Jump to content

- - - - -

How Exactly Elo Works


24 replies to this topic

#1 Tereva

    Member

  • PipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 120 posts

Posted 17 July 2014 - 06:14 PM

Hi all,

I was leaving a happy life, I finally figure out my style play, tweaked my mechs and I was cruising on the battelfield.
I was killing 3 or 4 times more that I was killed and my W/R was going up like crazy. In over word it was as close to heaven as it could be.

And then suddenly something changed. From one day to another, life become difficult. Now I have to work really hard to keep a K/D and W/R around 1. My guess is that I get bumped to a higher group of players.

I know you cannot check your ELO rating, but do we know ow it is calculated ?

#2 Redshift2k5

    Welcoming Committee

  • PipPipPipPipPipPipPipPipPipPipPip
  • Stone Cold
  • Stone Cold
  • 11,975 posts
  • LocationNewfoundland

Posted 17 July 2014 - 06:21 PM

You begin your Elo journey at 1500. How much you gain or lose per match is a mystery, and it varies with:
  • your team's average Elo(impossible to deduce)
  • the matchmaker's prediction whether you will win or lose
Essentially your Elo is adjusted when the matchmaker failed to predict the outcome- if your team is the predicted winner (higher Elo than the other team) and you win, then the prediction was correct and your Elo doesn't change. If your team had a higher Elo and you lose, then your Elo is adjusted.


A player who's performance increases over time will expect to have their Elo gradually move upwards.

View PostKarl Berg, on 15 July 2014 - 11:03 PM, said:


Hey Cimarb. It's true somewhat that Elo is related to your win/loss ratio. It adjusts based on your predicted chance to win however, and it turns out that this is a really important property. By adjusting in the manner proposed in Elo's algorithm, the system actually converges over time to a value that is not dependent on your win/loss ratio.

Simply put, if your predicted chance to win a game is 100%, and you actually win, your Elo doesn't increase as a result. Put another way, your Elo changes significantly only when you surprise the matchmaker.

The matchmaker is actually training itself. It's making predictions about whether or not you are going to win or loose a game beforehand, and depending on the actual outcome of the game, it's adjusting it's estimation of your skill.

Edited by Redshift2k5, 17 July 2014 - 06:36 PM.


#3 cleghorn6

    Member

  • PipPipPipPipPipPipPip
  • The Patron Saint
  • The Patron Saint
  • 511 posts

Posted 17 July 2014 - 11:50 PM

Redshift has given you answer you wanted but for more information than you will ever need: http://en.wikipedia....o_rating_system (with maths!)

#4 Karamarka

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 809 posts

Posted 18 July 2014 - 12:39 AM

Sounds dodgey to me. (Elo system)

After a 2.2 kd and hundreds of 700+ games i still get mouthbreathers on my team who suicide and make us lose the whole game

edit: yup just lost 3 games in kitfox, 2 in adder, 1 in nova and 1 in cent all in a row, average damage over 7 games = 688dmg what a joke for losing 7 in a row...

Edited by Karamarka, 18 July 2014 - 01:28 AM.


#5 Roughneck45

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Handsome Devil
  • The Handsome Devil
  • 4,452 posts
  • LocationOutreach

Posted 18 July 2014 - 06:59 AM

View PostKaramarka, on 18 July 2014 - 12:39 AM, said:

Sounds dodgey to me. (Elo system)

After a 2.2 kd and hundreds of 700+ games i still get mouthbreathers on my team who suicide and make us lose the whole game

edit: yup just lost 3 games in kitfox, 2 in adder, 1 in nova and 1 in cent all in a row, average damage over 7 games = 688dmg what a joke for losing 7 in a row...

Because it tries to make both teams have a close average ELO.

If your ELO is high and there are not enough high ELO players to group into one game it will balance your ELO with said mouthbreathers.

#6 Bulletsponge0

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Vicious
  • The Vicious
  • 2,942 posts

Posted 18 July 2014 - 11:51 AM

View PostKaramarka, on 18 July 2014 - 12:39 AM, said:

Sounds dodgey to me. (Elo system)

After a 2.2 kd and hundreds of 700+ games i still get mouthbreathers on my team who suicide and make us lose the whole game

edit: yup just lost 3 games in kitfox, 2 in adder, 1 in nova and 1 in cent all in a row, average damage over 7 games = 688dmg what a joke for losing 7 in a row...

highlighted the parts that have nothing at all to do with your Elo

#7 Xeno Phalcon

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • 2,461 posts
  • LocationEvening Ladies

Posted 18 July 2014 - 12:21 PM

ELO is like karma but more complicated and im pretty sure comming back as a cockroach would be preferable to some of the things elo will inflict on you.

After awhile you sort of get a feel for when your going to get a series of stomps, usually 2-3 good wins and I can expect a full on counter roll the next couple of matches, give or take a little luck.

Good example is iv had about three back to back pretty good games (not counting the fact that I spent the last few minutes of the third one under the effect of the mysterious lag bug) If I were to hop in my warhawk and launch right now odds are I would end up in a roll, or at the very least a hard fight followed by a roll.
-----------------

Its a sad truth that would should balance out skills instead controls how likely your team is going to be filled with droolers, the end result is terrible players get carried occasionally and very good ones can get 2-3 guys that do 0-40 dmg. I pity newer players or those that have the capacity to improve their skills, they will always get the bewildered effect of wondering why they did so well one match and the next their entire team got wiped off the face of the planet.

(this also has the effect of the tryhard players raging at the rest of the team when they die, creates a very hostile setting iv seen one guy in particular that crops up in my elo range occasionally that just dogs on everyone, especially if hes losing.)

#8 Redshift2k5

    Welcoming Committee

  • PipPipPipPipPipPipPipPipPipPipPip
  • Stone Cold
  • Stone Cold
  • 11,975 posts
  • LocationNewfoundland

Posted 18 July 2014 - 07:37 PM

View PostKaramarka, on 18 July 2014 - 12:39 AM, said:

Sounds dodgey to me. (Elo system)

After a 2.2 kd and hundreds of 700+ games i still get mouthbreathers on my team who suicide and make us lose the whole game

edit: yup just lost 3 games in kitfox, 2 in adder, 1 in nova and 1 in cent all in a row, average damage over 7 games = 688dmg what a joke for losing 7 in a row...


KD is irrelevant, damage done is irrelevant.

Also you have a different Elo score per weight class, so you could be above average Elo for Assaults but below average Elo for Lights.

Elo is averaged on a per team basis; both teams will have a similar average Elo. This does not mean all your teammates have the same Elo as you.

Your Elo is probably nowhere near as high as you think it is. Work on your fundamentals.

#9 Tereva

    Member

  • PipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 120 posts

Posted 18 July 2014 - 09:22 PM

View PostRedshift2k5, on 18 July 2014 - 07:37 PM, said:

Also you have a different Elo score per weight class, so you could be above average Elo for Assaults but below average Elo for Lights.


Are you sure about that ?
It seems to me that all my chassis were facing stronger players all of the sudden.
I might be wrong of course but I really get the feeling that the change was sudden and all across the board.

PS : I am not complaining about facing stronger players but the change felt really brutal to me

Edited by Tereva, 18 July 2014 - 09:25 PM.


#10 Redshift2k5

    Welcoming Committee

  • PipPipPipPipPipPipPipPipPipPipPip
  • Stone Cold
  • Stone Cold
  • 11,975 posts
  • LocationNewfoundland

Posted 18 July 2014 - 09:24 PM

Yes, different Elo per weight class is confirmed, that's been known since the beginning of Elo implementation.

#11 Lukoi Banacek

    Member

  • PipPipPipPipPipPipPipPipPip
  • WC 2018 Top 12 Qualifier
  • WC 2018 Top 12 Qualifier
  • 4,353 posts

Posted 18 July 2014 - 10:14 PM

Unfortunately, the MM also has valves that allows it to take a wider Elo range of players than is optimal if its at a time where enough players aren't in search mode.

It's all very nebulous and fairly obnoxious. The opacity is what really gives it the bad name that it has, imo. But anyway, learn to not worry about it and just fight hard every match. You'll probably enjoy the game more that way.

#12 Kjudoon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Rage
  • Rage
  • 7,636 posts
  • LocationWisconsin

Posted 18 July 2014 - 10:22 PM

Apparently I've won more than a few significant games against opponents I should never have beaten.


Crap.

#13 GunnyKintaro 01

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 3,072 posts

Posted 18 July 2014 - 10:39 PM

View PostLukoi, on 18 July 2014 - 10:14 PM, said:

Unfortunately, the MM also has valves that allows it to take a wider Elo range of players than is optimal if its at a time where enough players aren't in search mode.

It's all very nebulous and fairly obnoxious. The opacity is what really gives it the bad name that it has, imo. But anyway, learn to not worry about it and just fight hard every match. You'll probably enjoy the game more that way.

Indeed !

#14 TheCaptainJZ

    Member

  • PipPipPipPipPipPipPipPipPip
  • The CyberKnight
  • The CyberKnight
  • 3,655 posts
  • LocationUnited States

Posted 19 July 2014 - 12:18 AM

View PostTereva, on 18 July 2014 - 09:22 PM, said:


Are you sure about that ?
It seems to me that all my chassis were facing stronger players all of the sudden.
I might be wrong of course but I really get the feeling that the change was sudden and all across the board.

PS : I am not complaining about facing stronger players but the change felt really brutal to me

This also happens (facing seemingly stronger opponents) when you play at a different time of day or day of the week. The player dynamics just mix things up, partly due to population at various times. Maybe you normally play when fewer players are on, and so you are matched up against weaker opponents strugging with ping issues. Or something else.

#15 LauLiao

    Member

  • PipPipPipPipPipPipPipPip
  • 1,591 posts
  • LocationCalifornia

Posted 19 July 2014 - 09:34 AM

View PostTereva, on 17 July 2014 - 06:14 PM, said:

Hi all,

I was leaving a happy life,


No... you were leading or living a happy life. If you were leaving a happy life that means you were dying.

#16 Void Angel

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Marauder
  • The Marauder
  • 6,575 posts
  • LocationParanoiaville

Posted 19 July 2014 - 09:54 AM

Allow me to clear up some confusion in the thread:

Elo ranking is a rating system developed by Arpad Elo to match players of equivalent skill against each other in competitive chess.

ELO is the Electric Light Orchestra:
Posted Image

You're welcome.

#17 Kjudoon

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Rage
  • Rage
  • 7,636 posts
  • LocationWisconsin

Posted 19 July 2014 - 10:48 AM



Ode to Elo by ELO.

(when considering the song in this context, It gets even funnier as I listen to the lyrics)

Edited by Kjudoon, 19 July 2014 - 10:49 AM.


#18 Sug

    Member

  • PipPipPipPipPipPipPipPipPip
  • The People's Hero
  • The People
  • 4,626 posts
  • LocationChicago

Posted 19 July 2014 - 11:02 AM

The range for matching up Elo scores is 1400 on a scale of 0 to 2800. That's how it's possible for some of the best to be placed on a team with some of the worst.

Also, it was said earlier that if the MM calculates that you're "supposed" to lose and you do, your Elo score moves very little if at all. So while bad games can be frustrating it's probable that you were "supposed" to get stomped and that game doesn't "count" as far as affecting your Elo too much.

Same can be said for easy wins.

#19 Karamarka

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 809 posts

Posted 19 July 2014 - 06:30 PM

View PostBulletsponge0, on 18 July 2014 - 11:51 AM, said:

highlighted the parts that have nothing at all to do with your Elo



I was just stating that the Elo system and matchmaking in this game is horrible

why if i pull all the weight and just because i'm good, i have to carry every single game? Flawed design

If i play a light or medium mech i am guaranteed to lose because i cant carry like i can in a timberwolf or direwhale.

View PostRedshift2k5, on 18 July 2014 - 07:37 PM, said:


KD is irrelevant, damage done is irrelevant.

Also you have a different Elo score per weight class, so you could be above average Elo for Assaults but below average Elo for Lights.

Elo is averaged on a per team basis; both teams will have a similar average Elo. This does not mean all your teammates have the same Elo as you.

Your Elo is probably nowhere near as high as you think it is. Work on your fundamentals.


Wow another person saying L2P , is that what you tell yourself when you cant break 300damage every game?

heh

Pretty much Russ and others on the 15th July podcast stated how it works, and from deduction

1 pro and 5 noobs, 6 average on Team A

vs

12 average on Team B

Yet the matchmaker due to bell curve of new players and average Elo when starting cant deduct if they are new or average, so really half the time its massive noob (trial/poor mech lab mechs/cadets) on either team weighing them disproportionately hence, you get 3-4 people on your team suiciding and another 3-4 doing less than 150 damage.

ggclose

Edited by Karamarka, 19 July 2014 - 06:34 PM.


#20 EgoSlayer

    Member

  • PipPipPipPipPipPipPipPip
  • Wrath
  • Wrath
  • 1,909 posts
  • Location[REDACTED]

Posted 19 July 2014 - 10:30 PM

View PostKaramarka, on 19 July 2014 - 06:30 PM, said:



<snip>


Wow another person saying L2P , is that what you tell yourself when you cant break 300damage every game?

heh

Pretty much Russ and others on the 15th July podcast stated how it works, and from deduction

1 pro and 5 noobs, 6 average on Team A

vs

12 average on Team B

Yet the matchmaker due to bell curve of new players and average Elo when starting cant deduct if they are new or average, so really half the time its massive noob (trial/poor mech lab mechs/cadets) on either team weighing them disproportionately hence, you get 3-4 people on your team suiciding and another 3-4 doing less than 150 damage.

ggclose


Not sure what podcast you are getting that deduction from, because Russ said in the NGNG podcast that the majority of the solo queue games have a variance of 50 Elo. As in the lowest Elo on the team is 50 or less than the highest Elo on the team. Not a team average of 1 pro and 6 noobs - the only way that happens is in a player created group.

Group queue he didn't give numbers, but did say it's worse because the pools are smaller and there are team averages. So right now group queue is actually having worse match-ups than solo.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users