56 replies to this topic

[GalaxyBluestar]

WolvesX, on 17 March 2013 - 06:13 PM, said:

http://mwo.smurfy-ne...b0da1b1a7c92b22

Easy mode for tried, stoned or drunk people. or me.

oh great! you had to spill that secret, everyone used to laugh at me for running such simmilar set up on my stalker, now everyone will flock to it and the splat cat, pop tart, 3ls will be considered balanced. oh well nice to see simillar minds doing well...

http://mwo.smurfy-ne...4092479fa03e1bb

my 5 point bluestar sig build, sniping and standoffs, with tons more armour and std "survivability" engine, though the speed on that cataphract looks really good. it's your choice but yes they're easy to use mechs and weild great results!

Edited by GalaxyBluestar, 17 March 2013 - 09:45 PM.

[flying1ace]

How much C-bills did you get?
Wait, how much EXP did you recieve for that tidy little game?
How do I take screen shots? When I Print Screen, All I get is a black pic.
HELP!

[Targetloc]

Narcisoldier, on 17 March 2013 - 09:34 PM, said:

Yes but that's up to the matchmaker and out of my control. Presumably the matchmaker attempts to match everyone's Elo as closely as possible in every game. So the only action that I can take to get the maximum possible Elo would be to win every game that the matchmaker serves up. If the matchmaker isn't fair, and is in fact systematically (not randomly) matching players against much worse players, then there is a technical problem with the matchmaker.

In other words, you're wrong.

You can't tell what your elo is from number of wins in a game with such a loose matchmaker.

There have been dozens of posts from players claiming 75% win rates complaining about getting trial mechs and dual TAG Stalkers on their teams. Even with groups we can assume to be moderately high ranked, the game dips low for their opponents fairly often.

When you win against underhive opponents you get no points.

When you accidentally lose to underhive opponents because of disconnects or cap rushes or they're just playing a lot better than their rank, you lose a lot of rank.

Elo hell exists because as you approach the far right edge of the curve there are exponentially fewer opponents who are actually worth points to you.

In the worst cases, you'll have to slog through endless matches worth 1 point for a win, just hoping you drop against (and beat) someone who will give you a decent boost... but any unlucky losses will wipe out 20 games worth of progress.

There are games where you can have an 80% win rate and not budge your ranking for weeks. But once you get to that sweet golden tier you can bum along at 50% because you don't pay hardly any points for losing to someone in the same bracket.
It all depends on how the elo system is set up. And we don't have the source code for how PGI does it.

[Revorn]

GalaxyBluestar, on 17 March 2013 - 09:40 PM, said:

http://mwo.smurfy-ne...4092479fa03e1bb

my 5 point bluestar sig build, sniping and standoffs, with tons more armour and std "survivability" engine,

I modified it for you, for some additional Heatsinks and a AMS


http://mwo.smurfy-ne...03b60847010bef0

[Grand Ayatollah Kerensky]

Targetloc, on 17 March 2013 - 10:14 PM, said:

You can't tell what your elo is from number of wins in a game with such a loose matchmaker.

There have been dozens of posts from players claiming 75% win rates complaining about getting trial mechs and dual TAG Stalkers on their teams. Even with groups we can assume to be moderately high ranked, the game dips low for their opponents fairly often.

When you win against underhive opponents you get no points.

When you accidentally lose to underhive opponents because of disconnects or cap rushes or they're just playing a lot better than their rank, you lose a lot of rank.

Elo hell exists because as you approach the far right edge of the curve there are exponentially fewer opponents who are actually worth points to you.

In the worst cases, you'll have to slog through endless matches worth 1 point for a win, just hoping you drop against (and beat) someone who will give you a decent boost... but any unlucky losses will wipe out 20 games worth of progress.

There are games where you can have an 80% win rate and not budge your ranking for weeks. But once you get to that sweet golden tier you can bum along at 50% because you don't pay hardly any points for losing to someone in the same bracket.
It all depends on how the elo system is set up. And we don't have the source code for how PGI does it.

You don't understand what I am trying to say. If the matchmaker is matching you against better players as often as it matches you against worse players then that variable has been sufficiently randomized.

[Jonathan Paine]

Dropped on the same side as Narcisoldier earlier. I think the matchmaker was confused and should have put us on opposing sides. The match wasn't even close to being close.
Some random sub routine with one mixed up variable. Would explain the inexplicable matches where one side has all trial mechs and the other a bunch of psychos:-)

[Alexandrix]

Targetloc, on 17 March 2013 - 10:14 PM, said:

You can't tell what your elo is from number of wins in a game with such a loose matchmaker.

There have been dozens of posts from players claiming 75% win rates complaining about getting trial mechs and dual TAG Stalkers on their teams. Even with groups we can assume to be moderately high ranked, the game dips low for their opponents fairly often.

When you win against underhive opponents you get no points.

When you accidentally lose to underhive opponents because of disconnects or cap rushes or they're just playing a lot better than their rank, you lose a lot of rank.

Elo hell exists because as you approach the far right edge of the curve there are exponentially fewer opponents who are actually worth points to you.

In the worst cases, you'll have to slog through endless matches worth 1 point for a win, just hoping you drop against (and beat) someone who will give you a decent boost... but any unlucky losses will wipe out 20 games worth of progress.

There are games where you can have an 80% win rate and not budge your ranking for weeks. But once you get to that sweet golden tier you can bum along at 50% because you don't pay hardly any points for losing to someone in the same bracket.
It all depends on how the elo system is set up. And we don't have the source code for how PGI does it.

Well,If I had to make a guess i'd say the first few lines look something like :
"Derp a derpderp = Herp a derp"

[Grand Ayatollah Kerensky]

Jonathan Paine, on 17 March 2013 - 11:13 PM, said:

Dropped on the same side as Narcisoldier earlier. I think the matchmaker was confused and should have put us on opposing sides. The match wasn't even close to being close.
Some random sub routine with one mixed up variable. Would explain the inexplicable matches where one side has all trial mechs and the other a bunch of psychos:-)

I'm dropping with a well balanced 4 man and we're cleaning up.

[5th Fedcom Rat]

Narcisoldier, on 17 March 2013 - 11:20 PM, said:

I'm dropping with a well balanced 4 man and we're cleaning up.

Ooh ooh let me guess... a Raven 3L, Stalker LRM Boat, Atlas DDC, and Poptart 3D or Splat Cat. Yes, such groups are very well "balanced" indeed.

[Grand Ayatollah Kerensky]

5th Fedcom Rat, on 17 March 2013 - 11:27 PM, said:

Ooh ooh let me guess... a Raven 3L, Stalker LRM Boat, Atlas DDC, and Poptart 3D or Splat Cat. Yes, such groups are very well "balanced" indeed.

I was refering to the skill level of each member of the group. We only have one raven 3l, no LRMs, a raven 4x, an atlas d and a stalker 5m.

[Quxudica]

Narcisoldier, on 17 March 2013 - 07:09 PM, said:

The Elo of a group is an average of all the members of the group. So if you have three group members with, say, an Elo of 3000 each and one group member with an Elo of , say, 500 that means that your group's Elo will be 2375. Now the system is going to place you in a match with a target Elo at around 2375. So now you're in a high level match and you've brought along dead weight. Your 500 Elo group member is going to be completely useless and you might as well have dropped one man short. The only hope you have is that the other team also has a premade that has also dragged a useless player into the high level match.

The way around this is to match the skill level of each member of the premade so that there is no dead weight. This simple trick will enable you to go back to your pre-elo pugstomping since good pugs who aren't talking are far more likely to be rolled by a good premade that is talking.

Elo broke casual 4 man pugstomping but if you are mindful of the skill level of each member of your four man, you can still stomp the crap out of pugs and harvest their sweet tears of rage and disappointment.

If some is putting in that much effort just to fight easy groups with little challenge, I kind of feel sorry for them. Seems like it's missing the point of the game to me.

Edited by Quxudica, 17 March 2013 - 11:55 PM.

[Grand Ayatollah Kerensky]

Quxudica, on 17 March 2013 - 11:53 PM, said:

If some is putting in that much effort just to fight easy groups with little challenge, I kind of feel sorry for them. Seems like it's missing the point of the game to me.

The idea is to fight against people with the same Elo as you so that you can capitalize on the advantage you get from using teamwork.

[MischiefSC]

Narcisoldier, on 17 March 2013 - 10:59 PM, said:

You don't understand what I am trying to say. If the matchmaker is matching you against better players as often as it matches you against worse players then that variable has been sufficiently randomized.

No, here's the problem.

If you generally drop with premade teams and you win a lot your Elo will be pretty high. What this means is that it'll have to cast a wider net to get enough people with as close a weight match as possible. Since there are fewer people above you than below you this means you'll get matches weighted with lesser and lesser skilled people - except when you get pulled into matches against people much better than you who crush you easily.

The wins against the inferior people? Little gain. The losses to people much better than you? Little if any loss.

So this is Elo hell. Too few people in your own skill range for the 50/50 matching to work out. Most of your games (and nearly all your wins) will be against people further down the scale which results in little improvement. Even worse is when you lose those games - you take a hefty Elo hit for losing to someone it expected you to beat. Conversely the people up the Elo ladder from you have it even worse. The higher up you get the thinner the ranks get the harder it is to see real movement without beating teams who are ranked considerably higher than yourself.

Does that make sense? If you're high up the Elo ladder you pretty much have to drop in premades and bring your A game to advance. Otherwise even with a positive win/loss ratio you're probably just treading water - at best.

[Xetherius]

The main issue I've got with ELO is that they've removed weight matching for the most part. I dropped in several games earlier where it was especially prevalent. The worst one was where our team consisted of 3 lights, 4 meds, and 1 heavy. The other team was 3 Atlas, 2 Stalkers, 2 Awesome and a 'Phract. We got wiped hardcore. Lack of weight matching also has a tendancy to mess up ECM distribution between teams, due to ECM being pretty much all lights or D-DCs.

Personally, I find this disappointing (although there are some really good matches to be had) but hopefully it will be fixed around CW hits.

Still:

[GalaxyBluestar]

Revorn, on 17 March 2013 - 10:26 PM, said:

I modified it for you, for some additional Heatsinks and a AMS


http://mwo.smurfy-ne...03b60847010bef0

thanks i never thought to modify anymore {ran out of money when building it} i just leveled it up, moved on and never thought about it again, that or i just forgot what i have in my mechlab and looking at it all those empty spaces, i know i didn't bother with ams but hey i never keep track of anything. thatnks for the reminder jolt, i needed that!

Edited by GalaxyBluestar, 18 March 2013 - 12:34 AM.

[Grand Ayatollah Kerensky]

MischiefSC, on 18 March 2013 - 12:01 AM, said:

No, here's the problem.

If you generally drop with premade teams and you win a lot your Elo will be pretty high. What this means is that it'll have to cast a wider net to get enough people with as close a weight match as possible. Since there are fewer people above you than below you this means you'll get matches weighted with lesser and lesser skilled people - except when you get pulled into matches against people much better than you who crush you easily.

The wins against the inferior people? Little gain. The losses to people much better than you? Little if any loss.

So this is Elo hell. Too few people in your own skill range for the 50/50 matching to work out. Most of your games (and nearly all your wins) will be against people further down the scale which results in little improvement. Even worse is when you lose those games - you take a hefty Elo hit for losing to someone it expected you to beat. Conversely the people up the Elo ladder from you have it even worse. The higher up you get the thinner the ranks get the harder it is to see real movement without beating teams who are ranked considerably higher than yourself.

Does that make sense? If you're high up the Elo ladder you pretty much have to drop in premades and bring your A game to advance. Otherwise even with a positive win/loss ratio you're probably just treading water - at best.

Ok here's a little scenario and I'd like you to answer the three questions.

Jim has won 0 out of 100 matches
Sally has won 100 out of 100 matches

Who has the higher Elo?
What can Sally do to INCREASE her Elo?
What can Jim do to DECREASE his Elo?

Edited by Narcisoldier, 18 March 2013 - 12:44 AM.

[Egomane]

Please continue your Elo discussion in the officiall matchmaker phase 3 feedback thread.
http://mwomercs.com/...making-phase-3/

Thank you and closed!