39 replies to this topic

[Rekoro]

Hi,

I am absolutly unhappy about how the elo system is implemented at the moment. Heck i actually doubt it is implemented at all. Here are some of my sightings i ve made so far.

Doesn´t matter if grouped (up to 3 ppl in total) or alone, almost every match (~90%) i find myself matched with or against players that doesn´t belong in the match. I am talking about 2-5 ppl per match who are underperforming the rest.

These people are new players and bad players alike. Dying with less than 90 dmg dealt, and i am not talking about light mech pilots here, i am talking about heavy and assault mechs. Now someone could say, well these guys had bad luck. But i disagree on that, since this is happening almost every match i played so far, i highly doubt this is an coincidence. A friend of mine got confirmed that he has an high elo, doenst mean he is top notch but high elo none the less, and he ends up with the same kind of people and the same kind of outcome in their performance.

So here is my conclusion, an averange of 3 bad/new players per match is unacceptable for players with an mid or high elo rating. Also i do think that the matchmaking system takes a big part in this as well. since you acutally have to wait in games like Fifa or LoL to get a balanced match started you get almost instantly started here, and i highly doubt this is because they figured a way to decrease the time it takes to match you with decent people . I wouldnt be suprised if they let the system do an extended search right of the bet to decrease waiting time which resulsts in a big gap in the skill of the players.

In every other elo system i know of, no mid-high players are getting matched with new or low elo players. So i asume something is very broken here.

Since i ve no information about if the elo system is really implemented or if it is working as intended i am in the dark here, the only thing i know so far is, that this cant be right the way it works.

I ld be glad if someone could shed some more light on this than i am able to do. I could back this up with screenshots, but i think you all know what i am talking about.

PS: pls dont get me wrong, i am not talking about 1 person who gets owned or one team getting stomped. i really do mean 2-5 people underperforming aka plain bad players across the teams.

[Rekoro]

i think i mentioned that

[Davers]

Lots of XL engine builds plus high alphas equal a quick ride to Spectatorville.

That said, the Devs have talked about starting new players with 'below average' Elo instead of the average Elo they are starting with now. This hasn't been implemented yet. Since we don't know how wide a net Elo is casting to find players it may be that you have to be exceptionally high (or maybe exceptionally low ) to avoid seeing new players at all.

We have only seen one round of Elo tuning, and that mostly seemed to consist of fixing a bug. I imagine that there will be more changes to come.

[Rekoro]

execptionally low seems a good place to be since there are no ppc boat trial mechs .
btt, i didnt say they dont work on it, i just want to point out that this even happens to players with high elo, and that this is something that needs attention. But i dont expect them to do something about this in the next months since they are slow with everything

[jeffsw6]

Neverfar, on 29 April 2013 - 07:50 PM, said:

Part of the problem might be that the matchmaker wants to fill up slots too quick instead of actually trying to make fitting matches with well-paired players.

Many matches are near-instantaneous, which tells me "lets get these guys who do 30 damage a match or shoot their own team and shove them where we can".

I've said the same thing over and over. I would rather wait 30 seconds, or 60 seconds, for a drop to begin than end up in a match where the opposing team has a huge advantage. When my team cleans up 8-0 it's okay, though.

With all due seriousness, there are a multitude of match-maker problems that the devs have not even attempted to fix. One team regularly getting 4 lights where the other is 2 heavies and 6 assaults; sync drops where you end up playing against two 4-mans who dropped at the same time and are literally an 8-man vs PUG; total noobs in trial mechs facing average or better players; 1 or 2 players who are good being tossed in with 6 dummies and expecting a fair chance at a win.

They're not really trying. They seem to have this ******* fantasy that ELO is brilliant. If we all had the same mech and in-game VOIP then that would be true. It's not. Not only is ELO not brilliant, it's probably worse than just a simple tonnage system plus making certain that 4-mans never drop against PUGs.

Besides this: What if the 4 mechs you have marked "ready" were available for the match-maker to select on your behalf? Really want to play an Assault? Fine, don't ready any non-Assaults; but you might have to wait 30-60 seconds for a drop. Put 1 light, 1 medium, 1 heavy, 1 assault on Ready status, and you would be more likely to drop quickly.

They're not trying to fix this. They are just kidding themselves with ELO.

Know why you can't see your ELO score? They don't want players to compare numbers and learn how broken the system is. ELO is designed for 1-vs-1 games where the outcome is not influenced by teammates, equipment/armaments, map, or anything else. Want to use ELO? Put in a 1v1 duel mode. That's all it's good for.

Edited by jeffsw6, 29 April 2013 - 10:33 PM.

[Rekoro]

while i agree that they dont show the elo rating because it is not working, i ve to disagree that the elo system is bad or just a 1vs1 thing. its used in sports and in plenty of online games as well and i would call MLB or League of legends hardly an 1v1 game

[Lorcan Lladd]

Rekoro, on 29 April 2013 - 11:07 PM, said:

while i agree that they dont show the elo rating because it is not working, i ve to disagree that the elo system is bad or just a 1vs1 thing. its used in sports and in plenty of online games as well and i would call MLB or League of legends hardly an 1v1 game

Ah-ah!
In League of Legends, Elo-based ratings exist in a system which is completely independent from their leaderboards and tournaments - players also possess different ratings in different game modes and for solo or group play, as well, I think.

Matchmaking and stats-tracking are imperfect, and the developers need to acknowledge this.
Perhaps increasing queue times or evenly distributing groups thorough distinct teams would be sufficient to address some issues in MWO; weight-based matching as opposed to class-based matching is also an option.

But at some point the system will fail - during off-hours, for example, there are likely to be too few players online to put up balanced matches...

Edited by Lorcan Lladd, 29 April 2013 - 11:26 PM.

[Fooooo]

Neverfar, on 29 April 2013 - 07:50 PM, said:

Part of the problem might be that the matchmaker wants to fill up slots too quick instead of actually trying to make fitting matches with well-paired players.

Many matches are near-instantaneous, which tells me "lets get these guys who do 30 damage a match or shoot their own team and shove them where we can".

That there tells me the matches should be balanced, or at least, balanced by what the Elo system says.

The faster you get a match, the more balanced the match should be as there were enough players of similar rating and weight to instantly launch a match.

The longer it takes, the opposite is usually true.....at least in my experience and by what the devs have said....

I only started playing my mediums again the other day, they have never had matches since Elo came in and boy, the matches are pretty funny compared to the ones my heavys and assaults get.

I also get a match almost instantly when using my mediums, most times.....

Most times I wait up to 1 - 2mins with heavy / assault and get the odd instant match.

Total opposite of mediums.

Thats because I have played a lot of games in those weights so there is less players to choose from initally. (compared to starting elo etc) The time I play is usually around 9am in the states I think, however all weight classes would generally have the same waiting times if it were playing any kind of major role.

Edited by Fooooo, 30 April 2013 - 12:01 AM.

[Rekoro]

Lorcan Lladd, on 29 April 2013 - 11:21 PM, said:

Ah-ah!
In League of Legends, Elo-based ratings exist in a system which is completely independent from their leaderboards and tournaments - players also possess different ratings in different game modes and for solo or group play, as well, I think.

Matchmaking and stats-tracking are imperfect, and the developers need to acknowledge this.
Perhaps increasing queue times or evenly distributing groups thorough distinct teams would be sufficient to address some issues in MWO; weight-based matching as opposed to class-based matching is also an option.

But at some point the system will fail - during off-hours, for example, there are likely to be too few players online to put up balanced matches...

well i just made some example to show that elo is working if used correctly, nothing more . as of off peak hours, we are all connecting to the us server, so in theorie there shouldnt be big off peak times

[Rackminster]

They're tracking stats. A 'bad' player must weigh out over time and be shown statistically. Why not match them with other 'bad' players? If I'm a bad player, I'd rather be pitted against equal foes and slowly churn my way up into more difficult enemies than to load into rounds where I turn a corner into a squad of people who focus-fire on my chest and send me face planting.

At this point I have no idea what kind of player the game scores me as. I have good days where I'm a face-stomping demigod and bad days where I load up and march out and suddenly find myself alone in the middle of a hostile bullet-fest because my team all fell back and I was too busy trying to kill the first guy I saw - not noticing that all my blue icons had been replaced by red ones.

My match score averages around 60, I do anywhere from 150 to 600 damage in any given round (at my peak before quitting I could get up to 900-1000 with my (F)Atlas), my average experience per match is 590 and my kill/death ratio is an abysmal 1.17. I'm all over the place on Mechs and haven't settled into one I really like so I'm not building any familiarity to get good with.

So am I good, bad, or just average?

[Keifomofutu]

I'm seeing the majority of matches as 8-2 or worse. Elo does not work in any kind of fashion. Of course it's hard to objectively rate it when you've got a dc every match.

[Keifomofutu]

Neverfar, on 30 April 2013 - 11:23 AM, said:

I find it interesting that almost everyone assumes they are only grouped with "bad" people, and that they themselves must be the "good" ones. It might explain the countless matches where the losing side has a guy saying out loud "my team were scrubz gg"

No assumption involved but yours. If you do 400-500 damage and the second best guy on your team did 225 then yeah you got hosed by your team. Or did you forget the end of game stats were a thing?

[Keifomofutu]

Neverfar, on 30 April 2013 - 12:26 PM, said:

You're emitting lethal amounts of smug. Roll up to a service station and have it checked.

No, many times the one screaming about his team is the one who died first, with about 30-80 damage to his name.

And I'm saying there is an easy way to check without assuming anything one way or another. If someone says they are getting continuously dropped with baddies and trial mechs then "shocker" maybe they are. I've certainly had my share of matches like that today.

[hammerreborn]

I hope I never see a single person in this thread do less than 150 damage or I'm screening it and posting it here. See, matchmaking sucked, I got paired with this guy!

People have ****** games, people get sniped, people get random headshot from 1600 out. Deal with it. If you were so badass you would have won the match, but didn't, which means you were just slightly less awful than they were.

And an 8-0 match means nothing if every single person on the other team is below 60% health. 8-0 matches don't take into account premades, DCs, AFKs, hud bugs, poor tatics, cap rushes, mismatched mechs, tonnage differences, or a multitude of domino effects that take place in each and every match.

In summary of this thread:
HURR DURR I WON 8-0 ELO SUCKS!

Keifomofutu, on 30 April 2013 - 12:25 PM, said:

No assumption involved but yours. If you do 400-500 damage and the second best guy on your team did 225 then yeah you got hosed by your team. Or did you forget the end of game stats were a thing?

That's just as likely to prove that you were the last one shot at rather than the most skilled player on the team.

Edited by hammerreborn, 30 April 2013 - 12:41 PM.

[frag85]

I have noticed that you end up in cycles of good teams and bad teams. There isn't really a balance so you end up with hours of good followed by hours of bad frustrating gameplay. I feel this is not a good thing for a game because you shouldn't be spending a lot more time to get the same XP/CBills for upgrades with the same mech/builds. A few games is OK, but hours of getting put in bad matches is pushing my patients for this game.

It seems like I always get 15-25 round cycles of good, cooperative, working together team games, then 15 rounds of being paired with trial mechs and players that derp off in their own direction. Very rarely do I get in matches that seem each side is equal. The only other game I play with a 'matchmaker' of sorts is iRacing, and that always puts you in where you belong.

With MWO's matchmaker though, you are either on a team that pretty much demolishes the enemies, or are completely demolished yourself. So many matches are one sided. On the bad streaks I constantly end up on the top of my team's leaderboard, quite often with more damage and kills than the rest of the team combined.
When on a win streak its the complete opposite going against teams with multiple trial mechs and poor cooperation/communication, Damage and kills are spread pretty evenly throughout the team. Even on my mechs that I have many more matches than the Jagers(heavies) and Cents(mediums) I used below. I have seen these cycles of good and bad teams since ELO was implemented and then 'fixed'. I also had similar results when I played through the Stalkers.

As some background info, I generally buy a mech, play through the basics, go to the next variant, play through the basics and so on through Elite to get the Basic 2x bonus. It usually works out where I have 4-8k Xp towards elite on the first mech while I save up for the 2nd mech of that chassis. On the data below, I ran nearly identical builds between the mechs on advice from friends and some quick trips to the training grounds.

Dual A/C-20 Jags. Same everything.
Played in this order: S, then A, then DD

Results:



And Zombie Cents with all the upgrades running 2ML and 2 SRM6+A (or 3 normal SRM6s on the CN9-A since it has 3 Missile slots). The CN9-D and AL benefited from 1 more DHS and more armor.
Played in this order: A, then D, then AL.

Results:



Similar the results of my Stalkers I ran a month ago. The first one through basics (and some of elite so I could save enough Cbil to buy the 2nd)-good, 2nd one-bad, 3rd one-good. Then same thing through Elite.

Now with the CN9-A, I unlocked the basics after a string of 12 good matches. There were a couple loses, but they were good losses (good fighting, capping, teamplay on BOTH teams), however, most of the wins were one sided. I don't expect to win every match, I know that is unfair and wouldn't ever happen, its about being put into fair matches with even gameplay, but while playing the CN9-D and being on a 6 game losing streak and having the D stuck in a match I decided to jump back into the CN9-A for another 4 game losing streak of just overall bad matches. After the A I expected that my ELO probably increased and i would be put in more evenly matched games (isn't the sample supposed to be off something like many games in that weight class? I have played dozens, even hundreds of rounds in another medium chassis), but it went from teams that worked together and communicating to horrible teams that went in 4 or 5 different directions, everyone shooting a different mech. Out of the D and into the AL and I was thrust into a string of one sided winning matches.

When I got to the CN9-D and Jager A, its like I got put on complete crap teams, not that I was put in a skill bracket too high/good for me, It was the complete opposite of what I was expecting to happen. Its like it puts you with either new people, or people with a lower skill level in order to boost their ELO by putting them with somone that has been doing OK (compared to terrible). I was constantly doing more damage than the rest of my team, sometimes more damage than the rest of my team combined. For the CN9s this past weekend it took me 1h30min to unlock the basics on the A then 2h 27 on the AL. It was almost the same for the Jagers and Stalkers.

I would like to be able to at least see my own ELO score. If I can see the rest of the team's rating that would be great as well, but its as if we aren't allowed to see it because of how broken it is, or there are too few players for it to matter.

[frag85]

Neverfar, on 30 April 2013 - 12:49 PM, said:

You make some valid points, but you need only stroll over to the World of Warcraft forums and see all the locker-room comparison contests invoked the moment someone has an opinion to see why a public Elo is a bad idea. Armory score apparently measures validity of opinion, quality of love life, real-life income, and much more on that forum. It's childish.

Yeah, I've heard about that to a degree. I've never taken the plunge into WoW myself, but have had some friends (both RL and internet) that are hardcore into it and will go off on certain aspects about it.

With MWO, I think a lot of it has to do with overall game balance. If you randomly get put on a team of support roll mechs, you are screwed. If you get put on a team of brawlers against long range snipers on an open map, you are screwed Same goes if you end up with players that are new to a chassis, or are just overall low-skill level players. No harm against them, I play games that I suck at and I know I suck at them. One of my best friends is a horrible MWO pilot, but I like playing with him because its fun and we can laugh about it. I can make sure that I am near him to help him out, but there is something about being put on a team with 7 random people, most who you've never seen, some who you'll never see again, and definitely no one you will really remember later.

In the end, I don't think there will ever be a way to completely balance the game, but it would be nice to have fewer one sided matches. When I have done 8 man matches, we have always fared pretty well even against the 'best clans' because its an organized team vs an organized team. If you are lacking in one area, someone can take a different chassis or alter their build. In pub matches, it is too much random vs too much random. Throw in a good 4 man, heck even a good 2 man and that team has a HUGE advantage.

Edited by frag85, 30 April 2013 - 01:01 PM.

[Rekoro]

iirc the system trys to keep you at a 50% win/loss ratio, which is for example why you get matched against better people when you start winning more games than you lose. and your elo rises ofc. here it seems the system matches you with or against "low skilled" players to achieve this.
Frag85 is close to this 50% ratio so the system should have placed him correctly for now, but than there are the new and bad players in either your or the enemy team where i ve to scratch my head how exactly they made it into the game. And i guess we can all agree that the elo system is in someway broken. it would be very nice to get some devs in here to clear things up, or give us some information, but i dont expect that to happen. I very much lost interest in playing mwo at the moment because the launch button often equals a "match me with reta rds" button

[Rackminster]

Rekoro, on 01 May 2013 - 02:28 AM, said:

iirc the system trys to keep you at a 50% win/loss ratio, which is for example why you get matched against better people when you start winning more games than you lose. and your elo rises ofc. here it seems the system matches you with or against "low skilled" players to achieve this.

That ... that seems like cruel and unusual punishment.

Wow, you're doing great - here's a team of newbies and a kick in the shins. Good luck with that!

What disturbs me is that you're spot-on, at least for my stats. I'm currently sitting at 221 wins and 224 losses.

When I was into World of Tanks, the player base rated the general 'skill' of another player based strictly on their W/L ratio for the most part. The theory was that, over thousands of matches, the value of a player would average out in their win/loss score. If you were good, your team would naturally win a little more. If you were bad, your team lost more. If you were terribad, you lost a lot.

There were even mods you could run that would show you the general theme someone had. I was a 52% win player - someone you'd want on your team. A 54% player was like the cream of the crop. A 50% player was just average. A 48% player was bad. A 45% player was a plague on their team.

... and it proved true. If you saw a 54%'er he was usually rockin' the battlefield. If you saw a 45%'er you'd expect him to be stuck on a wall and have an equal chance of shooting his team mates as much as the enemy. You could look at the percentages for players on each side, and there would be a high degree of probability that the team with the higher average would come out on top.

To think that MWO is striving to give me a 50/50 ratio is to be a little concerned that it will never matter how well I do - I'll always be subjected to the game forcing me to score the way it wants.

Edited by Rackminster, 01 May 2013 - 05:51 AM.

[hammerreborn]

frag85, on 30 April 2013 - 12:47 PM, said:

I have noticed that you end up in cycles of good teams and bad teams. There isn't really a balance so you end up with hours of good followed by hours of bad frustrating gameplay. I feel this is not a good thing for a game because you shouldn't be spending a lot more time to get the same XP/CBills for upgrades with the same mech/builds. A few games is OK, but hours of getting put in bad matches is pushing my patients for this game.

It seems like I always get 15-25 round cycles of good, cooperative, working together team games, then 15 rounds of being paired with trial mechs and players that derp off in their own direction. Very rarely do I get in matches that seem each side is equal. The only other game I play with a 'matchmaker' of sorts is iRacing, and that always puts you in where you belong.

With MWO's matchmaker though, you are either on a team that pretty much demolishes the enemies, or are completely demolished yourself. So many matches are one sided. On the bad streaks I constantly end up on the top of my team's leaderboard, quite often with more damage and kills than the rest of the team combined.
When on a win streak its the complete opposite going against teams with multiple trial mechs and poor cooperation/communication, Damage and kills are spread pretty evenly throughout the team. Even on my mechs that I have many more matches than the Jagers(heavies) and Cents(mediums) I used below. I have seen these cycles of good and bad teams since ELO was implemented and then 'fixed'. I also had similar results when I played through the Stalkers.

As some background info, I generally buy a mech, play through the basics, go to the next variant, play through the basics and so on through Elite to get the Basic 2x bonus. It usually works out where I have 4-8k Xp towards elite on the first mech while I save up for the 2nd mech of that chassis. On the data below, I ran nearly identical builds between the mechs on advice from friends and some quick trips to the training grounds.

Dual A/C-20 Jags. Same everything.
Played in this order: S, then A, then DD

Results:



And Zombie Cents with all the upgrades running 2ML and 2 SRM6+A (or 3 normal SRM6s on the CN9-A since it has 3 Missile slots). The CN9-D and AL benefited from 1 more DHS and more armor.
Played in this order: A, then D, then AL.

Results:



Similar the results of my Stalkers I ran a month ago. The first one through basics (and some of elite so I could save enough Cbil to buy the 2nd)-good, 2nd one-bad, 3rd one-good. Then same thing through Elite.

Now with the CN9-A, I unlocked the basics after a string of 12 good matches. There were a couple loses, but they were good losses (good fighting, capping, teamplay on BOTH teams), however, most of the wins were one sided. I don't expect to win every match, I know that is unfair and wouldn't ever happen, its about being put into fair matches with even gameplay, but while playing the CN9-D and being on a 6 game losing streak and having the D stuck in a match I decided to jump back into the CN9-A for another 4 game losing streak of just overall bad matches. After the A I expected that my ELO probably increased and i would be put in more evenly matched games (isn't the sample supposed to be off something like many games in that weight class? I have played dozens, even hundreds of rounds in another medium chassis), but it went from teams that worked together and communicating to horrible teams that went in 4 or 5 different directions, everyone shooting a different mech. Out of the D and into the AL and I was thrust into a string of one sided winning matches.

When I got to the CN9-D and Jager A, its like I got put on complete crap teams, not that I was put in a skill bracket too high/good for me, It was the complete opposite of what I was expecting to happen. Its like it puts you with either new people, or people with a lower skill level in order to boost their ELO by putting them with somone that has been doing OK (compared to terrible). I was constantly doing more damage than the rest of my team, sometimes more damage than the rest of my team combined. For the CN9s this past weekend it took me 1h30min to unlock the basics on the A then 2h 27 on the AL. It was almost the same for the Jagers and Stalkers.

I would like to be able to at least see my own ELO score. If I can see the rest of the team's rating that would be great as well, but its as if we aren't allowed to see it because of how broken it is, or there are too few players for it to matter.

That's kinda easily explained. When you first step into a mech class (heavies for example with the Jager), your Elo score is at x, which is potentially a low or near average number.

You then proceeded to win 12 matches, which raises your Elo. Go into the next variant, and it went up more til you reached somewhere around where you were supposed to be, and then you proceeded to have "balanced skill" matches in your last variant.

[Rekoro]

You are basicly saying, every mech got its own elo. asuming this is true, that would be just bad, so i dont think this is the case. if they use common sense, you gain elo for your account not for a single mech. For example in League of Legends this would mean every hero has his own elo oO . have you something to back this up? because i call BS at this.

Edited by Rekoro, 01 May 2013 - 10:28 AM.