Jump to content

This Is Why I Can't Take Complaints About Matchmaker Seriously


69 replies to this topic

#41 N0MAD

    Member

  • PipPipPipPipPipPipPipPip
  • 1,757 posts

Posted 11 March 2015 - 06:39 PM

But doesnt WOT primarily base its MM on class or tier of tank? meaning that if you are driving say a Tier that is in the higher tiers you come against guys that have a certain number of games up their belt?using a similar tier tank? and these tiers are are achieved via playing games to unlock that tier, are your put in a game in your tier 1 tank vs a tier 5 tank? not really sure not played it but there are quite a few guys in our TS that play it and i hear them say " hey lets jump into this tier tank"

#42 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 11 March 2015 - 06:41 PM

View PostNik Reaper, on 11 March 2015 - 06:24 PM, said:


True, but that is not what is being discussed as the main point here, you can spectate someone and see just why that game will not end well .

More so the real thing most are interested in is why do we even have a system that trys to keep you at 50/50, now don't get me wrong , I'm not asking for a chance to repitedly euthanize the braindead so my w/l would sky rocket, but why is it so much to ask that we separate players based on elo and w/l and match them against each other? The only practical reason , for now, is the player population of specific values is not enough , so we might have too many of ones and not enough of the others , but doesn't WoT have very high player populations and even they still seems to just mix players of every kind of stat.

The whole 'keeping you at 50%' is a by-product of the system, not the point of the system.

#43 Tarogato

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Civil Servant
  • Civil Servant
  • 6,558 posts
  • LocationUSA

Posted 11 March 2015 - 07:26 PM

View PostN0MAD, on 11 March 2015 - 06:39 PM, said:

But doesnt WOT primarily base its MM on class or tier of tank? meaning that if you are driving say a Tier that is in the higher tiers you come against guys that have a certain number of games up their belt?using a similar tier tank? and these tiers are are achieved via playing games to unlock that tier, are your put in a game in your tier 1 tank vs a tier 5 tank? not really sure not played it but there are quite a few guys in our TS that play it and i hear them say " hey lets jump into this tier tank"

Kinda. I don't play it either, but I pay quite a bit of attention to it and I've been thinking of hopping in.

Tanks come in Tiers, and MMer puts together matches that comprise of a narrow range of tiers. Basically, any tank can go into a match two tiers higher if necessary. A tier 5 and tier 7 can actually see each other in the same match (actually a tier 4 light can see a tier 7 heavy, too, because tier 4 and higher lights get bumped up a tier for some reason)

Edited by Tarogato, 11 March 2015 - 07:27 PM.


#44 1453 R

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Little Devil
  • Little Devil
  • 5,815 posts

Posted 11 March 2015 - 07:43 PM

View PostDavers, on 11 March 2015 - 06:00 PM, said:

Even LoL has to relax it's MM and allow people from different tiers into the same game.

But the nice thing about being able to look up anyone's stats at any given time has it's benefits. I saw a thread once, another 'I am a good player who gets stuck with bad teams'. But then this Diamond player comes along and took the time to look up the posters stats and analyze them. He straight up showed the guy he wasn't a 'good player'- bad buys, bad masteries, and bad CS- and that was on a Champion that is really good at CS. But once he cut past the guys misconceptions and BS, he was able to show him how to actually improve.


There really isn't any 'nice thing' about being able to look up someone's stats. There's enough "SHUT UP NOOB HAHAHA WAT DO U NO ABOUT BLANCE SCRUB" bullscheissen around these parts, we don't need to give people concrete 'proof' that someone has no business playing MWO everz and would be doing the game the greatest possible favor by shutting up, uninstalling, and hanging themselves quietly in a closet.

No matter if someone's actually an intelligent poster with a lot of good points that just has fat fingers, is older than nineteen, or has a job and a life that jeeps him from setting up an IV drip and playing MWO twenty-two hours out of the day - if he's not in the top five percentile of the game, he no longer gets an opinion, period.

And if ye don't think that's where public Elo would go, then you haven't been around this sewage pit nearly as long as it otherwise looks like ye have, man.

#45 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 11 March 2015 - 08:28 PM

View Post1453 R, on 11 March 2015 - 07:43 PM, said:

There really isn't any 'nice thing' about being able to look up someone's stats. There's enough "SHUT UP NOOB HAHAHA WAT DO U NO ABOUT BLANCE SCRUB" bullscheissen around these parts, we don't need to give people concrete 'proof' that someone has no business playing MWO everz and would be doing the game the greatest possible favor by shutting up, uninstalling, and hanging themselves quietly in a closet.

No matter if someone's actually an intelligent poster with a lot of good points that just has fat fingers, is older than nineteen, or has a job and a life that jeeps him from setting up an IV drip and playing MWO twenty-two hours out of the day - if he's not in the top five percentile of the game, he no longer gets an opinion, period.

And if ye don't think that's where public Elo would go, then you haven't been around this sewage pit nearly as long as it otherwise looks like ye have, man.


I think if someone makes a valid point, it will most likely be upheld by good players.

#46 dethikeed

    Member

  • PipPip
  • The Merciless
  • The Merciless
  • 25 posts

Posted 11 March 2015 - 10:00 PM

This is not a scientific observation but I have noticed that I am not seeing as lengthy losing streaks in the solo que that I used to.

I cannot attribute this to any improvement of my skill or ELO, as I have recently returned from a four month break. I suppose that I will simply enjoy it while it lasts.

#47 Nik Reaper

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 1,273 posts

Posted 12 March 2015 - 06:26 AM

View PostDavers, on 11 March 2015 - 06:41 PM, said:

The whole 'keeping you at 50%' is a by-product of the system, not the point of the system.


Are you sure, as I see it as the opposite, the system is there to help the low skilled ones and to add dificulty to the high skilled ones as it doesn't seem to be leaning to creating games of different elo values instead trys to create games of an global average value.

So instead of trying to create a game of 2000 elo because i got a few such players in the initial pass in to a game, it will still try to create a 1500 elo game by finding some lower elo players to average it out, while we could, if there were enough players of needed elo, have a game of about 2000 elo average by having a few ~2200 and a few ~1800 instead of having a few ~2000 and a few ~1200.

I don't know for a fact that the system goes for a 1500 elo average for every game, but that is what it feels like and it would explain much. If anyone heard devs say otherwise to link it.

#48 CygnusX7

    Member

  • PipPipPipPipPipPipPipPip
  • 1,803 posts
  • LocationA desolate moon circling a desolate planet

Posted 12 March 2015 - 07:22 AM

Everyone push saddle! lol

#49 Water Bear

    Member

  • PipPipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 1,137 posts

Posted 12 March 2015 - 07:26 AM

View PostTed Wayz, on 11 March 2015 - 03:30 PM, said:

My issue with matchmaker is that it is inconsistent, or maybe too consistent. I have had streaks of 17 straight losses. And it is not unusual for the wins and losses to come in streaks of six or seven. This would suggest that the matchmaking system contains bias and given certain conditions you will end up always on a good team or always on a bad team.


Water Bear said:

You can't listen to MM complaints for many reasons. Among them is the following:

If it's working perfectly and you're playing at your typical skill level, then you should lose about 50% of your games. So if each match is independent, when you queue up your success is kind of like flipping a coin.

Fun fact: Arbitrarily long runs of consecutive heads and tails appear with positive probabilities in long coin tossing games. I forget the exact details, but basically it is not impossible (nor even really that uncommon) to see, say, a string of 100 consecutive heads if you toss a coin for a long time. Computing this probability is one of those fun first-week-of-class problems in any graduate probability course.


#50 Water Bear

    Member

  • PipPipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 1,137 posts

Posted 12 March 2015 - 07:29 AM

View PostDavers, on 11 March 2015 - 04:21 PM, said:

They have though.

It starts with one player. Then the MM tries to pull in 23 other players with similar Elo, compensating for the various hard and soft caps (such as 3/3/3/3 (soft cap) and no more than 4 of any weight class (hard cap)). It specifically does NOT do what people claim it does- specifically put a good player on a bad team and expect them to carry. The only time it does something similar to that is when one of the 'handful' (PGI's word) of high Elo players are looking for a group- there just isn't enough of them on at any given time to reliably match them against each other.


I would pay good money for a source quote for one of these MM explanations. Every time there's a thread about the MM, I read at least 2-3 conflicting explanations of how the MM works, and no one seems to realize that not everyone in the thread can be right at the same time.

#51 Lily from animove

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • The Devoted
  • The Devoted
  • 13,891 posts
  • LocationOn a dropship to Terra

Posted 12 March 2015 - 07:36 AM

View PostCorbenik, on 11 March 2015 - 10:31 AM, said:

People just like to complain or talk smack for no reason anywhere, had a guy in one pug match he was one of our assaults and we were in River City we spawned at the docks so first thing the other assaults and myself leave the docks and head north to get away from the enemy that usually comes to the docks and the King Crab starts walking towards them, we tell em "hey don't go that way all our team is up north you will die" and sure as heck he dies in a couple secs. and starts talking crap not even game related crap just straight out crap talk.


this, sometiems you have poeple in your teams or the opponents that seem to have no idea about the game and then get crushed within a few seconds. Ad then you know something with the elo distribution went horribly wrong. Espeically when in a match an entire team has only 2 or 3 people goind past 100 damage.

But I think this is the issue with the "skill curve" There is that small area where people are rather similar good/bad, then there is an area where skill suddenly has a good bunch of increase.

and so on low elo, the guy being 50 elo below the other low elo guy may not differ much, but in higher elo, those 50 elo make probably a big difference and so when a MM makes a 100 elo discrepancy in a match on low mid elo, its not that bad, but then when this happens on higher elo ranks, it can have a lot more impact.

#52 Davers

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,886 posts
  • Facebook: Link
  • LocationCanada

Posted 12 March 2015 - 08:55 AM

View PostWater Bear, on 12 March 2015 - 07:29 AM, said:


I would pay good money for a source quote for one of these MM explanations. Every time there's a thread about the MM, I read at least 2-3 conflicting explanations of how the MM works, and no one seems to realize that not everyone in the thread can be right at the same time.


View PostNik Reaper, on 12 March 2015 - 06:26 AM, said:


Are you sure, as I see it as the opposite, the system is there to help the low skilled ones and to add dificulty to the high skilled ones as it doesn't seem to be leaning to creating games of different elo values instead trys to create games of an global average value.

So instead of trying to create a game of 2000 elo because i got a few such players in the initial pass in to a game, it will still try to create a 1500 elo game by finding some lower elo players to average it out, while we could, if there were enough players of needed elo, have a game of about 2000 elo average by having a few ~2200 and a few ~1800 instead of having a few ~2000 and a few ~1200.

I don't know for a fact that the system goes for a 1500 elo average for every game, but that is what it feels like and it would explain much. If anyone heard devs say otherwise to link it.


Karl Berg has a thread in Offtopic that is very informative.

View PostKarl Berg, on 09 April 2014 - 10:41 AM, said:

On deep investigation, for many specific cases, we find that Elo deltas are actually very close; but the problem here is perception. Elo is purposefully a hidden variable, because we really wanted this to be a match quality metric, an invisible hand to help make your experience better, not a stat for users to compete over. Users are making assumptions about Elo ratings that are often highly inaccurate on examination.

View PostKarl Berg, on 27 September 2014 - 04:46 PM, said:


This is correct. The release valves are a function of the oldest request currently assigned to any pending game. The matchmaker is constantly breaking and reforming potential matches around those oldests requests to determine if it can produce an optimal game given it's current resource set and constraints.

That said, if you yourself happen to be the oldest request, Elo doesn't really start loosening until after roughly two minutes of searching. By default, you enter the queue with around a 50 Elo range. That widens to an absolute maximum of around 1400 after 5 minutes searching for solo queue, although I don't think I've ever seen a match much older than about 3.5 minutes in solo queue.


Edited by Davers, 12 March 2015 - 08:56 AM.


#53 Amer

    Member

  • PipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 113 posts

Posted 12 March 2015 - 09:12 AM

WOT doesn't include winning rate or ELO or anything like that in their matchmaking calculations.

Edited by Amer, 12 March 2015 - 09:13 AM.


#54 Water Bear

    Member

  • PipPipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 1,137 posts

Posted 12 March 2015 - 09:13 AM

View PostDavers, on 12 March 2015 - 08:55 AM, said:




Karl Berg has a thread in Offtopic that is very informative.



Fascinating, thank you!

#55 Bigbacon

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,108 posts

Posted 12 March 2015 - 09:16 AM

WoT is just a craptastic game to compare to anyway. you want to talk about BS gameplay.. it doesn't get any better than WoT

it also loves to stick me against tanks many tiers above my own where I can do literally nothing in a round but die in one shot...Least in MWO that doesn't happen.

Edited by Bigbacon, 12 March 2015 - 09:17 AM.


#56 Lostdragon

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 2,713 posts
  • LocationAlabama

Posted 12 March 2015 - 09:24 AM

View PostWater Bear, on 12 March 2015 - 07:29 AM, said:


I would pay good money for a source quote for one of these MM explanations. Every time there's a thread about the MM, I read at least 2-3 conflicting explanations of how the MM works, and no one seems to realize that not everyone in the thread can be right at the same time.


It starts with a target ELO value and tries to put in people close to that, if it can't it expands its tolerance.

Source: http://mwomercs.com/...-making-update/


How does the match maker compose a teams Elo rating, is it average rating or closest to a target?

It's closest to a target value, so the match maker starts trying to make a match for an Elo of say 1300 and will pull in players to those teams closest to those values; however, as mentioned earlier within growing thresholds and those curves will be tuned. Currently it may be a bit 'sloppy' about how it's filling those buckets but over time it will be tuned to be much more precise.
We need to do this carefully over time as generally the cost of precision is time to find a match we want to slowly find a very nice balance between time to find a match and the number of matches that are correctly composed.

#57 Almond Brown

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 5,851 posts

Posted 12 March 2015 - 09:38 AM

View PostDavers, on 11 March 2015 - 12:09 PM, said:

The MM does what it can, but there are too many other factors that affect matches. Maybe we should have check boxes to adjust Elo, such as Drunk, AFK For Doorbell, Distracting Child, Attention Starved Pet, and Wife Aggro? :D


I would add:

Still Raging at the Screen from the last Match and that ugly red haze over my eyes hasn't sufficiently cleared just yet, but I will try. LOL ;)

View PostFate 6, on 11 March 2015 - 12:27 PM, said:

There are times when I say I lost and I could have played better and maybe won, but that doesn't change the fact that either way I'm playing to carry that guy that did sub 50 damage in an assault mech. It's hard to blame myself for the loss sometimes, even if I know it is my job to be the carry, because it shouldn't be my job to be the carry and make up for other people. If I fail to carry, though, it is still my fault, and that's why I think the system is broken.


So it would be safe to say that you have "never" had a bad game. One that others would look at and say. "ouch, he sucks large balls" while of course not having seen the many many Leet 1500 damage 7 kill games that make up your "Gold level elo" rating.

That be about right then?

#58 Nightmare1

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 7,636 posts
  • Twitch: Link
  • LocationPeeking over your shoulder while eating your cookies.

Posted 12 March 2015 - 09:46 AM

No MM is perfect, but this one could use a bit of work. I'd recommend that it account for the individual Mech chassis used in addition to player elo. Two pilots of equal skill fighting each other in disparate Mechs will not a fair outcome make. For example, if I attempted to tackle a Firestarter with a Locust, there's a good chance I would lose if my enemy was my clone (i.e. - equal skill).

Assigning a numeric value to Mechs, similarly to Combat Value, and then applying this with Elo may help provide better balance. I saw something like this happen last year with Ghost Recon: Phantoms. Elo was being matched, but equipment wasn't. As a result, I was often fighting at a significant disadvantage since I have poor equipment (Good fighter but too poor to buy nice gear). When a teams with disproportionate levels of "good" gear were pitted against each other, the team with the better gear normally won. GR:P fixed this by accounting for equipment in the MM settings. Now teams are more evenly matched and the matches tend to be very close battles. My W/L went back to being about 50/50 too.

Not sure how well this would translate to MWO, but it's worth a look I think.

#59 Almond Brown

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 5,851 posts

Posted 12 March 2015 - 09:50 AM

View PostNik Reaper, on 12 March 2015 - 06:26 AM, said:


Are you sure, as I see it as the opposite, the system is there to help the low skilled ones and to add dificulty to the high skilled ones as it doesn't seem to be leaning to creating games of different elo values instead trys to create games of an global average value.

So instead of trying to create a game of 2000 elo because i got a few such players in the initial pass in to a game, it will still try to create a 1500 elo game by finding some lower elo players to average it out, while we could, if there were enough players of needed elo, have a game of about 2000 elo average by having a few ~2200 and a few ~1800 instead of having a few ~2000 and a few ~1200.

I don't know for a fact that the system goes for a 1500 elo average for every game, but that is what it feels like and it would explain much. If anyone heard devs say otherwise to link it.


MM Phase 4 results.

http://mwomercs.com/...ted-april-19th/

Not everyone will believe even those facts, but they don't really want to know the facts as they interfere with their fantasy world. go figure. ;)

#60 Almond Brown

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 5,851 posts

Posted 12 March 2015 - 09:56 AM

View PostCygnusX7, on 12 March 2015 - 07:22 AM, said:

Everyone push saddle! lol


Didn't you know. A survey of 5000 people indicated that more than 32% of respondents did not include themselves when the term "Everyone" was used. LOL! ;)

So when you hear "everyone" do something in MWO, expect only 8 of the 12 to respond at all. YMMV :)

View PostWater Bear, on 12 March 2015 - 07:29 AM, said:


I would pay good money for a source quote for one of these MM explanations. Every time there's a thread about the MM, I read at least 2-3 conflicting explanations of how the MM works, and no one seems to realize that not everyone in the thread can be right at the same time.


Search the Forums. The Dev's did a great breakdown of exactly how their MM works. It didn't prevent players from losing so it was "globally" decided, by osmosis in most cases, that becasue of that, it sucked.

Because it could never be the players, in groups that were more equal to "derpwads" than actual "Teams". ;)

Edited by Almond Brown, 12 March 2015 - 09:57 AM.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users