202 replies to this topic

[Felicitatem Parco]

Please resolve this for me:

If your Elo Rank it too high, then you get mixed with crappy players who can't fight worth crap and shoot LRMs into walls at point-blank,

AND

If your Elo Rank is too high, then you only play with Meta Elitists and the games become overly competitive and one-dimensional with precision sniping being the norm.


So, which is it?

[Felicitatem Parco]

Prosperity Park, on 06 April 2014 - 08:53 AM, said:

Please resolve this for me:

If your Elo Rank it too high, then you get mixed with crappy players who can't fight worth crap and shoot LRMs into walls at point-blank,

AND

If your Elo Rank is too high, then you only play with Meta Elitists and the games become overly competitive and one-dimensional with precision sniping being the norm.


So, which is it?

Oh, yeah, also please try to work-in the facts that Elo Ranks range from 0<->2300, and that your Elo Rank can only change by +/- 50 in a single match, for those who think there is a system that automatically matches you with losers when you win a couple games.

Edited by Prosperity Park, 06 April 2014 - 09:00 AM.

[Adiuvo]

Prosperity Park, on 06 April 2014 - 08:53 AM, said:

Please resolve this for me:

If your Elo Rank it too high, then you get mixed with crappy players who can't fight worth crap and shoot LRMs into walls at point-blank,

AND

If your Elo Rank is too high, then you only play with Meta Elitists and the games become overly competitive and one-dimensional with precision sniping being the norm.


So, which is it?

It's both. 'High Elo' doesn't consist of only high Elo players. People in trial mechs are frequently pulled in to balance things out.

[aniviron]

Profiteer, on 05 April 2014 - 03:03 AM, said:

So I play a bunch of games and stomp the enemy 12-0, 12-1. It's fun for a few games, then boring. (zero challenge)

All of a sudden the MM decides I've won too much and puts our 4-man with 8 complete newbs. No amount of carrying is enough, we get smashed.

Rinse and repeat.

I'm lucky to get 1-2 games out of ten that resemble anything like a good fight.



I DON'T WANT TO PLAY WITH OR AGAINST NEWBS!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!



I've been playing over a year and consider myself a good pilot. I should not have to put up with casuals, newbs, etc (no offense) in my games. It's BS.

There should be tiers: Soldier, Veteran, Elite, to separate players.

Don't give me "there aren't enough people playing", if the community is that small is time to close shop.

I just want good fights with/against people of a similar skill level as me.

You poor guy, it must be hard with your four-man.

Imagine this, but with solo dropping. It happens, and it's awful and completely unfun. At least in the rare instance I can get three of my friends to play this game, it's fun doing well as a lance, even if the team is awful. Dropping alone is hell.

[Davers]

MWn00b, on 06 April 2014 - 05:19 AM, said:

I think there should be a choice of gameplay added before you go into match - an Arcade gameplay, or Hardcore gameplay.
If you don't feel like getting into serious, thoughtful gameplay you go Arcade. If you want some serious stuff with tactics and players that share your current approach you select Hardcore. Maybe add some other differences to those modes to make them more unique and matching the game style.

Now if only they would get around to putting a Hardcore mode in.

People don't understand math and think the MM is some kind of magic. Most people think they are 'above average', but statistically that can't be true.

[Ordellus]

Abivard, on 05 April 2014 - 05:40 PM, said:

The MM certainly does not decide to place one player on the 'losing' side and another on the 'winning' side.

Actually it does.

Once you get over 1.00 win/loss, it places you in a situation in which you are expected to lose.

We can argue that having a 88% chance of losing =/= "being placed on the losing side", but that's pretty much what it means.

Abivard, on 06 April 2014 - 01:57 AM, said:

Taken out of context.

Except it's not.

Abivard, on 06 April 2014 - 01:57 AM, said:

What It does not do is pick one team as the winning team,with which MM will stack with the players it wants to be winners, and one team MM wants to lose so it is assigns players MM wants to lose this game.

Actually, that's essentially what it does.

[Ordellus]

Prosperity Park, on 06 April 2014 - 09:00 AM, said:

Oh, yeah, also please try to work-in the facts that Elo Ranks range from 0<->2300, and that your Elo Rank can only change by +/- 50 in a single match, for those who think there is a system that automatically matches you with losers when you win a couple games.

So if you have a player with a 2000 score and you put him with eleven 100 scores, you have an average team score of 258.

Put them against a team composed of twelve people with a score of 258 and guess which teams wins.

That's right, the 100s will die off leaving the one good guy to fight 6v1....

But how can that be? The elo average meant each team had a 50/50 chance of winning.....

Surely the system intended for a 1v1 game in which both players must use
the exact same rules,
with the exact same pieces,
with the exact same capabilites,

Can directly compare to a 12v12 in which all players use
vastly different pieces
with vastly different capabilites
that operate on vastly different rulesets....(don't try to pretend missles and lasers use the same ruleset)

[Roland]

Ordellus, on 06 April 2014 - 04:02 PM, said:

Actually it does.

Once you get over 1.00 win/loss, it places you in a situation in which you are expected to lose.

This is not even remotely how it works.

There isn't even any way that it COULD work. How could the matchmaker put everyone who had a positive win/loss ratio in games where they are expected to lose? By picking people with a negative win/loss ratio? And those people would somehow be favored to win?

No, that is quite obviously not how it works.

Winning merely increases your elo rating, and that rating is used to try and find opponents who have an equal rating.

There are multitudes of reasons who elo is poorly suited to this type of game, and doesn't work very well, but your description is clearly wrong.

[WarZ]

Roadkill, on 05 April 2014 - 04:51 AM, said:

Wrong. And more importantly, mathematically proven to be wrong.

Elo works just fine in multiplayer games, and has been repeatedly proven to do so. It just takes longer for your rating to stabilize.

The problem with MWO is that the matchmaker uses too wide of a rating spread when creating matches, which is necessary to create matches in a reasonable amount of time because the player base is so small.

So your quoted player is right, and you're wrong. The elo system is broken and producing bad matchmaking.

Rationalizing how it might be happening, does not excuse the fact that it is happening. Most matches. All day long.

[Abivard]

Jman5, on 06 April 2014 - 08:04 AM, said:

Let's be crystal clear here so people don't misinterpret what you're trying to say. The matchmaker is not looking at your Win/Loss Ratio and trying to adjust your games to keep you at 1.0 WLR. The Matchmaker does not care what your WLR is. Win/Loss is merely a side-effect of Elo Matchmaking. The matchmaker is trying to find you games with players who have Elo as close to yours (or your premade's average) as possible.

Of course finding your team's identical twin rarely happens so MM gets it close and then predicts which side will win based on team's average Elo. If the predictions come true, everyone's Elo does not move. If it does not go as predicted, you see shifts in Elo score. Victorious underdogs go up, losers go down. Now the underdogs face slightly better players, and better players until you start losing. Hence why for most players WLR will head toward 1.0.

Unfortunately Elo does have limits. At the very high and very low ends, there is simply no where left to go. If there are players who regularly win against the tip top or regularly lose against the bottom barrel, You're stuck with that. At the extremes you get really wacky WLR.

A couple myths here.
MM does not match your Elo with a player of a similar Elo, instead it uses your Elo to get an Average for the team that is close to the average for the other team. When rating premades it averages the premades Elo, adds a insignificant bonus, then uses that as the Elo for each of them.

This would work ok sorta maybe IF the player base was HUGE, I mean 100's of THOUSANDS of players looking for matches at the same time. But this is not the case with MWO. So it will grab solo players with Hi Elo, match them with one or MORE players of relatively low Elo to average them out to be the same as an equal amount of individuals whose true Elo is the Average for that Drop.

Your Elo will ALWAYS go up or down, the amount at which it does fluctuate is derived by the difference in the TEAM's AVERAGED Elo. Your Elo only counts as an averaged score. This is the only point at which any 'Prediction' is made. It is a prediction of how much you will go up if you win, and how much you will go down if you lose. That amount is written in stone, based off a mathematical equation using an Elo type system.

But of course Elo doesn't work in a randomly assigned team mutli-player game, and the matchmaker ignores every other facet of matchmaking or team balance outside of the Elo rating it assigned player's.

Shit in, shit out. No surprise matches are generally not balanced outside of the few that randomly occur due to simply the odds of chance and not MM's efforts.

[Targetloc]

WarZ, on 06 April 2014 - 04:28 PM, said:

So your quoted player is right, and you're wrong. The elo system is broken and producing bad matchmaking.

Rationalizing how it might be happening, does not excuse the fact that it is happening. Most matches. All day long.

I'd accept that maybe the matchmaking portion of the system is broken, or that the population might be too small to make decent matches at some levels, but Elo itself isn't the problem.

It's proven to work in other games, and mathematically sound regardless of team size.

These threads are really confusing when half the participants are railing about the evils of Elo when it's the one part of the system that's easy to understand and proven to work.

The common thread between games that use Elo or a very Elo-like system seems to be this:

There are players that whine about Elo spend most of their time trying to blame their problems on 'Elo hell' or a matchmaker that is trying to force them to lose.

Then there are players that don't whine about Elo because they're too busy trying to figure out what they could have done differently to force a win in that last match, or just shrugging and playing more, and for some strange reason they seem to get better matches with 'evenly skilled' players.

[Ordellus]

Roland, on 06 April 2014 - 04:21 PM, said:

There isn't even any way that it COULD work. How could the matchmaker put everyone who had a positive win/loss ratio in games where they are expected to lose? By picking people with a negative win/loss ratio? And those people would somehow be favored to win?

By tracking win/loss ratio - which it does...
By tracking elo - which it does.....
By creating a team of players that need to lose against a team of players that need to win...

Simply choose the 'need to win' team from a higher elo pool than the 'need to lose' team..

Ta Da. The "impossible" situation just happened.

If you somehow magically win the above match, your elo goes up and they go on pretending that the system works.

If you get stomped (as the MM is designed for) your win/loss evens out, and they go on pretending the system works.

You done throwing logic out the window, just so you can pretend that you aren't just stubborn?

Edited by Ordellus, 06 April 2014 - 06:10 PM.

[Ordellus]

Targetloc, on 06 April 2014 - 05:48 PM, said:

It's proven to work in other games, and mathematically sound regardless of team size.

"it works in other games" means absolutely nothing.... unless of course those other games are twitch based shooters, in which each player customizes their combatant, and teamwork only exists for a small group of people that use third party software...

Since that's not the case, that argument is completely useless.

Just because there isn't a factor for team size doesn't mean "it's sound for team size"....

That actually means they never bothered to consider that b/c it was never meant to be used for a team game.

And if you've ever actually looked at the elo formula you'd know it's terrible for this sort of game..... seeing as there are only a couple factors...

1) Supposed skill
2) Arbitrary constant (hint: these are generally a sign that the theory is majorly flawed)
3) Wins
4) Losses

You'll notice that nowhere in that list did you see:

5) Anything about the other 11 guys on your team
6) The "who gets to use voice comms and who doesn't" factor
7) The fact that each mech (and thus the game for those players) operates under dozens of variables that don't hold from player to player (aka not everyone has to follow the same rules)

+more things I'm too lazy to think of now...

Or to summarize: Elo is a terrible system to try and force onto this type of game b/c that's not what it was created for

Edited by Ordellus, 06 April 2014 - 06:20 PM.

[Roland]

Dude, you seriously have no idea how elo works, do you?
The system isn't trying to make you lose. It's just matching you against people who it thinks have the closest rating to you. It's actually trying to create a situation where it's as close to a 50/50 matchup as possible.

There are all kinds of reasons why it doesn't work right, but your understanding of how it works is not even close to correct.

[Targetloc]

Abivard, on 06 April 2014 - 05:24 PM, said:

A couple myths here.
MM does not match your Elo with a player of a similar Elo, instead it uses your Elo to get an Average for the team that is close to the average for the other team. When rating premades it averages the premades Elo, adds a insignificant bonus, then uses that as the Elo for each of them.

This would work ok sorta maybe IF the player base was HUGE, I mean 100's of THOUSANDS of players looking for matches at the same time. But this is not the case with MWO. So it will grab solo players with Hi Elo, match them with one or MORE players of relatively low Elo to average them out to be the same as an equal amount of individuals whose true Elo is the Average for that Drop.

Pretty sure this is wrong. Granted how the matchmaker actually builds the teams is one of the most poorly explained parts of the whole system, but given Paul's discussion about 'buckets' and matchings having an expanding 'acceptable Elo' the longer it takes to fill a drop the effective way it works is more likely this:

1) Matchmaker checks if any games are possible accepting unmatched players 1600-1900 Elo.
2) Not enough players found after 1 minute of trying
3) Matchmaker checks if any games are possible accepting unmatched players in 1500-2000 Elo.
4) 24 players found. Awesome. Grab them out of the queue.
5) Alternate assigning pre-made groups to teams, then alternate assigning lone wolves to teams.
6) Game happens. Team 2 wins.
7) Calculate Team 2's average elo. Calculate Team 1's average Elo. Adjust everyone based on expected outcome.

Anything else would be needlessly complicated, and the idea of purposely pulling in low Elo players to average out high Elo players doesn't match what we know about the 'acceptable Elo' expanding over time.

The game doesn't try to get you to 50/50 win loss. The title of this thread is completely out to lunch on that. It tries to match you against players in the same skill bracket, and continually moves that bracket up every time you win. Which means, unless you are literally the best player in the game, with no possible equal, your W/L will eventually go down to 50/50.

[anonymous161]

Why is this such a constant complaint? Of course you will get paired with new ppl once in a while. Why is that such a big deal? Why dont you kids take a break if you are playing a game so much that you find things to complain that really dont matter in the long run I'd say you need to add a couple more hobbies or get a real career going.

[Targetloc]

Ordellus, on 06 April 2014 - 06:09 PM, said:

By tracking win/loss ratio - which it does...
By tracking elo - which it does.....
By creating a team of players that need to lose against a team of players that need to win...

Simply choose the 'need to win' team from a higher elo pool than the 'need to lose' team..

Ta Da. The "impossible" situation just happened.

If you somehow magically win the above match, your elo goes up and they go on pretending that the system works.

If you get stomped (as the MM is designed for) your win/loss evens out, and they go on pretending the system works.

You done throwing logic out the window, just so you can pretend that you aren't just stubborn?

That's a really hilarious way of describing how a bracketed tournament works.

You win a whole bunch of times, so they put you up against guys from the other side that are also winning too much. And then you lose. Because the system wanted you to.

Someone should tell Sweden they need to start whining on the hockey forums until the Olympic committee fixes this.

[Ordellus]

Roland, on 06 April 2014 - 06:14 PM, said:

Dude, you seriously have no idea how elo works, do you?
The system isn't trying to make you lose. It's just matching you against people who it thinks have the closest rating to you. It's actually trying to create a situation where it's as close to a 50/50 matchup as possible.

There are all kinds of reasons why it doesn't work right, but your understanding of how it works is not even close to correct.

It matches you against a team whose SUM is closest to your teams SUM.....

http://en.wikipedia....o_rating_system

I know exactly how the formula works, it's barely algebra.

Also, EXTREMELY easily abused as shown directly above your post.... please try reading before commenting

Edited by Ordellus, 06 April 2014 - 07:02 PM.

[Roland]

Yes, one of the critical faults of the implementation of matchmaking in this game is the fact it tries to average out elo ratings...

However, your notion of it trying to make you lose is clearly false.

[Deathlike]

aniviron, on 06 April 2014 - 12:02 PM, said:

You poor guy, it must be hard with your four-man.

Imagine this, but with solo dropping. It happens, and it's awful and completely unfun. At least in the rare instance I can get three of my friends to play this game, it's fun doing well as a lance, even if the team is awful. Dropping alone is hell.

I will say that going solo with high Elo is asking for trouble. It will pull out more of the "underhive" than you'd want.

Going with 3 other people, you'll have "the group average" to save you a bit, but yet on the other hand, the roflstomps tend to go both ways with indifference to much generally speaking. I guess losing as a team is fine... as long as I'm not stuck on Mordor... (except there was this one night well I ended the night playing Mordor the last 6 out of 7 matches).

Roland, on 06 April 2014 - 07:07 PM, said:

Yes, one of the critical faults of the implementation of matchmaking in this game is the fact it tries to average out elo ratings...

However, your notion of it trying to make you lose is clearly false.

It's more like "how likely is it going to troll you (aka lopsided match, where you or the opposing team fought generally unopposed)?"

The answer is "more often than not".

Edited by Deathlike, 06 April 2014 - 07:26 PM.