Jump to content

This Elo System Is Working Perfectly


19 replies to this topic

#1 Tarantoga

    Member

  • PipPipPip
  • Knight Errant
  • 68 posts

Posted 22 February 2013 - 08:47 AM

especially for solo players

if you dont like criticism close this page right now

i queued solo, no, the queue didnt take 2+ minutes to find a game

now if the game wasn't relaying on some artificial scoring model made for a freaking SINGLEPLAYER game for team blalancing it would be 480 tons vs 415, not only tons dont match but classes too (obviously) 5 meds + light +2 heavies vs 2 assault + 2 heavies + 2 meds + 2 lights

i dont care bout your freaking elo or how it works as statistic in large numbers, im not a large number im a single person i get pissed whenever i get screwed by broken matchmaking, tell me how much you can take before you give up improving your elo as a player? do you have patience for 10k games? maybe 100k? how many it takes for elo to kick in and give some effects for solo player?

if you insist on implementing elo this way ignoring other factors that should add up to thing called game balance you are hurting mostly solo players as they cant really cooperate with team w/o voice communiation, not sure whats worse for solo players this elo or old matchmaking

Posted Image

#2 Skull Ringer

    Member

  • PipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 132 posts
  • LocationTexas

Posted 22 February 2013 - 08:54 AM

Agreed... ELO sucks for this game..

JUST FRIGGIN GIVE US PUG VS PUG!!!!

Let the Premades group up and take their chances that their group is better than the groups that simple ton/battle score balancing provide..

For pug vs pug balancing; take the grouped lances out of the match and all you need is simple ton/battle score..period.

#3 Mazzyplz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,292 posts

Posted 22 February 2013 - 08:56 AM

OP:
please check this thread out, it's a suggestion to fix this problem of classes.

http://mwomercs.com/...n-filled-match/

#4 Darwins Dog

    Member

  • PipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 1,476 posts

Posted 22 February 2013 - 09:00 AM

I'm not sure what the picture is trying to show. That looks like a really good game to me.

Pug vs. pug needs Elo as much as the premade groups do. New and learning players will still get rolled by teams with one or two elite lone wolves, and they will still have a hard time learning.

Check the developers' posts about Elo. They will be making changes to the rank and weight matching systems in the next couple of weeks and as time goes on. It's not perfect right now because they still need the data from live players.

#5 Wingbreaker

    Troubadour

  • PipPipPipPipPipPipPipPip
  • The Sureshot
  • The Sureshot
  • 1,724 posts
  • LocationThe city that care forgot

Posted 22 February 2013 - 09:01 AM

And yet the average damages were 299.875 versus 257.75.

2399 damage versus 2062 damage.

Effectively, it was actually a very even match regardless of tonnage. You had a single outlier AWS-8R, which was probably a missile boat setup, eh?

That's actually how Elo is supposed to work.

#6 jakucha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • 2,413 posts

Posted 22 February 2013 - 09:02 AM

I'm getting more balanced games than 8-0 blowouts now. I still get some bad games, but the Elo system is pretty new and will take tuning like the devs said. And if you read the command chair post you'd see that the matchmaking system does take weight into account: http://mwomercs.com/...-making-update/

Edited by jakucha, 22 February 2013 - 09:04 AM.


#7 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 09:12 AM

I am bewildered by the number of people who think that matchmaking should exactly match weight classes and skill levels, do so instantaneously, and ensure that the end results are an 8-7 win with no disconnects. How does that happen exactly? Can you show me the formula, because as someone who develops web applications that often involve participant matchmaking, I'd be intrigued to know how you would solve the issue.

Obviously it's a work in progress. This is not the sort of thing that ANYONE gets 100% flawlessly correct right out of the gates. However, look around at the boards. Whereas before people seemed to pretty universally bash the system, now you're seeing a lot of people singing its praises and talking about the great matches it produces.

Now, let's dissect your match a bit.

First, how do you know that the folks on your team weren't, in general, better players in terms of W/L records? You don't. All you have to go on is the weight class info and the final scoreboard. Yes, perhaps the weight balancing isn't precise here, but I'd say 2 heavy/5 med/1 light versus 2 assault/2 heavy/2 medium/2 light isn't THAT bad. And compare the damage done and assists. Beyond the obvious data outliers of the top two guys on the other team, only one of which was an assault mech, the rest of the field looks pretty even. The guy in the #2 spot was even driving a mech you had FOUR of.

Ultimately, I just don't understand how this serves as any evidence that the system is broken.

#8 Zolaz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 3,510 posts
  • LocationHouston, Tx

Posted 22 February 2013 - 09:13 AM

Looks like you are suppose to run an Atlas or an Assault so that you can fight Meds and Heavies and have up to a 300 ton weight advantage in the games.

#9 LT Kinslayer

    Member

  • PipPipPipPipPip
  • 137 posts

Posted 22 February 2013 - 09:14 AM

So, let me get this straight.

-total damage done for both teams was pretty close to equal
-match took 8:22 minutes
-It was an 8-4 game
-tonnage was 480-415

What on this one screenshot exactly leads you to the conclusion that Elo must be bad?

I sure hope the reasoning to get to your conclusion was not: "I lost, therefore Elo sucks donkey balls"

Because this one game was far from a roll...


edit to correct tonnage, I somehow thought spider was 20 tons.

Edited by LT Kinslayer, 22 February 2013 - 09:23 AM.


#10 Mazzyplz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,292 posts

Posted 22 February 2013 - 09:16 AM

View PostGallowglas, on 22 February 2013 - 09:12 AM, said:

I am bewildered by the number of people who think that matchmaking should exactly match weight classes and skill levels, do so instantaneously, and ensure that the end results are an 8-7 win with no disconnects. How does that happen exactly? Can you show me the formula, because as someone who develops web applications that often involve participant matchmaking, I'd be intrigued to know how you would solve the issue.



i'll explain to you real quick!

the matchmaker populates 16 players with ELO.

but instead of launching the game with 8 atlas vs 8 commandos;
in the time you chat before powerup, it would put switch 4 atlas with 4 commandos to the opposing team.

no ELO involved at this point, no players outside the match involved.
just some very basic in game code

#11 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 09:18 AM

View PostZolaz, on 22 February 2013 - 09:13 AM, said:

Looks like you are suppose to run an Atlas or an Assault so that you can fight Meds and Heavies and have up to a 300 ton weight advantage in the games.


And yet the enemy Trebuchet got as many kills and more damage than either of the friendly heavies or the enemy Atlas. In fact, the Awesome is the only mech that really stood out amongst the heavies and assaults in that match. It's the Trebuchets that look like they were dominant.

View PostMazzyplz, on 22 February 2013 - 09:16 AM, said:



i'll explain to you real quick!

the matchmaker populates 16 players with ELO.

but instead of launching the game with 8 atlas vs 8 commandos;
in the time you chat before powerup, it would put switch 4 atlas with 4 commandos to the opposing team.

no ELO involved at this point, no players outside the match involved.
just some very basic in game code


And when you end up with an imbalance in skill level, premade count, or time to find a match, you'd still be calling for PGI's heads.

Edited by Gallowglas, 22 February 2013 - 09:20 AM.


#12 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 22 February 2013 - 09:20 AM

that match looks like it was still better than 90% of the pre-Elo matches I ever had.

I had a match go 13:42 seconds yesterday. It was a cold, calculated battle with tons of tactics on both sides on Alpine. We lost but only barely. 1 or 2 poor choices by a couple of people swayed the match. 1 person overheated at the wrong time, one got caught away from the rest of his team or else it would have gone the other way.

I'm seeing that far, far more. Fights coming down to 1 or 2 mistakes not just a one-sided crushing skill advantage. Some matches still end up relatively one-sided in kills but that was more a domino effect than anything else. The winners didn't come out unscathed.

I love Elo. It's made the game far better for me.

#13 Mazzyplz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,292 posts

Posted 22 February 2013 - 09:20 AM

View PostGallowglas, on 22 February 2013 - 09:18 AM, said:


And yet the enemy Trebuchet got as many kills and more damage than either of the friendly heavies. In fact, the Awesome is the only mech that really stood out amongst the heavies and assaults in that match. It's the Trebuchets that look like they were dominant.



And when you end up with an imbalance in skill level, premade count, or time to find a match, you'd still be calling for PGI's heads.


no dummy.

no imbalance on skill level did you not understand? the match would still be populated by similar ELO on both teams.
just switch some from one team to the other at launch (chat time in game)

no added time finding a match either.
because it's still the same 16 players matched by ELO, just switch 4 around IN GAME (when elo is done with it's work making the match, just one more step toward balancing it IN GAME)

Edited by Mazzyplz, 22 February 2013 - 09:21 AM.


#14 Amsro

    Member

  • PipPipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 3,441 posts
  • LocationCharging my Gauss Rifle

Posted 22 February 2013 - 09:27 AM

This current ELO only matchmaker while flawed in many ways (also seems to have some benefit) mainly prevents established players from buying/building NEW mechs. If you don't take your mastered/perfectly tweaked build you will be very detered.

Quite simply its hit and miss with the current setup, from even matches to rediculous 6 Assault/2 Heavy vs 3 Medium/5 Light.

I understand the learing curve of ELO, but it needs more regulations. Some matches just shouldn't happen, others have been great.

#15 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 09:33 AM

View PostMazzyplz, on 22 February 2013 - 09:20 AM, said:


no dummy.

no imbalance on skill level did you not understand? the match would still be populated by similar ELO on both teams.
just switch some from one team to the other at launch (chat time in game)

no added time finding a match either.
because it's still the same 16 players matched by ELO, just switch 4 around IN GAME (when elo is done with it's work making the match, just one more step toward balancing it IN GAME)


Dummy? Insults are often the last refuge of the intellectually disadvantaged, but I'm not going to take that bait. Suffice it to say I don't feel a need to defend my intellectual prowess.

Your suggestion is based on the premise that the system is typically pitting 8 lights versus 8 assaults. Are you seeing this typically happening? I'm not. Even so, the possibility that some of each team may be premade groups coupled with which pilots in which mechs have what ELO complicates the issue. It may not be as simple as it seems. Sometimes, perhaps quite often, the puzzle pieces don't all fit nicely.

Edited by Gallowglas, 22 February 2013 - 09:38 AM.


#16 Mazzyplz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,292 posts

Posted 22 February 2013 - 09:38 AM

no the matches are not terribly imbalanced at the moment; youre right

but you still can spawn in alpine conquest mode with no light mechs on your team, while the opposing one might have one or many.

that's a guaranteed loss by cap.

how is that fun??

4 man can be a problem but average out their weighs and balance it in game (no match maker)
by an average of tons for 4 players.
(or pit them against another 4 man in the opposing team if 2 are present - disregarding weights in THIS particular case)
don't tell me a computer cannot calculate this in miliseconds, it can.


ps.
dummy is not a real insult, it's what you call your sibling.
i guess you haven't seen insults on the internet! hope you never do because trolls r nasty

at any rate i take it back and apologize for any unintended offense

Edited by Mazzyplz, 22 February 2013 - 09:43 AM.


#17 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 09:50 AM

View PostMazzyplz, on 22 February 2013 - 09:38 AM, said:

but you still can spawn in alpine conquest mode with no light mechs on your team, while the opposing one might have one or many.

that's a guaranteed loss by cap.


Those lights might be poorer players than the rest of the team (i.e. it's a top-heavy team), while the opposing team might be, on average, the same ELO. So by replacing the lights, you've imbalanced the ELO. That might be acceptable on Alpine/conquest, but it might be disastrous for River City/assault.

View PostMazzyplz, on 22 February 2013 - 09:38 AM, said:

don't tell me a computer cannot calculate this in miliseconds, it can.


Individual calculations are very fast, yes. However, when you're comparing the stats of thousands of players in hundreds of threaded processes, it's not always as fast as you might think.

#18 Mazzyplz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 3,292 posts

Posted 22 February 2013 - 10:36 AM

Quote

Individual calculations are very fast, yes. However, when you're comparing the stats of thousands of players in hundreds of threaded processes, it's not always as fast as you might think.


nope, it's very much fast as it could be, in game.

it's the same way damage is calculated server side, or how the triggers for base capping / round win works.

make calculations in game when 16 players have filled the ranks. no need to query thousands of anything.
i think you're still missing the point i'm trying to make.

Quote

Those lights might be poorer players than the rest of the team (i.e. it's a top-heavy team), while the opposing team might be, on average


how are they poorer players if the match is done by ELO scores?
all 16 players should be in ballpark of the same score from the start.

when matchmaker has filled the ranks - and is out of the picture. (MM would remain the SAME, no changes to MM just another balancing for just 16 players in match on top of that, unrelated to MM)

by the same triggers you use to start/end the match, calculate damage server side or allow base caps; you could do a last minute balancing with just 16 players to move around.

no it wouldn't take longer to calculate than calculating chances of criticals on several missiles at runtime (game time)

Edited by Mazzyplz, 22 February 2013 - 10:39 AM.


#19 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 11:25 AM

View PostMazzyplz, on 22 February 2013 - 09:38 AM, said:

at any rate i take it back and apologize for any unintended offense


No worries. No harm, no foul.

#20 Gallowglas

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,690 posts

Posted 22 February 2013 - 11:30 AM

[quote name='Mazzyplz' timestamp='1361558160' post='1941730']
how are they poorer players if the match is done by ELO scores?
all 16 players should be in ballpark of the same score from the start.

when matchmaker has filled the ranks - and is out of the picture. (MM would remain the SAME, no changes to MM just another balancing for just 16 players in match on top of that, unrelated to MM)

by the same triggers you use to start/end the match, calculate damage server side or allow base caps; you could do a last minute balancing with just 16 players to move around.

no it wouldn't take longer to calculate than calculating chances of criticals on several missiles at runtime (game time)
[/quote]

The current matchmaker, so far as I can tell from reading about it, doesn't match based on individual ELO numbers. It uses an average of all 8 players:

[quote name='Paul Inouye]At the start of a match' date=' all player's on ONE team have their Elo ratings totaled and divided by 8 (max players). You may realized that this is simply the AVERAGE of a team's total Elo. Team 1's average and Team 2's average are then used to calculate the probability of win (as per the formulas above). If Team 1 beats Team 2, then the appropriate math as above is applied to each player using the probability score calculated by the team averages.[/quote']

Source: [url="http://mwomercs.com/forums/topic/66079-matchmaking/"]http://mwomercs.com/...79-matchmaking/[/url]

So you could end up with one team that has all 1300 ratings and another team with some low ratings balanced by some high ratings. All the matchmaker cares about is that they average out to about the same number.

Edited by Gallowglas, 22 February 2013 - 11:34 AM.






6 user(s) are reading this topic

0 members, 6 guests, 0 anonymous users