38 replies to this topic

[Devlin Pierce]

Is there a way to see what ELO bracket you are placed in?

[OneEyed Jack]

ScarecrowES, on 06 July 2014 - 03:25 PM, said:

http://mwomercs.com/...302-02-jul-2014

• The matchmaker will attempt to match based on both Elo and weight class composition (3’s)
  
• Over time, the matchmaker will relax both Elo and weight class composition constraints. In the case of weight class composition, the matchmaker guarantees both teams end up matching.

The statements imply a player-for-player match for both points of composition.

I think you're maybe forcing your own perspective on the words. No where in there does it say "individual Elo", and it's been attempting (albeit poorly) to balance team Elo for a long time.

It's possible I'm wrong and you're right, but I don't believe that those statements mean what you think they do.

[ScarecrowES]

OneEyed Jack, on 06 July 2014 - 03:42 PM, said:

I think you're maybe forcing your own perspective on the words. No where in there does it say "individual Elo", and it's been attempting (albeit poorly) to balance team Elo for a long time.

It's possible I'm wrong and you're right, but I don't believe that those statements mean what you think they do.

It really doesn't matter, either way. Outside of organized competitive play between fixed teams, Elo is a meaningless stat. It has minimal impact on actual team composition but has ZERO reflection of actual player skill. It might as well be ignored entirely in matchmaking.

[Ghogiel]

ScarecrowES, on 06 July 2014 - 03:46 PM, said:

It really doesn't matter, either way. Outside of organized competitive play between fixed teams, Elo is a meaningless stat. It has minimal impact on actual team composition but has ZERO reflection of actual player skill. It might as well be ignored entirely in matchmaking.

I bet 100cbill that the best players in the game have high Elo.

[ScarecrowES]

Ghogiel, on 06 July 2014 - 03:51 PM, said:

I bet 100cbill that the best players in the game have high Elo.

I would bet you the same that the "best" players in the game also play in fixed teams vastly more often than not.

Put the same players in nothing but pug drops for a year, and their w/l ratio will remain 50% +/- 5% (give or take) for the year, and their Elo will drop accordingly. That's statistics 101.

Also, how would you even define "best" for this game? Given that it has no legitimate stat tracking system to see what players are actually doing in-game to positively or negatively impact their teams, there's no actual and impartial way to judge player skill in MWO.

Edited by ScarecrowES, 06 July 2014 - 03:56 PM.

[Ghogiel]

ScarecrowES, on 06 July 2014 - 03:54 PM, said:

I would bet you the same that the "best" players in the game also play in fixed teams vastly more often than not.

Put the same players in nothing but pug drops for a year, and their w/l ratio will remain 50% +/- 5% (give or take) for the year, and their Elo will drop accordingly. That's statistics 101.

Also, how would you even define "best" for this game? Given that it has no legitimate stat tracking system to see what players are actually doing in-game to positively or negatively impact their teams, there's no actual and impartial way to judge player skill in MWO.

All the good players I know will be pulling much higher W/L than 50% +/-5% while solo including myself. And universally the advice to others is "carry harder".


And yeah there is an actual impartial way to see who is better. There are lobbies. You can 1v1, 4v4, 12v12 to see who is actually better irrespective of Elo ratings.

[Sandpit]

Devlin Pierce, on 06 July 2014 - 03:39 PM, said:

Is there a way to see what ELO bracket you are placed in?

no

some people at the top of the pile have an educated guess as to what range their Elo is in but noone knows what their Elo is

[ScarecrowES]

Ghogiel, on 06 July 2014 - 04:01 PM, said:

All the good players I know will be pulling much higher W/L than 50% +/-5% while solo including myself. And universally the advice to others is "carry harder".


And yeah there is an actual impartial way to see who is better. There are lobbies. You can 1v1, 4v4, 12v12 to see who is actually better irrespective of Elo ratings.

If you managed better than say 55% w/l doing nothing but pugging solo, not only would it be unlikely, but also statistically meaningless. It would merely serve as an offset for similarly skilled players whose ratio would fall below 45% accordingly. Nothing more.

As to the second point... no that isn't actual or impartial. By definition you'd only be showing relative skill on a composition per composition basis... which is also statistically meaningless. "I beat that guy when I dropped with 3 of my friends one time" isn't a useful measuring tool for matchmaking.

You might want to check out how games like Battlefield and other modern multiplayer games do skill ranking to get an idea of how utterly pointless elo is.

[Ghogiel]

ScarecrowES, on 06 July 2014 - 04:19 PM, said:

If you managed better than say 55% w/l doing nothing but pugging solo, not only would it be unlikely, but also statistically meaningless. It would merely serve as an offset for similarly skilled players whose ratio would fall below 45% accordingly. Nothing more.

Pretty much every good player will be able to pull more than a 55% w/l while solo reletively easily is all I am saying. And yeah, it goes without saying that bads will have no trouble pulling sub 45% w/l. Like I was saying, bads gon b bad and good players will carry harder.

Quote

As to the second point... no that isn't actual or impartial. By definition you'd only be showing relative skill on a composition per composition basis... which is also statistically meaningless. "I beat that guy when I dropped with 3 of my friends one time" isn't a useful measuring tool for matchmaking.

You might want to check out how games like Battlefield and other modern multiplayer games do skill ranking to get an idea of how utterly pointless elo is.

Given enough attempts a pattern will form. If player B cannot win a match against player A, we can of course conclude that player A is better. Better players will statistically perform better and win more. This is irrespective of their Elo rating. Only when those wins are counted over time and a value given will they be relevant for MM purposes.

[ScarecrowES]

Ghogiel, on 06 July 2014 - 04:52 PM, said:

Pretty much every good player will be able to pull more than a 55% w/l while solo reletively easily is all I am saying. And yeah, it goes without saying that bads will have no trouble pulling sub 45% w/l. Like I was saying, bads gon b bad and good players will carry harder.


Given enough attempts a pattern will form. If player B cannot win a match against player A, we can of course conclude that player A is better. Better players will statistically perform better and win more. This is irrespective of their Elo rating. Only when those wins are counted over time and a value given will they be relevant for MM purposes.

I see the whole science and math thing has zoomed right over your head. But that's ok. PGI doesn't get it either, and they're "professionals."

[Ghogiel]

ScarecrowES, on 06 July 2014 - 05:02 PM, said:

science and math

[Jeb]

based on the reading I have done, the match maker tries to create teams from a range of Elo rated players. I can't seem to find what that range is currently set to (might be 1400 as that is the last post I can find on it, which was Jan 2014), or what it uses when it first starts creating a match (I would assume the it uses the first player it adds to the match but that is just me guessing)
It's not totally random as suggested, and not 1 for 1 from what I can tell...

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

Quote

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;

http://mwomercs.com/...79-matchmaking/
This one talks about the groups Elos being added up and averaged, and then how the Elo is adjusted based on if the team expected to win, wins or loses. I think this team average is where people get confused... based on these links, that is done AFTER the teams are created and is not used in creating the teams...


Now with 3/3/3/3 it may have changed, as they do now look for weight classes, but if so, it would be great if someone could post a link to those changes


Some more links on match making:
http://mwomercs.com/...ted-april-19th/
http://mwomercs.com/...chmaker-update/
http://mwomercs.com/...old-adjustment/
http://mwomercs.com/...sizes-and-more/

Edited by Jeb, 06 July 2014 - 11:44 PM.

[Stormyblade]

Ghogiel, on 06 July 2014 - 04:52 PM, said:

Pretty much every good player will be able to pull more than a 55% w/l while solo reletively easily is all I am saying. And yeah, it goes without saying that bads will have no trouble pulling sub 45% w/l. Like I was saying, bads gon b bad and good players will carry harder.


Given enough attempts a pattern will form. If player B cannot win a match against player A, we can of course conclude that player A is better. Better players will statistically perform better and win more. This is irrespective of their Elo rating. Only when those wins are counted over time and a value given will they be relevant for MM purposes.

For a short while maybe...then reality will kick in and no matter how good this "high ELO" player is, they will find they can't carry the other 11 members of the team and lose, over and over and over again.

I'd like to know how "better" players will be able to do that much for their team and drive a team of PUGs to more and more wins. I've mentioned before that I've been in matches with some of the self-proclaimed competitive/high ELO players and even with their group on our team we have been dealt horrendous losses.

The point is, if you have a team that is truly made up of PUGs and just one group of four on the other that's a pre-made, the coordination and strategy that can result from just the group of four easily makes up for the lack of the same for an entire group of twelve on the other team.

My own personal experience of the past several days (since the patch where MM was tweaked) has been even more of the same as it was before the patch -- one team pretty much destroys the other team and there haven't been many truly "good" matches where it comes down to 2 'mechs vs 2 'mechs. It's very discouraging, even as someone that has been playing for close to 2 years now to find that you are a single 'mech facing easily 6-7+ 'mechs on the other team, some of which have barely been scratched, and you're already having Betty inform you about your critical damage. Oh, and lastly, according to the patch notes, it appears that you can still face a team of 2-4 in a group as a solo PUG person -- so how did that part change for the better?

[Filane]

Stormyblade, I agree 100% with everything you said. Good post.

[Fooooo]

Jeb, on 06 July 2014 - 11:40 PM, said:

snip!!
http://mwomercs.com/...-making-update/

snip!!
http://mwomercs.com/...79-matchmaking/



Now with 3/3/3/3 it may have changed, as they do now look for weight classes, but if so, it would be great if someone could post a link to those changes


Some more links on match making:
http://mwomercs.com/...ted-april-19th/
http://mwomercs.com/...chmaker-update/
http://mwomercs.com/...old-adjustment/
http://mwomercs.com/...sizes-and-more/

This is pretty much it, unless with the new MM re-write they changed that part ofc.....


First the server creates a match. This match has a set elo level. Lets say 2500 for this match it made. (if its not static then some matches would start at 1500 etc etc)

The created "match" then searches for players in the queue (players who have clicked launch) at 2500( +/- 700 i think when it relaxes.?? )


So anyone with an elo in the range of 1800 - 3200 could be selected right away....thats if its +/- 700 spread at the start....it might only be 100 or so until the first relax..... (now they must also fit the 3/3/3/3 criteria also until it "relaxes" and puts 5 assaults in etc.....)



Now it may have only found say 3 players in that range. (1800 - 3200) and has been sitting there for a bit and has now hit its "relax" point.........the elo now widens (I have no idea by how much and if it only goes down or not...) and the match is now basically set at 2000.

Anyone in the 1300 - 2700 (it may ADD to the range instead tho, they never said....so 1300 - 3200 could be the relaxed spread) elo will now be picked........



This goes on and on until a match happens..............unless they changed it all ofc......




For groups things change a little. A groups elo is averaged. That elo is now used as if its a "single persons elo"



IE a group of 4 given an average elo of 1500 wont get into the match above until it relaxed and accepted 1300 - 2700 etc etc.......



The WHOLE TEAM is averaged after a match. (actually before probably to make things slightly faster, tho it doesnt need to be that way...) and it predicts the chances of each side winning based on the averaged elo.

The points (rise in elo or drop) are determined from this last WHOLE TEAM average.



Yes, many people got it all mixed up...........however like I said this may have changed a bit with the new MM re-write.......... =o

Edited by Fooooo, 07 July 2014 - 01:45 AM.

[OneEyed Jack]

ScarecrowES, on 06 July 2014 - 03:46 PM, said:

It really doesn't matter, either way. Outside of organized competitive play between fixed teams, Elo is a meaningless stat. It has minimal impact on actual team composition but has ZERO reflection of actual player skill. It might as well be ignored entirely in matchmaking.

But it's not completely meaningless. Better players will gradually rise in Elo an lesser players will gradually fall, until each reaches a point that they balance out. A player doesn't have to carry every team to have a statistical impact over the course of many games. A shot here, a dodge there, occasionally pulling off a "miracle" comeback that a lesser player wouldn't have, and it adds up.

ScarecrowES, on 06 July 2014 - 03:54 PM, said:

Put the same players in nothing but pug drops for a year, and their w/l ratio will remain 50% +/- 5% (give or take) for the year, and their Elo will drop accordingly. That's statistics 101.

ScarecrowES, on 06 July 2014 - 04:19 PM, said:

If you managed better than say 55% w/l doing nothing but pugging solo, not only would it be unlikely, but also statistically meaningless. It would merely serve as an offset for similarly skilled players whose ratio would fall below 45% accordingly. Nothing more.

ScarecrowES, on 06 July 2014 - 05:02 PM, said:

I see the whole science and math thing has zoomed right over your head.

"Even the best players W/L would even out, but if it didn't, and I was proven wrong, I'd still be right, because it wouldn't mean anything, because I'm right," isn't really a very cohesive argument.

[Willard Phule]

Sandpit, on 06 July 2014 - 04:03 PM, said:

no

some people at the top of the pile have an educated guess as to what range their Elo is in but noone knows what their Elo is

If you never see Trial Mech's alpha their teammates in the back in spawn while in 3PV, odds are good you're at the top of the pile. If there's actually a "top of the pile" anymore.

[Jeb]

Found a couple more links:
The first one talks about new players starting at a lower Elo for their first 25 games.

http://mwomercs.com/...23-21-may-2013/

Also here is confirmation that Elo is based on weight class:
http://mwomercs.com/...ost__p__2766628



I know it would cause some longer wait times to lower the Elo range match maker has to pick from, but I personally think the range is just too wide to get a properly balanced match most times...

I have read that you get a bell curve if you plot the Elo ratings... meaning the majority of players are somewhere in the middle...

Maybe they need to do something with the low and high ranges so that those ranges are more open to pulling in people if needed, but the matches created in the middle of the bell curve are more strict... so the wait times are not crazy on the low and high ends, but the ranges where there are lots of players to use in a match are more balanced.

[Ghogiel]

Stormyblade, on 07 July 2014 - 12:28 AM, said:

For a short while maybe...then reality will kick in and no matter how good this "high ELO" player is, they will find they can't carry the other 11 members of the team and lose, over and over and over again.

This player you describe is not a good player then. If he were, reality would kick in and they'd carry their w/l to be 55%+while solo without much issue.

Quote

I'd like to know how "better" players will be able to do that much for their team and drive a team of PUGs to more and more wins. I've mentioned before that I've been in matches with some of the self-proclaimed competitive/high ELO players and even with their group on our team we have been dealt horrendous losses.

The point is, if you have a team that is truly made up of PUGs and just one group of four on the other that's a pre-made, the coordination and strategy that can result from just the group of four easily makes up for the lack of the same for an entire group of twelve on the other team.

My own personal experience of the past several days (since the patch where MM was tweaked) has been even more of the same as it was before the patch -- one team pretty much destroys the other team and there haven't been many truly "good" matches where it comes down to 2 'mechs vs 2 'mechs. It's very discouraging, even as someone that has been playing for close to 2 years now to find that you are a single 'mech facing easily 6-7+ 'mechs on the other team, some of which have barely been scratched, and you're already having Betty inform you about your critical damage. Oh, and lastly, according to the patch notes, it appears that you can still face a team of 2-4 in a group as a solo PUG person -- so how did that part change for the better?

It's changed a bit since now a solo player isn't stuck between 4 premades fighting it out. Only 2. Plus usually the bulk of the good players are in group queue so the competition in solo queue is a bit thin.