Jump to content

Why Elo Doesn't Work Here


633 replies to this topic

#281 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:10 AM

View PostReXspec, on 24 January 2014 - 12:07 AM, said:

Battle Value acting as a base for matchmaking would work better... and would aid in curbing lob-sided matches that are currently determined by weight class.

BV is yet another tool that the matchmaker could use to improve matches, but it would not make a good base for matchmaking.

Do you really think that a 10,000 match expert and a 0 match noob, piloting identical Atlases, is a fair fight?

No, of course not. That's why BV is insufficient as the basis for matchmaking.

#282 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 09:12 AM

View PostNik Van Rhijn, on 24 January 2014 - 09:01 AM, said:

Battlevalue is not a horrible idea, it improves the parameters that the Mm has available in setting up the match. Skill will count for more even then. Two teams equally matched in skill, one with fully elited min/maxed mechs, the other in "stock".
Which would you expect to win the vast majority of the time?
Contrary to some people, the purpose of playing games is to have fun, not neccessarily to win at all costsStrangely, many people find getting killed rapidly most/all of the time to be the opposite. I have no problems in losing if I felt the team had a chance.
For some people the journey is more important than the destination.


What I said before was, sometimes a guy is better in a medium then he is in an assault mech. How do you account for that then?

It makes sense now why they didn't use weight limits like previous games. Because I bet the previous game didn't have amatchmaker based on some sort of skill rating.

ANd i guess we are assuming only skirmish mode, because I don't know how an all assault mech team is going to beat a smarter team in conquest that outcaps them....


Which is why most people will find out, since we are already matching on "skill" weight limits won't make much of a difference for them and they still be whining and crying about how unfair the game is...

Maybe I'm wrong we will have to wait and see.

Edited by RichAC, 24 January 2014 - 09:13 AM.


#283 RussianWolf

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 2,097 posts
  • LocationWV

Posted 24 January 2014 - 09:13 AM

View PostRoadkill, on 24 January 2014 - 09:02 AM, said:



People just need to forget that Elo exists. It's just a rating system. All of the perceived problems that people have with MWO matches come from the matchmaker, not from Elo ratings.


Yes, The matchmaker that uses ELO. Thus the original premise of the thread, ELO isn't working here. If the application of ELO were working then you'd see everyone with the magic 50/50, but we don't. ELO can't accurately rate in this environment and the matchmaker can't make even matches because its using flawed data as its premise.

#284 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 09:14 AM

View PostRoadkill, on 24 January 2014 - 09:10 AM, said:

BV is yet another tool that the matchmaker could use to improve matches, but it would not make a good base for matchmaking.

Do you really think that a 10,000 match expert and a 0 match noob, piloting identical Atlases, is a fair fight?

No, of course not. That's why BV is insufficient as the basis for matchmaking.



You just explained the opposite. reread what you said. I can't stop laughing at this sandbagger right now.
"Piloting the exact same atlases" ok so why would battlevalue matter here then? umm....

What you described is how a matchmaker based on ELO/rating alone would be sufficient. I think you mispoke here bud.

#285 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:14 AM

View PostKhobai, on 24 January 2014 - 08:06 AM, said:

Battlevalue could work the same way as elo, and have a dynamic value. For example, the more games a particular mech or weapon wins, the higher its battlevalue becomes. So youd get battlevalue that autoadjusts over time based how successful mechs/weapons are in the game. Itd be pretty difficult if not impossible to game that type of system.

In other words, use Elo to rate weapons and Mechs.

No really, that's what you just suggested.

And here's the kicker... it would work. If players, Mechs, and weapons all had independent Elo ratings, the matchmaker could use all of that data to set up very balanced matches.

But even then, you'd still get 12-0 ROFLstomps. Games like MWO are very momentum dependent. 12 vs 12 might be completely even. But when that first Mech dies to a lucky head shot, 12 vs 11 is suddenly really unbalanced and the match quickly cascades into a lopsided victory.

That's just the nature of games like MWO. People have to just accept that ROFLstomps happen, even when the match was balanced at the start.

#286 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 09:17 AM

View PostRussianWolf, on 24 January 2014 - 09:13 AM, said:


Yes, The matchmaker that uses ELO. Thus the original premise of the thread, ELO isn't working here. If the application of ELO were working then you'd see everyone with the magic 50/50, but we don't. ELO can't accurately rate in this environment and the matchmaker can't make even matches because its using flawed data as its premise.



Yes there really is no difference. Why he keeps arguing its the matchmakers fautl and not the ELO whent they go together is mind boggling.

BUt what is your W/L Russian? Mine is bettween 1.00-1.30 for all matches. So that seems fine to me.

Our W/L ratio is not the issue here. All our other stats are, and that is what PGI and most of you posters are not understanding. You are never going to satisfy the community, by matching them up according to something they don't care about.....

1. Because its a random team game.

2. Because the game is based around cbills as gauging a players worth.

#287 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:19 AM

View PostRussianWolf, on 24 January 2014 - 09:13 AM, said:

Yes, The matchmaker that uses ELO. Thus the original premise of the thread, ELO isn't working here. If the application of ELO were working then you'd see everyone with the magic 50/50, but we don't. ELO can't accurately rate in this environment and the matchmaker can't make even matches because its using flawed data as its premise.

*sigh*

No. Elo is working fine. If you believe that something's wrong, it's the matchmaker that isn't working.

Elo does not give everyone a 50% chance to win. Elo just rates everyone's skill. And it's doing that just fine. The matchmaker uses Elo ratings to try to set up matches so that everyone has a 50% chance to win.

If that's not happening, it's because the matchmaker isn't working. It's not because Elo isn't working. Elo is a proven system. It works. Stop trying to say that it isn't working because you're just flat-out wrong.

Blame the matchmaker if you think something's wrong, because that's where the blame belongs.

/deadhorse

#288 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:22 AM

View PostRichAC, on 24 January 2014 - 09:14 AM, said:

You just explained the opposite. reread what you said. I can't stop laughing at this sandbagger right now.
"Piloting the exact same atlases" ok so why would battlevalue matter here then? umm....

What you described is how a matchmaker based on ELO/rating alone would be sufficient. I think you mispoke here bud.

No, I said it right. He was saying that BV should be the basis for the matchmaker, and I showed why that wouldn't work.

Two identical Atlases have identical BVs, do they not? So if the matchmaker was based on BV, that would be a fair fight?

No, it wouldn't. Player skill - i.e. Elo rating - needs to be the basis for the system. BV can be layered on top of that, but in games like MWO skill is more important than equipment. So player skill should be the key component of the matchmaker with other things like equipment (i.e. BV) layered on top of it.

#289 Joseph Mallan

    ForumWarrior

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

Posted 24 January 2014 - 09:22 AM

View PostRichAC, on 24 January 2014 - 09:14 AM, said:



You just explained the opposite. reread what you said. I can't stop laughing at this sandbagger right now.
"Piloting the exact same atlases" ok so why would battlevalue matter here then? umm....

What you described is how a matchmaker based on ELO/rating alone would be sufficient. I think you mispoke here bud.

Actually BV compensated for Pilot and Gunnery differences as well. So identical Atlases would have different BV according to if a Pilot was a 3/4 or a 1/2 P/G.

Edited by Joseph Mallan, 24 January 2014 - 09:23 AM.


#290 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 09:24 AM

View PostRoadkill, on 24 January 2014 - 09:22 AM, said:

No, I said it right. He was saying that BV should be the basis for the matchmaker, and I showed why that wouldn't work.

Two identical Atlases have identical BVs, do they not? So if the matchmaker was based on BV, that would be a fair fight?

No, it wouldn't. Player skill - i.e. Elo rating - needs to be the basis for the system. BV can be layered on top of that, but in games like MWO skill is more important than equipment. So player skill should be the key component of the matchmaker with other things like equipment (i.e. BV) layered on top of it.


Ok i'm corrected, and I agree with you there.


But I still feel
Our W/L ratio is not the issue here. All our other stats are, and that is what PGI and most of you posters are not understanding. You are never going to satisfy the community, by matching them up according to something they don't care about.....

1. Because its a random team game.

2. Because the game is based around cbills as gauging a players worth.

Edited by RichAC, 24 January 2014 - 09:25 AM.


#291 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:27 AM

View PostRichAC, on 24 January 2014 - 09:17 AM, said:

Yes there really is no difference. Why he keeps arguing its the matchmakers fautl and not the ELO whent they go together is mind boggling.

What's mind boggling is that you keep arguing about stats, yet you can't acknowledge that Elo is just another stat.

Elo ratings aren't the problem. They're just data that can be used properly or used poorly.

The matchmaker is what uses that data to create matches. If you think that the matches aren't being set up properly, you should be blaming the matchmaker for doing a poor job. Blaming Elo - which is just data being input into the system - is silly.

You keep suggesting match score. Match score is just data, too. I'm sure someone could create a matchmaker that could use match score to create balanced matches. But they could also create a matchmaker that screwed it up. Would that be the fault of match score or the screwed up matchmaker?

Don't blame the data. Facts are facts. Blame the tool that's using (or misusing) the data.

#292 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 09:33 AM

View PostRichAC, on 24 January 2014 - 09:24 AM, said:

But I still feel
Our W/L ratio is not the issue here. All our other stats are, and that is what PGI and most of you posters are not understanding. You are never going to satisfy the community, by matching them up according to something they don't care about.....

1. Because its a random team game.

2. Because the game is based around cbills as gauging a players worth.

I agree with you on this. I'd like to see the matchmaker using a LOT more data to create matches. Elo by itself isn't sufficient unless you have a gazillion Elo ratings for every Mech that you pilot. (Which would take forever to stabilize, so it's not a reasonable approach.)

There's got to be some way to use all of the data - Elo, kill/death ratio, match score, damage, whatever - to help balance matches.

But that's beyond me. I've only ever set up simple systems that just used Elo. For me, those were hard enough to get working.

View PostJoseph Mallan, on 24 January 2014 - 09:22 AM, said:

Actually BV compensated for Pilot and Gunnery differences as well. So identical Atlases would have different BV according to if a Pilot was a 3/4 or a 1/2 P/G.

I think that depends on which BV you're talking about, doesn't it? I know there were at least 2 different ones over the years, and I think there were 3. The one I'm most familiar with (circa 1995-1998) didn't include P/G that I remember.

#293 Joseph Mallan

    ForumWarrior

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

Posted 24 January 2014 - 09:34 AM

View PostRoadkill, on 24 January 2014 - 09:33 AM, said:

I agree with you on this. I'd like to see the matchmaker using a LOT more data to create matches. Elo by itself isn't sufficient unless you have a gazillion Elo ratings for every Mech that you pilot. (Which would take forever to stabilize, so it's not a reasonable approach.)

There's got to be some way to use all of the data - Elo, kill/death ratio, match score, damage, whatever - to help balance matches.

But that's beyond me. I've only ever set up simple systems that just used Elo. For me, those were hard enough to get working.


I think that depends on which BV you're talking about, doesn't it? I know there were at least 2 different ones over the years, and I think there were 3. The one I'm most familiar with (circa 1995-1998) didn't include P/G that I remember.

To the best of my memory the BV of a mech is for an Average P/G... BV 2 was introduced as I was hitting my retirement age from TT, so I didn't use it.

#294 RussianWolf

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 2,097 posts
  • LocationWV

Posted 24 January 2014 - 09:42 AM

View PostRoadkill, on 24 January 2014 - 09:19 AM, said:

*sigh*

No. Elo is working fine. If you believe that something's wrong, it's the matchmaker that isn't working.

Elo does not give everyone a 50% chance to win. Elo just rates everyone's skill. And it's doing that just fine. The matchmaker uses Elo ratings to try to set up matches so that everyone has a 50% chance to win.

If that's not happening, it's because the matchmaker isn't working. It's not because Elo isn't working. Elo is a proven system. It works. Stop trying to say that it isn't working because you're just flat-out wrong.

Blame the matchmaker if you think something's wrong, because that's where the blame belongs.

/deadhorse
How can ELO rate my skill if I'm only 1/12th of my team?

Example. Lets say we were to set up a hypothetical new account. We drop in random pug matches with this account. But instead of fighting we simply run off to a quiet corner and hide for the duration, never DCing. Hypothetically, the enemy never finds this account and it never fires a shot. Now the 11 other members of the team fight it out with the enemy and sometimes they win and sometimes they lose.

This account would then have an ELO rating of x while having demonstrated no skill what-so-ever.

How has ELO rated this account? How can it?

ELO was not designed for this environment, so any data it generates is flawed. Using the data that ELO generates in any way will then give flawed results as well.

So if the ELO data is flawed, and the matchmaker uses that data, then the resulting matches will be flawed as well.

Does the ELO system work? yes. Does ELO work here as it has been applied? NO!

#295 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 09:52 AM

View PostRussianWolf, on 24 January 2014 - 09:42 AM, said:

How can ELO rate my skill if I'm only 1/12th of my team?

Example. Lets say we were to set up a hypothetical new account. We drop in random pug matches with this account. But instead of fighting we simply run off to a quiet corner and hide for the duration, never DCing. Hypothetically, the enemy never finds this account and it never fires a shot. Now the 11 other members of the team fight it out with the enemy and sometimes they win and sometimes they lose.

This account would then have an ELO rating of x while having demonstrated no skill what-so-ever.

How has ELO rated this account? How can it?

ELO was not designed for this environment, so any data it generates is flawed. Using the data that ELO generates in any way will then give flawed results as well.

So if the ELO data is flawed, and the matchmaker uses that data, then the resulting matches will be flawed as well.

Does the ELO system work? yes. Does ELO work here as it has been applied? NO!



You still haven't posted your W/L ratio bud.



@Roadkill,
I think all that would be different is they would use match score compared to other players on team + W/L to = ELO. Its only one more factor....not dozens...

or they can just rank people based on cbills earned period. sweet and simple.

The problem many have with this is they feel match score doesn't really reflect a players worth. Well thats been the age old debate over this game. They give more cbills for dmg done more then everything else combined, and no real reward at all for winning. But apparenlty thats how the majority players like it.

Since we have Skirmish now, for people who prefer damage and kills to matter more, they can even tweak the cbills given for the other game modes if they want. I still don't see how people can complain if they give a bundle for winning a match at least? Especially now that we have skirmish, caps and especially WINS should be worth a hell of alot more then they are now.

Edited by RichAC, 24 January 2014 - 10:02 AM.


#296 Fate 6

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 2,466 posts

Posted 24 January 2014 - 09:53 AM

View PostKhobai, on 24 January 2014 - 08:06 AM, said:


Battlevalue could work the same way as elo, and have a dynamic value. For example, the more games a particular mech or weapon wins, the higher its battlevalue becomes. So youd get battlevalue that autoadjusts over time based how successful mechs/weapons are in the game. Itd be pretty difficult if not impossible to game that type of system.

^
It would function sort of like the system that MW:T has, where your mech and weapons have battlevalue, and then the pilot is a %modifier. Good pilots in MWO will increase the battlevalue of their mechs more. Of course, you would have a different pilot value for each weight class.

And that brings up another issue - it takes many games to build a true Elo rating even in a system that works well. We have different ratings for each weight class, which means it takes at least 4x longer for matchmaking to be able to place us correctly.

#297 Joseph Mallan

    ForumWarrior

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

Posted 24 January 2014 - 09:55 AM

Mine is 58.9% Rich.

2,566/1,789

I lost 0.4% over the last 3 months!

Edited by Joseph Mallan, 24 January 2014 - 09:56 AM.


#298 RussianWolf

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 2,097 posts
  • LocationWV

Posted 24 January 2014 - 10:02 AM

Kills / Death 3,143 / 3,570 C-Bills 47,429,833 Experience Points 3,972,789 Wins / Losses 2,523 / 2,744 Kill / Death Ratio 0.88 Accumulative C-Bills Per Match 113,233.15 Avg. XP Per Match 754.28
As of today

#299 RichAC

    Member

  • PipPipPipPipPipPipPip
  • 661 posts

Posted 24 January 2014 - 10:06 AM

View PostRussianWolf, on 24 January 2014 - 10:02 AM, said:

Kills / Death 3,143 / 3,570 C-Bills 47,429,833 Experience Points 3,972,789 Wins / Losses 2,523 / 2,744 Kill / Death Ratio 0.88 Accumulative C-Bills Per Match 113,233.15 Avg. XP Per Match 754.28
As of today



ok so your a little below 1.00 but its really not that bad at all. Could be alot of factors. Alot of people get their ratio higher, by playing their best mech over and over again. Some people don't care about their stats and have dozens of mechs they constantly change for fun.

A little rule i have for myself. Is only change mechs if I die without a kill. Otherwise i see how long i can keep the streak going. and I do have my goto mechs, like my ctf or atlas or shadowhawk if I feel i'm getting stomped too much.

my ratios go down when I start trying out new mechs, but my goal is keep at 1.00


As far as battlevalue goes, I remember being told assault mechs are harder to use for new players. I know alot of guys better in shadowhawks then any assault they play. I don't know how PGI could make that work. talk about complicated.

The only value of players people care about, is how much they get paid in cbills. And the only math that matters is the scoreboard at the end of a match. Otherwise they better revamp this whole game from the ground up

Edited by RichAC, 24 January 2014 - 10:17 AM.


#300 Roadkill

    Member

  • PipPipPipPipPipPipPipPipPip
  • 3,610 posts

Posted 24 January 2014 - 10:09 AM

View PostRussianWolf, on 24 January 2014 - 09:42 AM, said:

How can ELO rate my skill if I'm only 1/12th of my team?

15 pages into this thread and you still don't understand?

It's been explained over and over backed up by enough statistics to make my head hurt. And I was a Math major. In statistics. Go back and re-read one of those explanations.





22 user(s) are reading this topic

0 members, 22 guests, 0 anonymous users