Jump to content

Proof That Elo Is Broken


99 replies to this topic

#61 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 28 January 2015 - 09:00 AM

View PostSarlic, on 28 January 2015 - 08:58 AM, said:

But the TEAM is one huge factor, not you in your single mech scoring over 400 damage.


It is a huge factor but it is not the ONLY factor.

IF team was the deciding factor, then your win/loss ratio would never be able to exceed 50% by any significant margin.

#62 Axeface

    Member

  • PipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 655 posts

Posted 28 January 2015 - 09:01 AM

I for one feel like i'm being singled out by the MM and have been for weeks, I'm really getting tired of it. Each day I play I usually end up rage quitting and throwing out a few insults over chat. I have a positive win loss at 2,084 / 1,730, all pugged. And i think the MM has decided to try to make me quit the game.

If it does single people out, make... it.... stop.

Edited by Axeface, 28 January 2015 - 09:02 AM.


#63 Joseph Mallan

    ForumWarrior

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

Posted 28 January 2015 - 09:04 AM

View PostMister Blastman, on 28 January 2015 - 09:00 AM, said:

It is a huge factor but it is not the ONLY factor.

IF team was the deciding factor, then your win/loss ratio would never be able to exceed 50% by any significant margin.

Not quite true. If you are somehow the only good player on a team, You can only carry so much. And will not have a winning record. If the team has 3-4 good players it can help you to have a good W/L record. One player cannot carry a team by themselves.

#64 Dino Might

    Member

  • PipPipPipPipPipPipPipPipPip
  • Little Helper
  • Little Helper
  • 2,030 posts

Posted 28 January 2015 - 09:12 AM

View PostLordKnightFandragon, on 28 January 2015 - 06:49 AM, said:


That is everything wrong with every Matchmaker EVER......it shouldnt take skill into account that way. It just hurts the players are play well....its like, you play well? Here is your reward.....your good self alongside 11 other horrible scrubs!!! You now face LORDS for the next 10 games!!!! HAVE FUN!!!!


That's everything right with matchmaker. Why would you want to stay playing against easy opponents? That would become very not fun very quickly. I like when I drop against difficult opponents/teams regularly, because it means I'm working my way up to their level. I get some satisfaction that I am getting better at the game. Playing those opponents makes me play better, because I have to make fewer mistakes to do well.

You'll never improve if you don't fight more difficult opponents.

#65 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 28 January 2015 - 09:41 AM

View PostJoseph Mallan, on 28 January 2015 - 09:04 AM, said:

Not quite true. If you are somehow the only good player on a team, You can only carry so much. And will not have a winning record. If the team has 3-4 good players it can help you to have a good W/L record. One player cannot carry a team by themselves.


Then how come it is possible to have a 60+% win ratio in the PUG queue playing solo by yourself, all the time? Over a statistically significant portion of games, too--say, upwards of 2000?

Edited by Mister Blastman, 28 January 2015 - 09:41 AM.


#66 LordKnightFandragon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 7,239 posts

Posted 28 January 2015 - 10:45 AM

View PostLordSkippy, on 28 January 2015 - 07:05 AM, said:

Elo and the matchmaker are not supposed to keep you at a 50/50 win/loss ratio. It's a myth that just won't die. MM uses Elo to try to make teams reasonably within the same skill bracket. NOTE: that's TRIES, it can't always do that, and seems to fail at it quite often.

At no point does the MM look at your win/loss ratio and say "hmmm, Joe needs a loss to keep at 50/50, so I'll put him on the derp team." It also won't put you on the "good" team just because you have more losses than wins. It looks at your Elo and tries to put you in a team of similar Elo players and then match your team with a similar team. That's it. No forced 50/50.

Now, with that said, for average players, a roughly 50/50 win/loss ratio over a large number of games is a side effect of an Elo rating system. As the system tries to place you at the proper skill levels, you can expect to win roughly half your games. You can also expect win and loss streaks, when you straddle bracket groups. Again, that's roughly 50/50, not exactly 50/50. 48/52 is roughly 50/50. 44/56 is reasonably close to 50/50.



Yeah, and ultimately, some players are just cursed to never get over the hump...I seem to have been one of the chosen ones cursed to forever remain below 50/50. even in games where I contribute pretty well its sometimes not enough. And as for the whole L2P or carry idea? No....I dont carry, I aim to contribute my share of the damage and kills. I expect that the other 11 guys on my team will have skill and brains enough to do the same.

But, I get put on the side with hte guy stuck in sniper mode, who walks into the fire pit on terra therma and commences to get a mostly healthy Timber Wolf annihilated by a mostly dead Blow-Job 1X....Those are the kinds of players I get put with........while the game was an 11/11 and we ended losing 11/12, the point is, that was our game to lose, and because of 1 idiot damn derp to stupid to realize how to even move the mech around, we end up losing. And the guy was no Trial mech noob, he had a TBR-S all tricked out and laser vomiting.....he just couldnt aim and had no sense of skill or direction. He literally walked around the map in max zoom mode........

#67 Joseph Mallan

    ForumWarrior

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

Posted 28 January 2015 - 11:10 AM

View PostMister Blastman, on 28 January 2015 - 09:41 AM, said:


Then how come it is possible to have a 60+% win ratio in the PUG queue playing solo by yourself, all the time? Over a statistically significant portion of games, too--say, upwards of 2000?

Same reason I used to have it. Your team was the better team 60% of the time.

#68 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 28 January 2015 - 11:18 AM

View PostJoseph Mallan, on 28 January 2015 - 11:10 AM, said:

Same reason I used to have it. Your team was the better team 60% of the time.


Impossible. The matchmaker seeks to put you at 50% as far as I can tell. If it is just a luck of the draw then after 2000 or so games you'd be at 50%, not continually climbing past 60%.

You have control over the outcome. You don't have total control, but you absolutely can have a big influence by yourself if you try hard enough.

(if you had total control then you'd see people winning 100% of the time but this isn't the case, either, in pugs)

#69 shadowrwolf

    Member

  • PipPipPip
  • Mercenary
  • 72 posts

Posted 28 January 2015 - 11:20 AM

Proof ELO is a giant steaming pile of ****. all you have to do is look at the stupid math system they use for these "Events"

#70 Joseph Mallan

    ForumWarrior

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

Posted 28 January 2015 - 11:22 AM

View PostMister Blastman, on 28 January 2015 - 11:18 AM, said:


Impossible. The matchmaker seeks to put you at 50% as far as I can tell. If it is just a luck of the draw then after 2000 or so games you'd be at 50%, not continually climbing past 60%.

You have NO control over the outcome. You don't have total control, but you absolutely can have a big influence by yourself if you try hard enough.

(if you had total control then you'd see people winning 100% of the time but this isn't the case, either, in pugs)
Fixed what was wrong with your statement. You do not have control over the win of a game you can contribute to it
but you cannot make it happen alone unless the stars align and the angels sing.

#71 Sarlic

    Member

  • PipPipPipPipPipPipPipPipPip
  • Hearing Impaired
  • Hearing Impaired
  • 4,519 posts
  • LocationEurope

Posted 28 January 2015 - 11:26 AM

View PostWintersdark, on 28 January 2015 - 08:46 AM, said:


Yeah, like I said above, the MM doesn't care about you. It doesn't arrange things so that you'll lose, in fact its goal with every match is to get as close to even odds as it can, and then adjust your Elo as appropriately as possible after.

It doesn't ever "make you lose", because you're not a special snowflake. You're just one of 24 players its cramming into a match.


True. Well i can always do my best i guess ;).


#72 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 28 January 2015 - 11:28 AM

View PostJoseph Mallan, on 28 January 2015 - 11:22 AM, said:

Fixed what was wrong with your statement. You do not have control over the win of a game you can contribute to it
but you cannot make it happen alone unless the stars align and the angels sing.


Statistics disagree with you. Angels sing would be a deviation of +/- .5 - 5% at best. These odds are defying even that possibility.

Besides, if we did a random walk we'd expect to see players with 75% - 90% win ratios and be average to below average players after 1000 games. We do not, however. It doesn't happen. Not in the pug queue at least.

#73 LordKnightFandragon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 7,239 posts

Posted 28 January 2015 - 11:31 AM

View PostDino Might, on 28 January 2015 - 09:12 AM, said:


That's everything right with matchmaker. Why would you want to stay playing against easy opponents? That would become very not fun very quickly. I like when I drop against difficult opponents/teams regularly, because it means I'm working my way up to their level. I get some satisfaction that I am getting better at the game. Playing those opponents makes me play better, because I have to make fewer mistakes to do well.

You'll never improve if you don't fight more difficult opponents.



You'll also never improve if your always on the side that gets mobbed in 5 minutes flat and you get to fire maybe a shot or two and the game ends.....sure, you can try to learn what just happened, but besides you realized ur side got absolutely destroyed, whats to learn?

I dont want harder players, rather players of similar skill and elo........

#74 PurpleNinja

    Member

  • PipPipPipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 2,097 posts
  • LocationMIA

Posted 28 January 2015 - 11:31 AM

Nope, Elo is dead, he passed away on november 5, 1992.

#75 Kookieslayer

    Member

  • PipPipPip
  • Bad Company
  • Bad Company
  • 50 posts

Posted 28 January 2015 - 11:34 AM

All I read was "I am bad it must be ELO's fault" .... Don't blame match making. If you don't like your performances, find some lancemates and practice with them.

Teamwork is OP

#76 Joseph Mallan

    ForumWarrior

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

Posted 28 January 2015 - 11:34 AM

View PostMister Blastman, on 28 January 2015 - 11:28 AM, said:

Statistics disagree with you. Angels sing would be a deviation of +/- .5 - 5% at best. These odds are defying even that possibility.

Besides, if we did a random walk we'd expect to see players with 75% - 90% win ratios and be average to below average players after 1000 games. We do not, however. It doesn't happen. Not in the pug queue at least.
Bet the top tier players are up around that level. And no the statistics do not disagree with me. The best team wins every match. There are many a story of players with good performances and being on the losing team. An individuals performance can Help a victory occur, but it is not a 50/50 proposal. I've been a 62% winning player, Thanks to good teams. I am a 48% winner in PUGs. The better team wins.

#77 LordKnightFandragon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 7,239 posts

Posted 28 January 2015 - 11:37 AM

View PostKookieslayer, on 28 January 2015 - 11:34 AM, said:

All I read was "I am bad it must be ELO's fault" .... Don't blame match making. If you don't like your performances, find some lancemates and practice with them.

Teamwork is OP



****, ive gone with groups and we simply get paired against guys who are even better yet.....so even in groups, same result...loss after loss...

#78 Almond Brown

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 5,851 posts

Posted 28 January 2015 - 11:39 AM

View PostApocryph0n, on 28 January 2015 - 06:04 AM, said:

Also: To qualify all you have to do is hit 7 different enemies for 1 damage. (If this is a rant towards the tournament)

stuff


I think it only fair to note that those 7 Mechs must ALL die in order to get those Assist points. That is what is wrong with most in this challenge. They do not KNOW how it actually works! (doh!)

If they don't die then those 7 wicked points of damage done are /15 and = (no MC for you!) LOL!

View PostMawai, on 28 January 2015 - 06:32 AM, said:

This appears to be a record from the current mercenary challenge where you need 130 points in a match to earn 20MC.

You do NOT need a win to qualify.

The formula is Kills * 10 + assists * 20 + (damage-team damage)/15

6 assists and 150 damage will net 130 points.

AND ... this has NOTHING to do with Elo.

Elo is supposed to be related to creating matches between players with roughly equivalent skills. However, it is actually based on creating matches between players with roughly equal chances to win. There are multiple ways to contribute to winning a match. Damage/kills/assists is one way. Organizing your pug group can be another. Team work is OP so a player that does nothing but can get the team moving together and organized may also end up with a higher Elo.

Anyway, as far as this post goes ... unless you add the data for whether you won or lost each match (it isn't there) ... then there is no relevance of this data to Elo.



Winning sure will increase the chances for those assist points to count though. ;)

Edited by Almond Brown, 28 January 2015 - 11:38 AM.


#79 Axeface

    Member

  • PipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 655 posts

Posted 28 January 2015 - 11:40 AM

View PostKookieslayer, on 28 January 2015 - 11:34 AM, said:

All I read was "I am bad it must be ELO's fault" .... Don't blame match making. If you don't like your performances, find some lancemates and practice with them.

Teamwork is OP


What the hell has that got to do with pug drops?

#80 Joseph Mallan

    ForumWarrior

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

Posted 28 January 2015 - 11:42 AM

View PostAxeface, on 28 January 2015 - 11:40 AM, said:


What the hell has that got to do with pug drops?

You will likely find better team work with regular lance mates than With random PUGs. It is PUGs greatest weakness at present. A hampered ability to work together.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users