205 replies to this topic

[Franchi]

FrDrake, on 25 March 2013 - 11:54 AM, said:

So you admit there are 30% of games that would be more fun if you could play better

FrDrake, on 25 March 2013 - 11:54 AM, said:

So you admit there are 30% of games that would be more fun if you could play better

eh maybe I could play better in the 25% I lose, but there is not a lot you can do when your team has no lights and the other team has 4 on alpine conquest, or when you are down by 500 tons on forest colony assault.

games were far more balanced before elo when matchmaker was tonnage and nothing else.

Edited by Franchi, 25 March 2013 - 12:26 PM.

[Tikkamasala]

FrDrake, on 25 March 2013 - 11:36 AM, said:

The average Elo of the match is 1500. 4man-elite-dudes are 2000 a piece, that means the other 4 are going to be 1000. That's how the process works if they can't find 8 2000 Elo guys to put you up against. Carry the pugs if you have to, someone with a 76% win rate should be able to.

That's not how MWO's Elo-system works though. If it's trying to fill a 1500 Elo match it will try to fill it with people with an Elo close to 1500. If that's not possible the acceptable Elo variance increases.
So, in the case of the 2000 Elo four-man it's rather unlikely the MM will put them in the 1500 match. They'll get matched with people close to 2000. It's not going to balance 2000 and 1000 to fill a 1500 match.

[Captain Stiffy]

ELO can't possibly work and here's why

I can utterly dominate a match with a team that doesn't just run all over like chickens with their heads cut off (literally they do this) almost all of the time or make a very good show of it. Depending on my team I'm top or bottom of the scoreboard and that's just the kind of player I am I guess. In the end, nobody can go-it-alone anyway.

That said, I also often dominate many matches in a row with certain classes or loadouts of mechs and because I'm always going from mech to mech playing something different and also because if I pull alpine or the new desert in my DDC I'm going to disconnect immediately about 30% of the time cause I refuse to deal with it.

So my ELO is in no way reflective of my ability level and I'll trash the game in groups where it assigns me low. The only groups I can't seem to fight are the team dropped ones at the very top of the ELO, and I totally 100% agree with what was said by the OP - you seem to very often get 4 pros and 4 completely and overtly terrible nubs on your team which does not a victory scenario make.

TLDR;
I play lots of different mechs and leave matches when I feel like it so my ability level and my ELO can never truly be in sync - thus ELO is not a quality by which you would want to match players.

ALSO REALLY QUICK:
THE ONLY WAY TO GET BETTER IS BY PLAYING AGAINST PLAYERS THAT ARE BETTER THAN YOU AND LEARNING THEIR TACTICS! ELO BASED MATCHMAKING STIFLES YOU INTO A STINKY CRACK THAT YOU CAN'T LEARN YOUR WAY OUT OF BY PLAYING IN GAME

Edited by Captain Stiffy, 25 March 2013 - 01:02 PM.

[EgoSlayer]

Funkadelic Mayhem, on 25 March 2013 - 11:11 AM, said:

Its funny you say this and support this 50/50 every one is equally special Elo system. LMFAO!

<snip>

Quit saying this 50/50 quote. All that proves is that you don't have any idea how Elo works.

Elo *never* looks at or cares about your W/L ratio.

Elo attempts to match you with players of near equal skill level.

Statistics being what they are, players of equal skill each have an equal chance of winning. This is completely different from trying for a 50/50 ratio.

Elo looks at one stat, and one stat only - Did you win or lose the current game. That is what changes your Elo Score. That is it, not your previous games, not what it thinks your future 10 games will be.

If you win or lose against an equal opponent then your score doesn't change much, if at all. If you beat someone rated significantly higher than you, then your score improves a large amount to reflect that your Elo score is too low. If you are beaten by someone who's score is dramatically lower than yours then your score will go down a great deal to reflect that you are rated too high.

That's it. No voodoo, no trying to make your W/L percentage 50/50. But if you are playing people who have the same skill, all things being equal, each of you has an equal chance of winning. Now, with the complexity/variety of the Mechs and classes, voice coms, pre-made vs pug, etc, all things are not equal. So it can take hundreds or thousands of matches for your elo to be an accurate reflection of your ability in each mech class, especially if you are changing things yourself e.g. different load outs or different mechs in the same weight class.

[Mazzyplz]

been experiencing this.

super low ELO players should not be allowed to play with average/high elo players.
PERIOD

[Agent 0 Fortune]

Mech_____________Played__Wins_Losses__Ratio__Hours
JAGERMECH JM6-DD___62_____35____27____1.30__5hr 10min
JAGERMECH JM6-A____55_____23____32____0.72__5hr 11min
JAGERMECH JM6-S____49_____25____23____1.09__4hr 22min
NOTE: don’t ask me why the total games played exceeds the total win/losses, but I suspect that kind of math explains a lot about the match maker.

Total Matches:___166
Total Win/Loss:__83/82
Ratio:___________1.01

These numbers are based on this weekend’s tournament since I purchased the Jaegers for the event and played almost exclusively Jaegermechs it is a pretty good snapshot of the ELO in action. It almost looks like the ELO is working as expected, that is a 1 to 1 win/loss ratio. Until you at the individual matches:

• Roughly 85% of the games played end in a brutal, soul-crushing, PUG-stomping defeat in one direction or the other (8-0, 8-1, 8-2).
  
• Another 12% ended in humiliating loss (8-4, 8-3) for one team.
  
• And 3% of the matches (about 5 of the games played over the course of 15 hours) was a respectable challenge with an exciting back and forth battle with heavy losses on both sides.

Basically ELO seems to do a good job of forcing PUG players into a 1:1 win/loss ratio, but the quality of those matches is questionable.

Edited by Agent 0 Fortune, 25 March 2013 - 01:50 PM.

[Papaspud]

EgoSlayer, on 25 March 2013 - 01:32 PM, said:

Elo *never* looks at or cares about your W/L ratio.

Actually the devs have said that is the MAIN thing ELO takes into account.

[Mazzyplz]

JAGERMECH JM6-DD___62_____35____27____1.30__5hr 10min
JAGERMECH JM6-A____55_____23____32____0.72__5hr 11min
JAGERMECH JM6-S____49_____25____23____1.09__4hr 22min

HAH!
useless missiles on that 6-a.
and people say they're just fine. missile variant is the worst

[Agent 0 Fortune]

Mazzyplz, on 25 March 2013 - 02:05 PM, said:

JAGERMECH JM6-DD___62_____35____27____1.30__5hr 10min
JAGERMECH JM6-A____55_____23____32____0.72__5hr 11min
JAGERMECH JM6-S____49_____25____23____1.09__4hr 22min

HAH!
useless missiles on that 6-a.
and people say they're just fine. missile variant is the worst

Only if you want to carry 3+ Ballistic points, otherwise it is much more vesititle than the other Jaegers.
I tried out the following, and ulitimatley I feel it was luck of the draw regarding win/loss.
JM6-A PPC, AC10, 4x LRM5
also tried: 2x Gauss +2 Md Laser
2x AC20
2x UAC5 + 2x Lg Laser
2x ER PPC + 4x SSRM2

[Papaspud]

I'm not sure how it works, but my matches since they started ELO are 241-238 damn close to 50/50. Problem is a lot of those matches I am put on teams that really stink, I don't claim to be great, but Daaaamn some of these guys really are new to the game or well you know. There seem to be way too many roflstomps one way or the other, really tired of watching 1 or 2 rambos take off right when the game starts..... probably game lost in the first 2 minutes.

[FunkyFritter]

This is generally an issue caused by low population. There aren't enough people playing at your level to get an even game, so the system has to stretch to make things remotely close.

[EgoSlayer]

Papaspud, on 25 March 2013 - 02:03 PM, said:

Actually the devs have said that is the MAIN thing ELO takes into account.

No they didn't, Elo never looks at your win lost ratio or history. Elo only looks at the current game and if you won or lost. Period. Your Elo score is a reflection of your history, but still not tied to any w/l ratio. It's all about the relative rating between you and the opponents.

http://mwomercs.com/...65#entry1626065

[Angus McBeef]

I thought "Elo" was a guy's name, not an abbreviation like "E.L.O." Otherwise, I have nothing constructive to say.

[SgtKinCaiD]

Another evening and the same thing, 11 matchs played, 9 of them are roflstomp (8-0 8-1 8-2) with each time several players doing less than 100dmg, 2 "balanced" matchs ending close, 8-7 and 8-5

[Papaspud]

EgoSlayer, on 25 March 2013 - 02:29 PM, said:

No they didn't, Elo never looks at your win lost ratio or history. Elo only looks at the current game and if you won or lost. Period. Your Elo score is a reflection of your history, but still not tied to any w/l ratio. It's all about the relative rating between you and the opponents.

http://mwomercs.com/...65#entry1626065

Did you read it, it is ALL about winning or losing, who you beat and who you lost to. That is your history....if you can point out where they say anything but winning or losing makes a difference please feel free to show me what it is.

Edited by Papaspud, 25 March 2013 - 03:30 PM.

[Mazzyplz]

SgtKinCaiD, on 25 March 2013 - 03:01 PM, said:

Another evening and the same thing, 11 matchs played, 9 of them are roflstomp (8-0 8-1 8-2) with each time several players doing less than 100dmg, 2 "balanced" matchs ending close, 8-7 and 8-5

did you experience rambo players taking off on their own in the first minute or so?

[Chemie]

I even see this with 2-man where 5 are total idiots with less than 100 damage each (or sometimes together). With 4-man it is either 8-0 because you won the lottery or you get the 4 idiots like OP said. ELO is averaging when queues are low (even though PGI say it is to target). Then throw in mismatched weights and I honest wish we were back to pre-ELO when things were better balanced.

[WVAnonymous]

EgoSlayer, on 25 March 2013 - 02:29 PM, said:

No they didn't, Elo never looks at your win lost ratio or history. Elo only looks at the current game and if you won or lost. Period. Your Elo score is a reflection of your history, but still not tied to any w/l ratio. It's all about the relative rating between you and the opponents.

http://mwomercs.com/...65#entry1626065

Important to be clear here on Elo versus matchmaking:

If your Elo score is combined with a matchmaking system that matches you to similarly scored individuals, claiming the coin toss is independent does mean that if you toss a coin 1000 times, the extremely strong tendency is that you will have something pretty close to 500 heads and 500 tails, or the coin is biased.

If we have people with 1000 games in the same weight class that are significantly distant from an even W/L split, I will take that as sufficient evidence that the matchmaker is NOT matching them to similarly scored individuals.

The failure to be matched with similarly scored individuals on both teams is the crux of the thread, i.e. matchmaking is broken in certain cases, not the Elo score. However, since the matchmaking is based on the Elo score, it is easy to blame the Elo score for the matchmaking failure. Particularly since no one knows what they have for an Elo...

For the record, post stats intro, I have 165 wins and 203 losses. I assume my Elo will keep inching down until that evens out.

[Targetloc]

People at the high end will keep a high win/loss ratio because there aren't enough high elo players launching in the same 2 minutes to always pair high elo players together.

Also, if the range of 'close enough' is wide enough, they'll also keep a positive win/loss (even if they're not near the top of the elo range) because the further you go out from the middle, the more extreme the difference between the number of players available to play from your left, versus number of players available to the right.

[MischiefSC]

I love this talk of 'balanced matches' meaning everyone does damage and kills people.

I drop with teams from time to time. Really good people in a bracket that is way above where I normally play. Even the best people get killed quickly sometimes. Sometimes you get out-positioned. Everyone throws up some sub-100 games sometimes. In fact when 1 or 2 people do the game can swiftly turn into a roll.

That doesn't mean their Elo is bad. Just means they had a bad game.