Jump to content

My Musings On Elo


50 replies to this topic

#1 CDLord HHGD

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 9,190 posts
  • Location"You're not comp if you're not stock."

Posted 15 June 2015 - 08:54 AM

So I've been thinking of ELO and how it works/affects me. This coincides with another though along the lines of "How many times do I have to lose to get out of "ELO Hell?".

This could be an "It's me, not you." outcome, but I feel it's more like "It's you, not me.". :P

I could suck at this game (which I have compelling evidence that I do not), and ELO puts me with other bads.

I could be awesome and the MM puts me with bads to average out the equation.

I could be in "ELO Hell" where I am above the bads, but below the goods that my ELO places me with.

And please, do NOT take offence at my use of the terms "goods" and "bads", I only use them to differentiate the performance I see from my team and opponents.

Any other musings about ELO? :)

#2 Vandul

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,342 posts
  • LocationYork, New

Posted 15 June 2015 - 09:05 AM

I dont really pay attention to it, but I do notice it. I took a 3 week break and when I came back, I picked up my Zeus' (which were all recently mastered). While I did okay, it was not a winning foray into puglandia. I attribute that to some rust, but...

when I switched to my Grasshoppers (which Ive never played to this point), it was a significant difference ELO wise. Not only were the names different, but the consistency and level of play was several marks below where I was at with my Zeus'.

#3 Ultimax

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 6,979 posts

Posted 15 June 2015 - 09:07 AM

I've come to the opinion that winning and losing in the solo PUG queue is based on which team the MM allocates fewer bads to - or those one in 20 or 30 matches when the MM is straight up havin a giggle and clearly puts a concentration of good players all on one team.


You can sometimes influence this with your personal performance, but usually 2 or 3 bads will just evaporate on contact with the enemy and the match is often already over at that point.

Edited by Ultimatum X, 15 June 2015 - 09:09 AM.


#4 El Bandito

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 26,736 posts
  • LocationStill doing ungodly amount of damage, but with more accuracy.

Posted 15 June 2015 - 09:14 AM

Elo Hell is just a myth, perpetuated by the MOBA players. As one's skill improves, so does his standings.

RNGesus though, is real.

Edited by El Bandito, 15 June 2015 - 09:14 AM.


#5 Kristov Kerensky

    Member

  • PipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 2,909 posts

Posted 15 June 2015 - 09:20 AM

Usually I find the MM has done it's job fairly well, both teams are evenly matched skill wise, don't see newbs brand new to the game in Trial Mechs. I do see a number of high comp team tags often, so I guess I'm slightly above average but not near top of the Elo brackets, which fits my WLR of 1.04(it's come up recently!).

I DO wonder about the MM when I see totally new to the game players in my drop, people in Trials who don't know how to get out of Zoom, don't understand weapon groups or ranges, or even know how to fire AT a target instead of just firing blindly and tearing the hell out of the landscape. I think many of us see these types of players on occasion, and we shouldn't EVER be seeing them in our drops. I actually had this happen multiple times yesterday and it was...annoying. I have nothing against new players, I'm more than happy to give them private match drops to teach them how things work, but to have them put in with players like me is a bit much, much less the 228 or EmP players I get in my drops, wtbf is the MM doing in these cases?

Stomps, I've noticed, tend to happen when both teams are actually fairly well balanced, while the close games come about when there's a noticable disparity in the skills of the two teams, ie : I see newbs on both teams along with skilled players. Stomps are to be expected when both teams are evenly matched, we see an opening and exploit it for that quick win, often without anyone saying anything to anyone else, we have the skills to see what needs to be done and do it. Unevenly skilled teams don't do that as often, usually because the newbs aren't aware of what they are seeing and don't respond to it, even when VOIP and chat commands are given to point it out.

#6 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 15 June 2015 - 09:38 AM

View Postcdlord, on 15 June 2015 - 08:54 AM, said:

So I've been thinking of ELO and how it works/affects me. This coincides with another though along the lines of "How many times do I have to lose to get out of "ELO Hell?".

This could be an "It's me, not you." outcome, but I feel it's more like "It's you, not me.". :P

I could suck at this game (which I have compelling evidence that I do not), and ELO puts me with other bads.

I could be awesome and the MM puts me with bads to average out the equation.

I could be in "ELO Hell" where I am above the bads, but below the goods that my ELO places me with.

And please, do NOT take offence at my use of the terms "goods" and "bads", I only use them to differentiate the performance I see from my team and opponents.

Any other musings about ELO? :)

Elo does not affect Attrition. in fact Attrition throws Elo out the airlock!

#7 Davers

    Member

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

Posted 15 June 2015 - 09:46 AM

View PostJoseph Mallan, on 15 June 2015 - 09:38 AM, said:

Elo does not affect Attrition. in fact Attrition throws Elo out the airlock!

How so? There is certainly attrition in Chess.

#8 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 15 June 2015 - 09:53 AM

View PostDavers, on 15 June 2015 - 09:46 AM, said:

How so? There is certainly attrition in Chess.

Yes and its also strategic. Every piece has a use, and one can choose to sacrifice pieces strategically. But Unlike Chess, for every one player lost, a double team can occur. Massed fire ends an enemy in less time. IF in chess I could move 2 pieces for every one you lose I increase my chances of victory over you.

Good question by the way. :)

Edited by Joseph Mallan, 15 June 2015 - 09:54 AM.


#9 Soy

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,689 posts
  • Locationtrue Lord system

Posted 15 June 2015 - 09:56 AM

Went out last night in a LRM Dire Wolf, watchin a movie real late just chillin. You know, 4am on a Sunday night/Monday morning. Real serious ****, right.

I lost for every round for 80 minutes straight, and all I could think was. "lol MM actually expects me to make a real DWolf and really try and really win" .........like fffffffffffffffffffff that, if I wanna carry I'll carry..... but if I take some random **** out at the oddest time, then Elo doesn't give a ****, it responds with a "get pumped a dozen rounds in a row, nerd!"

Like... ok, you almost got me game, I almost cared enough to try. Almost. When will this game be able to use some sort of advanced metrics to tell whether or not someone is even trying?

To quote Cyborx... where is the ******* "meters walked per minute" stat, where is the "shots fired per round" stat, when is it gonna take a look at the weapon you're using and say "he's used this weapon 6 times, ever, on any mechs" etc............................... "average damage per killing blow" etc

.....what about "SNEAKINESS" how often are you locked on? Give a stat on that PGI... Oh wow that LRM guy getting carried by teammates sitting in back, such high Elo, such high W:L, wait he hasn't moved in the past 17 rounds in 3 hours. Not targeted once. WTF? Same as gauss wolves in group play, what the **** is that garbage. Nice job walking for 60 secs, sitting behind a rock, and firing 9 times a game. Pro status; nobody even ******* shot at him entire weekend most likely...

Elo would work a lot better if there were some sort of actual stats being used to dictate things besides Win/Loss. Win/Loss means jack **** if you're being carried by people every session, or only pugging.

Edited by Soy, 15 June 2015 - 10:02 AM.


#10 DaZur

    Member

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

Posted 15 June 2015 - 10:03 AM

My Elo musing?

Like that sinking feeling in your stomach when your significant other responds to your query of how does she feel with a curt "fine"... One needs to recognize and embrace that gut feeling one gets when they drop into a match and you just know it's a Elo game.

#11 Kristov Kerensky

    Member

  • PipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 2,909 posts

Posted 15 June 2015 - 10:10 AM

Exactly Soy, some of the high end comp team players I see in my solo drops are obviously NOT the cream of the crop, some of them ARE good, but some of them are obviously being carried by their teammates. WLR is a horrible standard for ranking when dropping in the solo que, we have no personal control of that stat, so it doesn't reflect OUR ability at all.

But hey, it IS how Elo works, and that's the system PGI decided to use because LoL and others use it, despite the fact that even in those games it's shown repeatedly to be a bad system. Problem is, no one has a working ranking system where twitch skills AND tactical thinking are combined to produce somewhat effective rankings for a 1v1, much less for a team vs team setting. So we're stuck with the industry standard, broken as it is.

#12 Tiamat of the Sea

    Member

  • PipPipPipPipPipPipPipPip
  • Guardian
  • Guardian
  • 1,326 posts

Posted 15 June 2015 - 10:12 AM

Elo is a poor system for ranking players for this game.

Unfortunately, there is no existing, well known, tested system that would work better for this kind of game, so we're more or less stuck with it.

Just a reminder- Elo system was created for chess, a game with:
  • No inherent random elements.
  • One player per side so there can be no teaming up.
  • One player per side so vague things like 'cooperation' have no effect.
  • One player per side so it doesn't matter which part of the OPFOR you engage at any given moment.
  • One player per side so you literally cannot wind up crippled in terms of your capability.
  • No hidden units or other info of any sort on either side.

Edited by Quickdraw Crobat, 15 June 2015 - 10:13 AM.


#13 Templar Dane

    Member

  • PipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 1,057 posts

Posted 15 June 2015 - 10:12 AM

I find that I'll win more matches by being super aggressive and carrying narc or BAP. I play scout/cat herder. Seems most of these pugs have no clue where the enemy is unless they can see red doritos.

Like one incident where a pirate's bane came into a clumped up group of at least 8 of us, and started circling a whale. The whale turned around and looked at us like "why you shoot meh" and the others had no idea what was happening, even while I was spamming AC rounds at the locust. I kept thinking any second now these guys are going to see the locust, laser death awaits this little locust. Some of my adder's UAC rounds hit the locust and he ****** off. I don't remember seeing anybody but me firing at him even once.

Conversely, if you have BAP and target an enemy the herd will wander in that red dorito's direction. Narc also works really well if you can nail something so that it looks like it's by itself so that even the cowardly ones go mine?mine?mine?mine? mode. They'll deathball in that direction almost without fail.

#14 Wintersdark

    Member

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

Posted 15 June 2015 - 10:12 AM

View PostVandul, on 15 June 2015 - 09:05 AM, said:

I dont really pay attention to it, but I do notice it. I took a 3 week break and when I came back, I picked up my Zeus' (which were all recently mastered). While I did okay, it was not a winning foray into puglandia. I attribute that to some rust, but...

when I switched to my Grasshoppers (which Ive never played to this point), it was a significant difference ELO wise. Not only were the names different, but the consistency and level of play was several marks below where I was at with my Zeus'.
You have a different Elo rating for every weight class. Thus, your assault (Zeus) and Heavy (Grasshopper) could be in entirely different ranges.

#15 Mawai

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,495 posts

Posted 15 June 2015 - 10:21 AM

Elo is not win/loss but it is related.

1) The amount of change in your Elo is related to which team is predicted to win and by how much ... then by whether you fill those expectations or not.

Actual Result/Expected Result/ Change in Elo

Win/Win/Elo up a little
Win/Loss/Elo up more
Loss/Loss/Elo down a little
Loss/Win/Elo down more

The exact change in Elo depends also on how much you were actually expected to win or lose.

2) A player can not be carried forever. It doesn't work that way. In PUG land there is one common element in every match you play ... that is you. Over some number of matches, it is possible to extract the "signal" value of your average contribution to matches ... that is your Elo. Mixing group and solo matches will break Elo since you will have a different Elo on a team than separately. Honestly, they should never have had one Elo value for group and solo. The other factor they have taken into account is weight class. So you have a different Elo in each weight class ... and it can show ... my Elo in heavies is in a decent place at the moment ... in lights that is another story :)

Finally, the composition of any given match depends on several factors ..
- the Elo distribution of folks in the queues
- how LONG the MM wants players to wait for a match ... the tighter the Elo match the longer you may have to wait


The actual play of a match is another thing altogether ...

1) Sometimes good players peek around a corner and are alpha'd by the enemy lance that they did not know was there
2) Sometimes good players are tired/intoxicated/moody :) and decide that a Steiner scout role would be fun :)
3) Sometimes good players come across an opponent and decide to duel them :) ... fun eh? ... until his lancemates come around the corner and leave your team down one
4) Sometimes good players take bad builds (see LRM Dire Whale above) for whatever reason :)
5) Sometimes in a PUG, good players don't push since they don't want to go first ... and it is a PUG so no one might follow ... so you lose :)

... and these are just some of the actions that good players might take all the time ... never mind the semi-mythical "bad".

(There are some bad players out there ... mostly I think because they are learning the game ... folks who don't see the indicator showing they are being cored out from behind while they chase the light mech in front of them ... or whose aim is sub-par ... but I think there are far fewer "bads" than good folks doing not so good things ... at least "not so good" in terms of maximizng win probability ... maybe not it terms of having fun.)

#16 AdamBaines

    Member

  • PipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 1,384 posts

Posted 15 June 2015 - 10:30 AM

View PostSoy, on 15 June 2015 - 09:56 AM, said:

Went out last night in a LRM Dire Wolf, watchin a movie real late just chillin. You know, 4am on a Sunday night/Monday morning. Real serious ****, right.

I lost for every round for 80 minutes straight, and all I could think was. "lol MM actually expects me to make a real DWolf and really try and really win" .........like fffffffffffffffffffff that, if I wanna carry I'll carry..... but if I take some random **** out at the oddest time, then Elo doesn't give a ****, it responds with a "get pumped a dozen rounds in a row, nerd!"

Like... ok, you almost got me game, I almost cared enough to try. Almost. When will this game be able to use some sort of advanced metrics to tell whether or not someone is even trying?

To quote Cyborx... where is the ******* "meters walked per minute" stat, where is the "shots fired per round" stat, when is it gonna take a look at the weapon you're using and say "he's used this weapon 6 times, ever, on any mechs" etc............................... "average damage per killing blow" etc

.....what about "SNEAKINESS" how often are you locked on? Give a stat on that PGI... Oh wow that LRM guy getting carried by teammates sitting in back, such high Elo, such high W:L, wait he hasn't moved in the past 17 rounds in 3 hours. Not targeted once. WTF? Same as gauss wolves in group play, what the **** is that garbage. Nice job walking for 60 secs, sitting behind a rock, and firing 9 times a game. Pro status; nobody even ******* shot at him entire weekend most likely...

Elo would work a lot better if there were some sort of actual stats being used to dictate things besides Win/Loss. Win/Loss means jack **** if you're being carried by people every session, or only pugging.


Interesting thoughts on stats and how they can be used to help determine a players intent in a specific round, and how that could help the MM.

Im really dumb on this as I really dont have time to play much else then sporadic MWO, but what other games use metrics like these to help its match maker?

#17 STEF_

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nocturnal
  • The Nocturnal
  • 5,443 posts
  • Locationmy cockpit

Posted 15 June 2015 - 10:31 AM

Classifing players by elo (so they can amuse matches more, as they, supposedly, are going to play with similar skilled players) I think is something that comes from T-Ball generation.

Never ever played mechwarrior or battletech with a elo system before, and never felt the need to have elo.

Throw elo MM away.

#18 Water Bear

    Member

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

Posted 15 June 2015 - 10:40 AM

View Postcdlord, on 15 June 2015 - 08:54 AM, said:

So I've been thinking of ELO and how it works/affects me. This coincides with another though along the lines of "How many times do I have to lose to get out of "ELO Hell?".

This could be an "It's me, not you." outcome, but I feel it's more like "It's you, not me.". :P

I could suck at this game (which I have compelling evidence that I do not), and ELO puts me with other bads.

I could be awesome and the MM puts me with bads to average out the equation.

I could be in "ELO Hell" where I am above the bads, but below the goods that my ELO places me with.

And please, do NOT take offence at my use of the terms "goods" and "bads", I only use them to differentiate the performance I see from my team and opponents.

Any other musings about ELO? :)


Edit: I realized something and changed my mind.

If you are "above the bads, but below the goods that my ELO places me with," then it sounds like you have converged to your true Elo, up to whatever range the Matchmaker tolerates. I'm not sure I see how that's some kind of hell.

Edited by Water Bear, 15 June 2015 - 10:45 AM.


#19 Mawai

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,495 posts

Posted 15 June 2015 - 10:43 AM

View PostStefka Kerensky, on 15 June 2015 - 10:31 AM, said:

Classifing players by elo (so they can amuse matches more, as they, supposedly, are going to play with similar skilled players) I think is something that comes from T-Ball generation.

Never ever played mechwarrior or battletech with a elo system before, and never felt the need to have elo.

Throw elo MM away.


Life before Elo MM was worse. Sorry, but that is the truth. ROFL stomps were the order of the day ... more frequent than they are now ... part of that was 8 v 8 which is less forgiving of mistakes and part was just a much wider range of balance in matches.

Separated group and solo queues using weight class and Elo matching give better matches than any system they have used to date. Perfect? No, far from it ... but until something better comes along it is what we have and saying "Let's chuck Elo MM because I don't like it" isn't very effective since it is better than what came before and you haven't made any suggestion for a better system.

By the way ... all the systems suggested based on kills/damage done/match point score ... all avoid the question of whether YOUR actions help your team win ... whatever they may be. I have seen a Timberwolf get 1100 damage ... 5 kills, 3 assists ... carry harder you say? Well they got most of that while I spectated them sniping into a melee where we were losing badly and there were only 2 team mates left on our side. The remaining 9 or so opponents finished off the last of our team and then came after the Timberwolf who had been hiding most of the match. I watched him get 4 of the 5 kills on mechs already damaged by his team mates while he sprayed laser fire over and over again ... mostly on the mechs but not on one section ... he was just lucky they were already damaged. I have seen lots of examples of this ... so I am not a fan of ranking systems based on actions that may or may not contribute to the win.


Edit: It takes about 180 damage to core out an assault through the CT. In some ways, I wonder if really high damage numbers are as much indications of poor aim or inability to focus fire down opponents than anything else ...

Edited by Mawai, 15 June 2015 - 10:46 AM.


#20 STEF_

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Nocturnal
  • The Nocturnal
  • 5,443 posts
  • Locationmy cockpit

Posted 15 June 2015 - 10:53 AM

View PostMawai, on 15 June 2015 - 10:43 AM, said:


Life before Elo MM was worse. Sorry, but that is the truth. ROFL stomps were the order of the day ... more frequent than they are now ...


My point is that rolf stomps happen, and if they happen it's not the end of the world.

T-Ball generation guys want elo system because they feel the need to have a possibility.

Instead, imo, my skills improve only when I meet better pilots than me, watching well how they pilot. I don't care if I get stomped.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users