winning the fight but the weight disparity caught up with us in the end when we couldn't take down the Atlas, Hunchback, and CTF with just 4 heavily damaged mediums.

None of that takes into account premades. If 4 of the mediums were a single group and 2 heavies and assaults were another group theres no way that match would have been balanced tonnage wise.

The fact of the matter is that at the end of the day, my 150 sets of data, ops data, and all the other results I've seen that isn't arbitrarily posted nonsense (I can post more matches with 0 tonnage differences than pretty much half the "MATCHMAKING SUX LOLOL" threads posted in a week) shows that even with disparities as high as 165 tons, there's still only roughly a 50% chance for the heavier team to win (in fact OP wins more games than loses even when being significantly outtonned, which shows that the skill based matchmaker is working).

[East Indy]

What if players designated up to three 'Mechs for games, and the matchmaking system offered slots first-come, first-serve based on lobby class needs?

You'd ready your Centurion, your Blackjack and your Cataphract. The searching animation would gyrate, and most of the time a message would appear after a very short time reading

------------------------------

Battle found. Deploy your:

( ) Centurion

( ) Blackjack

-------------------------------------

If you really wanted, you might only ready your Cataphract, but your wait time would be somewhat longer because matches would be balanced for a plurality of medium 'Mechs. But no matter what, you'd enter a battle consisting of about 50% medium 'Mechs all the time.

The game could also encourage modeled team weights by allowing friends to group light-medium-medium-heavy, or light-light-medium-assault, to be put at the front of the queue.

[blinkin]

Ezekeel666, on 01 June 2013 - 11:11 PM, said:

Your argument is invalid. The matchmaker exists so it should be able to gracefully handle every setup thrown at it. Why? Because that is its job. The data presented clearly show that there are cases were the matchmaker fails horrendously indicating that the algorithm used is unstable and needs improvement.

PhoenixFire55, on 02 June 2013 - 04:46 AM, said:

They should be able to drop in whatever they want and matchmaker should find them an appropriate balanced match. If it does not its broken.

if there aren't enough light mechs looking for matches there is nothing match maker can do. would you just like it to tell you that you are SOL whenever there aren't enough light mechs to match up with?

PhoenixFire55, on 02 June 2013 - 05:15 AM, said:

True. But thing is right now tonnage part is the only part thats flexible, thus a lot of tonnage mismatches. Funny part is that the % of matches (I speak about my matches) that aren't anywhere near (aka 8:0-8:1 and visa versa) is way way higher then the % of matches where I have tonnage mismatching, which in turn means that Elo matching means little towards game being close game or not. Should make Elo part flexible and tonnage part non-flexible (inside 20-30 tons tops).

in a tactics based game a single mech in the right place can easily mean the difference between victory and slaughter. if one mech goes down then the team suffers a major loss in DPS. it is the exact same reason why 7v8 matches tend to be so unbalanced.

all it takes is a little push in one direction and everyone dies. i have been playing since the founders program started and i could probably count on one hand the number of matches that came down to the last two mechs, while the 7:1 or the 8:0 matches make up the vast majority of my experience. a team that suffers from a bad plan or is simply unlucky is very likely to be horribly slaughtered regardless of their skill.

this is just the nature of a somewhat deadly tactics based game.

Fate 6, on 02 June 2013 - 07:27 AM, said:

The average weight of their lance was 50 tons (a couple tons under if you want to be exact). That IS balanced. That is basically the definition of balanced. We're looking at average weight here, matchmaking doesn't care about individual weights.

Game is broken if you can't take a 50 ton average lance into MM and come out with a balanced game.

the field is heavily weighted towards assault mechs right now but 50 tons wouldn't be the middle of the road anyway. currently the lightest mech on the field is 25 tons and the heaviest is 100 tons. the middle of the field with the current mech selection is about 62.5 tons (just shy of a catapult).

Asakara, on 02 June 2013 - 08:43 AM, said:

I would like to think my 5,400+ games in a Jenner or Raven since I started in CB may have given me a slight idea on how to play one. Having played with Herms and StuffYouFear in 4 and 8 man groups 3-6 nights a week for the last 6 months seems to help a bit with being coordinated.

Then again, as you probably know, there is only so much you can do in a light sometimes... Especially when so many heavy and assault builds need just a single solid alpha to kill you.

i know the saying "practice makes perfect" gets spouted all of the time, BUT for practice to work you have to actually see the mistakes you are making and come up with solutions. the much more common scenario is people find something that they perceive as a "winning" tactic and then they follow it zealously. people decide that they know everything there is to know and don't put any effort into advancing their abilities, even to the point of detesting any new ideas that might be presented.

now for a long time i have run light mechs WITH a good connection. WITHOUT using the lag shield it is very possible to kill most assault and heavy mechs reliably with a jenner. when i play my jenner i really like facing assault mechs far more than light mechs, because most of the time larger mechs are facing the wrong direction and i get to shoot them in the back regularly.

Asakara, on 02 June 2013 - 10:02 AM, said:

Sadly, these "Horrendously Bad" match-ups are common in my experience.. Even when dropping with more of a mix or less than 4. But whatever.


Difference: +135 Tons - "Horrendously Bad"

Spoiler

Difference: +160 Tons - "Horrendously Bad"

Spoiler

Difference: +115 Tons - "Horrendously Bad"

Spoiler

Difference: +80 Tons - "Horrendously Bad"

Spoiler

Difference: +80 Tons - "Horrendously Bad"

Spoiler

Difference: +265 Tons - "Horrendously Bad" - The matchmaker was really trolling us this one. LOL

Spoiler

these stats aren't much better. i am still seeing lots of ravens. your total number of grouped mechs in that list is 21 out of that 15 are light mechs (ravens) 4 are mediums (trebuchet) 1 heavy (illya) and 1 assault (highlander). 6 matches with one being a 2 group and one being a 3 group, while the other 4 are 4 groups.

this data is still horribly skewed.

Asakara, on 02 June 2013 - 10:15 AM, said:

Read the OP again. I am just rating the matches according to PGI's standards. PGI stated 7% of the matches were what they call "Horrendously Bad" match-ups (i.e. the difference between one and two assault mechs). To PGI a "Tolerable" match-up is up to 40 tons difference.

All I am asking for is better weight balancing to make matches more fair for everyone. Do you think that is a bad thing?

Edit: Please ignore the victories and defeats as the point is the seemingly common (to me) "horrendously bad" match weight imbalances when PGI states it only happens 7% of the time.

PGI also states (from what i have heard) that assault mechs make up 80% of the field, yet you seem to be consistently running matches with 25-50% of your team as light mechs. i suspect your matches make up a sizable portion of that 7%, because you are pushing the match maker beyond it's limits. if there are not enough other light mechs in the list there is absolutely nothing any system could do.

CancR, on 02 June 2013 - 11:00 AM, said:

It;s about time people are starting to realize that match up should be by tonnage limits and not this BS 'weight matching' or 'class matching'

that has almost nothing to do with the current conversation. they are matching light mechs against assault mechs. the classes aren't even being matched.

Asakara, on 02 June 2013 - 11:02 AM, said:

Yet even when there are other lights it still fails:


Difference: +115 Tons - "Horrendously Bad"

Spoiler

ok this one is actually valid and shows that there is a real problem.

Fate 6, on 02 June 2013 - 12:10 PM, said:

I just had a drop where my team was 5 mediums (4x50 tons, 1 BJ), 1 light (JR7), 1 assault (HGN), 1 heavy (DRG). Enemy team has 2 lights (SDKs), 1 medium (HBK), 3 heavies (2xCTF, 1xJM6), 2 assaults(HGN and AS7). How is this balanced? It EASILY could have been balanced with that setup, but instead we get a team that's well over 100 tons less.

The matchmaking system needs to account for weight more than Elo. Sure we could have tried to cap, but the Spiders are way faster than our mediums. We started off winning the fight but the weight disparity caught up with us in the end when we couldn't take down the Atlas, Hunchback, and CTF with just 4 heavily damaged mediums.

were there any groups on your side? i could easily see that happening if there was a four man on each side and there would be nothing that any match maker could do other than tell you to f*** off.

Edited by blinkin, 02 June 2013 - 10:48 PM.

[PhoenixFire55]

blinkin, on 02 June 2013 - 10:45 PM, said:

if there aren't enough light mechs looking for matches there is nothing match maker can do. would you just like it to tell you that you are SOL whenever there aren't enough light mechs to match up with?

No, I want matchmaker to have a minimum limit of how many mechs of same time are in game. If you are in a light mech you'll be able to find a match really quick because they will be in deficit. If you are in an assault mech you'll be waiting for a long time till matchmaker finds you light mechs for your match. Should make people use different chassies, don't just stick to assault pop-tarts and ppc boats.

[Hammertrial]

PhoenixFire55, on 03 June 2013 - 12:46 AM, said:

No, I want matchmaker to have a minimum limit of how many mechs of same time are in game. If you are in a light mech you'll be able to find a match really quick because they will be in deficit. If you are in an assault mech you'll be waiting for a long time till matchmaker finds you light mechs for your match. Should make people use different chassies, don't just stick to assault pop-tarts and ppc boats.

That's a bad idea. Balancing using a MM queue time is as bad as R&R, and just makes people not want to play because they can drop in Hawken in 30s rather than waiting 10mins for a light to show up.

Things that would bring at least some lights back into the game is:

i) A complete rework of rewards rewarding objectives rather than shooting. If by capping I can earn more c-bills than a 300 damage 5 assist match people will actually want to do this. This makes the light trolls come out in force, and either makes the enemy teams start dropping lights to counter, or a turtle meta will appear which will still make scouting supreme because it'll let you know about the turtle in the first place.

2) Some sort of balance with high alphas. Getting one shot is pretty much the least fun thing you can have in a non-respawn relatively slow paced game. There's a reason they are called AWPwhores in CS.

3) Get rid of speed cap. Lights should go fast, and removal of the weight based speed cap (and keeping the stock engine based one) would allow lights to really show their "quirks", rather than being inferior Jenners or 3Ls. Commandos and spiders at high speeds will be able to outrun the streak boats attacking it, which will make them far less gimped than they are now.

[OneEyed Jack]

PhoenixFire55, on 02 June 2013 - 05:04 AM, said:

Then don't claim your matchmaker is working according to plan. If its not doing its job for whatever reason its broken. If players pool is small and can't allow for Elo and tonnage matching get rid of Elo.

First off, it's not my matchmaker. Second, it could make perfectly balanced matches, at least insofar as skill level can be measured and calculated, but at the cost of longer wait times. [redacted]

PhoenixFire55, on 02 June 2013 - 05:15 AM, said:

True. But thing is right now tonnage part is the only part thats flexible, thus a lot of tonnage mismatches. Funny part is that the % of matches (I speak about my matches) that aren't anywhere near (aka 8:0-8:1 and visa versa) is way way higher then the % of matches where I have tonnage mismatching, which in turn means that Elo matching means little towards game being close game or not. Should make Elo part flexible and tonnage part non-flexible (inside 20-30 tons tops).

The fact that you're even bringing up final scores as a measure of balanced matches just shows you have no clue what you're talking about. Perhaps at the low end of the Elo range, where it's still a pure brawl-fest battle of attrition, with no tactics involved, then that would be accurate. At moderate skill levels and above, you will find some matches like that, because they go back and forth, but many matches will simply cascade into a 8-2, or worse, once the first mechs drops.

Nicholas Carlyle, on 02 June 2013 - 12:33 PM, said:

If a 4 man deciding to play 3 lights is "skewing" the results, and "messing up" the matchmaker...then the matchmaker sucks.

Or the population is very low, and the mech meta is all sorts of fubar.

Take your picks, but they are all bad.

Wrong. The matchmaker doesn't have access to every player in the game to pick from. It doesn't even have access to every player currently logged in. In can only choose from the players looking for a match right that moment. As has been said multiple times, it could be much more accurate, but at the cost of waiting for the right mix of players/mechs to all be waiting for a match at the same time, which would create wait times that the player-base finds unacceptable.

Edited by Dakkath, 03 June 2013 - 12:23 PM.
redacted unconstructive

[Derffe]

Hmm. I see the point on the weight limits leading to increasing wait times, and yet I've been fully aware (this past week in fact) that teams are getting mis-matched weight wise by quite a bit (100+ tons).

Not sure what the solution would be. Create an option where you can drop in weight-restricted matches OR the current set-up? Possibly, though it begs the question as to how many players would actually go for a weight-restricted drop. If not many, its might not be something high on the priority list for development.

[Asakara]

I hope people are not confusing this for some science experiment.. It is not. This was exactly what I said in the OP, 3 friends joining with a 4th to do some drops. We played what we liked best... And this was the result.

I am not saying this is what everyone experiences... This is just what my friends and I experience daily most of the time, even though PGI says it only happens 7% of the time..

I say the matchmaker, whose sole job is to create balanced matches, must be able to handle a few friends playing this team game together in what they like best without coming up with "horrendously bad" match-ups.. That is its sole function, and for me.. It constantly seems to fail. There are only 4 weight categories after all.. If having 3 people on a side in 1 of the 4 weight categories is such a major problem for this system then it is broken in my book.