78 replies to this topic

[RichAC]

Flyby215, on 07 January 2014 - 11:17 AM, said:

I would tend to think the MM is simply hell-bent on giving players 1:1 W/L records. I was on a 7-match straight winning streak when I landed in a match where I managed 300-something damage, but not a single teammate could get above 100, 0-12 loss.

The following match we ended up with 1 assault, 1 heavy, the rest assorted mediums/lights vs 9 assault mechs (I posted the pic in another thread). 1-12 loss.

This trend generally continued until I was down to about 8-6 W-L for my evening gaming session.

MM is doing its job then. Like the above poster says though, good teamwork definitely helps keep it over 1.0

Edited by RichAC, 10 January 2014 - 10:34 AM.

[Khobai]

Quote

Its a numbers game. Once your team is down (or up) by 2-3 mechs things tend to snowball.

This. Lack of respawns causes an avalanche effect. Generally once a team gets down by 4 mechs theres almost no way they can come back.

[Nicholas Carlyle]

RichAC, on 10 January 2014 - 10:33 AM, said:

MM is doing its job then. Like the above poster says though, good teamwork definitely helps keep it over 1.0

Guess you don't contain you bad posts to one thread.

That is NOT the job of a matchmaker.

A matchmaker should be doing it's best to create situations where matches end 12-8, 12-9, etc etc.

When matches are ending 1-12, 4-12 that is NOT fun, and NOT good matchmaking. It creates situations where new players say "Wtf, this game is dumb" and leave. And it leaves experienced players shaking their heads wondering why they got matched with people who are obviously not on the level of their opponents.

Striving for a 1:1 W/L record is not the goal of a good matchmaker.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 10:38 AM, said:

Guess you don't contain you bad posts to one thread.

That is NOT the job of a matchmaker.

A matchmaker should be doing it's best to create situations where matches end 12-8, 12-9, etc etc.

When matches are ending 1-12, 4-12 that is NOT fun, and NOT good matchmaking. It creates situations where new players say "Wtf, this game is dumb" and leave. And it leaves experienced players shaking their heads wondering why they got matched with people who are obviously not on the level of their opponents.

Striving for a 1:1 W/L record is not the goal of a good matchmaker.

Sorry Nick but expecting a MM to mitigate rolls is a fallacy. The MM is intended to pair reasonably similar "teams"... What those teams do after that is largely their own doing.

The biggest influence into rolls is "Combat Grouping" or what amounts to a tipping point where team attrition leads to insurmountable odds of overcoming the opposing force due to team attrition.

Proof of this is provided in the fact that 12-man team play, where by selection teams are as balanced as possible and most if not all players subscribe to the forgoing meta.... STILL have outcomes of 1-12, 4-12 rolls... Granted, not as often, but fact remains... it's happens and happens more often than folks appreciate.

IMHO... the likelihood of a 12 to 8 or 12 to 10 is a greater outlier than not because of this singular statistical anomoly.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 10:58 AM, said:

Sorry Nick but expecting a MM to mitigate rolls is a fallacy. The MM is intended to pair reasonably similar "teams"... What those teams do after that is largely their own doing.

The biggest influence into rolls is "Combat Grouping" or what amounts to a tipping point where team attrition leads to insurmountable odds of overcoming the opposing force due to team attrition.

Proof of this is provided in the fact that 12-man team play, where by selection teams are as balanced as possible and most if not all players subscribe to the forgoing meta.... STILL have outcomes of 1-12, 4-12 rolls... Granted, not as often, but fact remains... it's happens and happens more often than folks appreciate.

IMHO... the likelihood of a 12 to 8 or 12 to 10 is a greater outlier than not because of this singular statistical anomoly.

So your example of the match maker being fine is the 12 man queue? REALLY?

So what inside information do you have that...

Quote

where by selection teams are as balanced as possible and most if not all players subscribe to the forgoing meta....

...is actually happening?

Do you have some special insight that says that the matchmaker in 12 mans is better than the matchmaker for PUG's?

You're insinuating that the 12 man queue which we all know has a very small population compared to the very small PUG queue population is matching teams fairly and in the most balanced fashion?

Dear lord.

Edited by Nicholas Carlyle, 10 January 2014 - 11:04 AM.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 11:04 AM, said:

So your example of the match maker being fine is the 12 man queue? REALLY?

You're insinuating that the 12 man queue which we all know has a very small population compared to the very small PUG queue population is matching teams fairly and in the most balanced fashion?

Dear lord.

If a 12 man drops the MM is by default looking to match their composite Elo... which unless the draw pool is too shallow due to available numbers or equitable Elo in that same pool... "yes" 12-mans should in practice have a reasonably matched pairing.

That said... No inside information other than discussion within the 12-man competetive team ranks...

Point is... your blaming the MM exclusively for rolls and I'm telling you the battle of attrition and the numerical tipping point is a greater contribution to this result than the equity of the Elo.

Edited by DaZur, 10 January 2014 - 11:21 AM.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 11:20 AM, said:

If a 12 man drops the MM is by default looking to match their composite Elo... which unless the draw pool is too shallow due to available numbers or equitable Elo in that same pool... "yes" 12-mans should in practice have a reasonably matched pairing.

That said... No inside information other than discussion within the 12-man competetive team ranks...

Point is... your blaming the MM exclusively for rolls and I'm telling you the battle of attrition and the numerical tipping point is a greater contribution to this result than the equity of the Elo.

I'm blaming MM for the amount of rolls.

And I'm saying the GOAL of MM should be to limit rolls, not to create a 1:1 W/L ratio.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 11:24 AM, said:

I'm blaming MM for the amount of rolls.

And I'm saying the GOAL of MM should be to limit rolls, not to create a 1:1 W/L ratio.

How do you expect the MM to control the outcome of a match.. seriously, are you hearing yourself?

I use the 12 man as example because they for lack of a better simile, are practiced "professionals". The fact that they, representing the apex in the Elo tier (presumably) still elicit rolls despite this is proof that the MM cannot reliably control the outcome... It's chaos theory in the finest example.

The MM is supposed to pair equitable teams based on composite Elo... not guarantee the participants take turns killing each other off to avoid the potential for rolls.

Absolutely... Elo should predicate some equity in skill level but outside of that... it's chaos baby!

Edited by DaZur, 10 January 2014 - 11:40 AM.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 11:39 AM, said:

How do you expect the MM to control the outcome of a match.. seriously, are you hearing yourself?

I use the 12 man as example because they for lack of a better simile, are practiced "professionals". The fact that they, representing the apex in the Elo tier (presumably) still elicit rolls despite this is proof that the MM cannot reliably control the outcome... It's chaos theory in the finest example.

The MM is supposed to pair equitable teams based on composite Elo... not guarantee the participants take turns killing each other off to avoid the potential for rolls.

Absolutely... Elo should predicate some equity in skill level but outside of that... it's chaos baby!

It's not asking it to "Control" the outcome.

A matchmaker is supposed to put you in conditions where lets say as a made up number, 70% of your matches are good matches. And 30% of your matches are going to have some random issue (someone's girlfriend stealing his attention, a disconnect, someone playing with a foolish weapons loadout) that creates some kind of roll.

But assuming everyone is playing and trying to win, 70% of your matches you should be very close (12-7 or better).

What I'm seeing and what most posters are seeing, is that is not happening. It's more like 12-1, 12-2, 12-5 wins, then 2-12, 3-12, 4-12 loses creating that magically stupid 1:1 W/L ratio.

It's not about winning or losing, it's about fun, meaningful matches. Which this matchmaker fails at and why most of the guild/mercenary groups have moved on.

It's why we need lobbies.

Edited by Nicholas Carlyle, 10 January 2014 - 12:05 PM.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 12:04 PM, said:

It's not asking it to "Control" the outcome.

A matchmaker is supposed to put you in conditions where lets say as a made up number, 70% of your matches are good matches. And 30% of your matches are going to have some random issue (someone's girlfriend stealing his attention, a disconnect, someone playing with a foolish weapons loadout) that creates some kind of roll.

But assuming everyone is playing and trying to win, 70% of your matches you should be very close (12-7 or better).

What I'm seeing and what most posters are seeing, is that is not happening. It's more like 12-1, 12-2, 12-5 wins, then 2-12, 3-12, 4-12 loses creating that magically stupid 1:1 W/L ratio.

It's not about winning or losing, it's about fun, meaningful matches. Which this matchmaker fails at and why most of the guild/mercenary groups have moved on.

It's why we need lobbies.

The MM is presenting you a roughly 50/50 chance of your team winning by matching your team with a similarly ranked team. There is nothing in that pairing that exclusively predicates your personal win percentage... The notion that the MM should favor your personal success is out and out incorrect.

In fairness it is presuming your Elo + the composite of you team equals that of the opposing team. Let's say you are a 10 and the rest of your team are 7s you composite is 87. You opponent could have an 8 and the rest 7s + a 2... The presumption is that your team should win right? All it takes is for 4 of your team to fall quickly in battle for that tipping point to kick in and screw your presumed win.

The reality is the MM sets the stage for an equitable match, hedging its bet one side should win.... Outside of that it's chaos theory... Anything can and will happen.

Edited by DaZur, 10 January 2014 - 12:35 PM.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 12:29 PM, said:

The MM is presenting you a roughly 50/50 chance of your team winning by matching your team with a similarly ranked team. There is nothing in that pairing that exclusively predicates your personal win percentage... The notion that the MM should favor your personal success is out and out incorrect.

In fairness it is presuming your Elo + the composite of you team equals that of the opposing team. This let's say you are a 10 and the rest of your team are 7s you composite is 87. You opponent could have an 8 and the rest 7s + a 2... The presumption is that your team should win right? All it takes is for 4 of your team to fall quickly in battle for that tipping point to kick in and screw your presumed win.

The reality is the MM sets the stage for an equitable match, hedging its bet one side should win.... Outside of that it's chaos theory... Anything can and will happen.

You seem to be functioning under the assumption I care about win/loss.

If you are reading my posts, you'd understand I'm interested in close, fun games.

Hard fought wins, and heart breaking losses.

When I play a match, and I've ended up with a good amount of damage, and a kill or two, and my team put a good effort in. I don't care if I lose. I had fun playing.

But when I play a match and I did good, and we get rolled because I was with a bunch of derp's against an obviously coordinated pre-made. I get no satisfaction.

The same way when I run a game and get 4 kills and we roll them 12-1. It just means we were better than them in all facets and should not have been matched up. It's fun to kill, but in the end a very empty win and does not entice me to play more.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 12:39 PM, said:

You seem to be functioning under the assumption I care about win/loss.

If you are reading my posts, you'd understand I'm interested in close, fun games.

Hard fought wins, and heart breaking losses.

When I play a match, and I've ended up with a good amount of damage, and a kill or two, and my team put a good effort in. I don't care if I lose. I had fun playing.

But when I play a match and I did good, and we get rolled because I was with a bunch of derp's against an obviously coordinated pre-made. I get no satisfaction.

The same way when I run a game and get 4 kills and we roll them 12-1. It just means we were better than them in all facets and should not have been matched up. It's fun to kill, but in the end a very empty win and does not entice me to play more.

I understand you're looking for hard fought fun / entertaining match results not whether you win or lose... But in fairness you're tying that to the numerical outcome ..i.e. tight game = "fun" versus rolls = "no fun".

I guess I'm not grasping how your presume any aspect of match pairing ensures what you are looking for... It's an attrition anomaly... Something the MM cannot predict nor influence...

You can have a perfectly matched pairing... all players "10"s all running the exact mech and configuration with no consumable influence... and if one team falls short 1/4th of their applied forces, it is almost impossible (had to say almost as I've been party to several rallies)... deter the inevitable outcome of a roll-stomp.

Absolutely... It's presumed equitable pairings will result in that magic hard fought fun / entertaining match... The problem is once the match starts, the MM is no longer part of the equation.

[wanderer]

Nicholas Carlyle, on 10 January 2014 - 12:39 PM, said:

You seem to be functioning under the assumption I care about win/loss.

If you are reading my posts, you'd understand I'm interested in close, fun games.

Hard fought wins, and heart breaking losses.

When I play a match, and I've ended up with a good amount of damage, and a kill or two, and my team put a good effort in. I don't care if I lose. I had fun playing.

But when I play a match and I did good, and we get rolled because I was with a bunch of derp's against an obviously coordinated pre-made. I get no satisfaction.

The same way when I run a game and get 4 kills and we roll them 12-1. It just means we were better than them in all facets and should not have been matched up. It's fun to kill, but in the end a very empty win and does not entice me to play more.

Here's the thing. The most potent modifiers are ones that the matchmaker cannot take into account, since they relate to parameters the game cannot possibly take into account.

That is, premades with communication ability. They usually cannot be seen, and third-party VOIP obviously isn't part of the player profile. One lance that can co-ordinate fire is worth eight 'Mechs that can't, especially when frequently I'll see those 12-1's with multiple 'Mechs one salvo from death- because everyone randomly engaged on one side and the other side decided to all shoot at one target instead with no idea the guy they didn't have locked would have gone down if they just turned and fired at them instead. Faceroll games happen up and down the ELO level- someone with a KDR of .5 will hit them just as much as someone with a KDR of 5.0, a W/L of 50% or 95%.

It's one thing I wish they'd add to the list at the end- the %damage stat for each survivor. People might be a lot more aware when they realize there's half a dozen targets in the "critical damage" range they could have nailed if they just shot at the same target.

Likewise, I wish lance and company commanders could easily mark targets with a single button press. R-lock, hit a button, designate as "lance priority" or "company priority" target.

[Nicholas Carlyle]

wanderer, on 10 January 2014 - 12:58 PM, said:

Here's the thing. The most potent modifiers are ones that the matchmaker cannot take into account, since they relate to parameters the game cannot possibly take into account.

That is, premades with communication ability. They usually cannot be seen, and third-party VOIP obviously isn't part of the player profile. One lance that can co-ordinate fire is worth eight 'Mechs that can't, especially when frequently I'll see those 12-1's with multiple 'Mechs one salvo from death- because everyone randomly engaged on one side and the other side decided to all shoot at one target instead with no idea the guy they didn't have locked would have gone down if they just turned and fired at them instead. Faceroll games happen up and down the ELO level- someone with a KDR of .5 will hit them just as much as someone with a KDR of 5.0, a W/L of 50% or 95%.

It's one thing I wish they'd add to the list at the end- the %damage stat for each survivor. People might be a lot more aware when they realize there's half a dozen targets in the "critical damage" range they could have nailed if they just shot at the same target.

Likewise, I wish lance and company commanders could easily mark targets with a single button press. R-lock, hit a button, designate as "lance priority" or "company priority" target.

I get that we can't see 3rd Party VOIP, but if the matchmaker isn't heavily favoring 4 man pre-mades when it's seeding teams, then once again it is not working properly.

The matchmaker sucks. End of story.

However you want to wrap it up.

And to Dazur, if the matchmaker is doing it's job. You should be seeing more hard fought fun matches than rolls. All evidence points to the opposite.

Therefore the matchmaker is not doing it's job.

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 01:21 PM, said:

If the matchmaker is doing it's job. You should be seeing more hard fought fun matches than rolls. All evidence points to the opposite.

Therefore the matchmaker is not doing it's job.

I'm at a loss...

Let's turn this around... You tell me, how a working MM can do whatever it needs to do to ensure a hard fought fun / entertaining match result and mitigate the opportunity for rolls-stomps.

Edited by DaZur, 10 January 2014 - 02:04 PM.

[Kekkone]

Nick Makiaveli, on 05 January 2014 - 07:54 PM, said:

First, Elo doesn't work simply off winning or losing. It predicts the winner, then if that prediction is wrong, and only if it's wrong, does the Elo score change. So if your team was predicted to lose, and you lose, nothing happens.

Wrong. Maybe correct. PGI's own statements on the subject contradict one another.

PGI version of Elo:

Quote

A player’s rating will only go down if they are beaten by a player who has a lower rating than theirs

Quote

If the Match Maker determines that you're going to win and you actually win, then your Elo score isn't going to change very much (if at all). The same applies to a prediction of loss and you actually lose, your score may drop but it will be slight.

Classic Elo:

Quote

After every game, the winning player takes points from the losing one. The difference between the ratings of the winner and loser determines the total number of points gained or lost after a game. In a series of games between a high-rated player and a low-rated player, the high-rated player is expected to score more wins. If the high-rated player wins, then only a few rating points will be taken from the low-rated player whereas if the lower rated player scores an upset win, then many rating points will be transferred.

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

Edited to correct information.

Edited by Kekkone, 11 January 2014 - 04:29 AM.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 01:30 PM, said:

I'm at a loss...

Let's turn this around... You tell me, how a working MM can do whatever it needs to do to ensure a hard fought fun / entertaining match result.

So you think that if we are dropping two teams of similar skill, with similar mechs...that we should expect more rolls than entertaining matches?

What draws you to that conclusion?

If we have two equally skilled teams with similar mechs and we are still having a lot of rolls, the underlying game is completely broken, and the matchmaker will NEVER work properly because the game is busted.

So are you saying the game is broken then?

[DaZur]

Nicholas Carlyle, on 10 January 2014 - 02:06 PM, said:

So you think that if we are dropping two teams of similar skill, with similar mechs...that we should expect more rolls than entertaining matches?

What draws you to that conclusion?

If we have two equally skilled teams with similar mechs and we are still having a lot of rolls, the underlying game is completely broken, and the matchmaker will NEVER work properly because the game is busted.

So are you saying the game is broken then?

Not broken... simply slave to the end result of force attrition.

With every numerical loss to a team, statistically the opposing team gains momentum in strength of force... which includes the concentration of firepower, ability to focus that firepower and convergence of forces (blob efficiencies).... i.e. the ubiquitous "steamroll".

This is why I consider the more even-handed matches ...i.e the 12 to 9 matches the true outliers.

The reason why these results are far more pervasive is because the statistical pool is only 12 players deep... 1/4th of force loss seems to be the magic tipping point where it's nigh impossible for the dwindling team to hold ground, least of all, recover and push back against the momentum...

If we had 100 players per side, statistically there would be more room for those 10 to 12 ( 60 to 40 ) matches to occur... The force attrition is just too shallow with 12 to allow ebbs and flows...

Edited by DaZur, 10 January 2014 - 02:48 PM.

[wanderer]

DaZur, on 10 January 2014 - 02:27 PM, said:

If we had 100 players per side, statistically there would be more room for those 10 to 12 ( 60 to 40 ) matches to occur... The force attrition is just too shallow with 12 to allow ebbs and flows...

It's notable that when it was 8v8, the "snowball" effect was -worse-. Even losing two 'Mechs early was a strong indication you were gonna lose the match.

[Nicholas Carlyle]

DaZur, on 10 January 2014 - 02:27 PM, said:

Not broken... simply slave to the end result of force attrition.

With every numerical loss to a team, statistically the opposing team gains momentum in strength of force... which includes the concentration of firepower, ability to focus that firepower and convergence of forces (blob efficiencies).... i.e. the ubiquitous "steamroll".

This is why I consider the more even-handed matches ...i.e the 12 to 9 matches the true outliers.

The reason why these results are far more pervasive is because the statistical pool is only 12 players deep... 1/4th of force loss seems to be the magic tipping point where it's nigh impossible for the dwindling team to hold ground, least of all, recover and push back against the momentum...

If we had 100 players per side, statistically there would be more room for those 10 to 12 ( 60 to 40 ) matches to occur... The force attrition is just too shallow with 12 to allow ebbs and flows...

Whether you mean too or not you are agreeing that the game is fundamentally flawed. Whatever bud, enjoy your 12-0 games. I think they suck. Until we get rid of them this game will never flourish.

And it sure as hell won't be an e-sport like they want it to be.