Jump to content

Elo Has Been Replaced !

News

263 replies to this topic

#41 Gas Guzzler

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 14,256 posts
  • LocationCalifornia Central Coast

Posted 17 August 2015 - 03:04 PM

View PostFupDup, on 17 August 2015 - 03:02 PM, said:

Bro, do you even Novatart? :P


Well yeah, but I believe his intent with "Those who are good at gundam wing online" were those who were proficient with meta mechs but didn't have amazing skills throughout the rest of the game, or something along those lines.

Since "gundam wing" isn't the meta anymore...

#42 ScarecrowES

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 2,812 posts
  • LocationDefending the Cordon, Arc-Royal

Posted 17 August 2015 - 03:29 PM

View Postbad arcade kitty, on 17 August 2015 - 03:03 PM, said:


can you explain me, if they can make a super objective hidden score why the visible match score is so crappy


Baby steps. Baby steps. We're just NOW getting a real stat-based skill system. With expanded stat categories, PGI will have a better idea of how skill relates to match performance and outcome, and can adjust the rewards systems accordingly... so that players are actually being rewarded for how they play.

All in time though. Baby steps.

#43 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 17 August 2015 - 03:30 PM

View PostThisMachineKillsFascists, on 17 August 2015 - 03:01 PM, said:

5 tiers


Right five tiers but separated by what? Just overall?

I mean, what if I suck donkey-crap in a BJ-1 and decide to take it for a spin. I don't want it to be mixed in with my TimberWhale match scores.

The score used should be independent for each chassis we bring so we don't get punished for taking an underhive non-GGnoscope420 'mech.

Edited by Mister Blastman, 17 August 2015 - 03:31 PM.


#44 Mechwarrior Buddah

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 13,459 posts
  • LocationUSA

Posted 17 August 2015 - 03:30 PM

View PostGas Guzzler, on 17 August 2015 - 03:28 PM, said:


Well I thought a lot of the community referred to flying mechs as Gundams, and poptarting was the primary reason why, because JJs made mechs more like Gundams than BattleMechs, but whatever it doesn't really matter.


a lot of the community refer to then as robots too -.-

#45 oneproduct

    Member

  • PipPipPipPipPipPip
  • 213 posts

Posted 17 August 2015 - 03:31 PM

The problem with this PSR is that it introduces multiple points of failure, in that every action that you receive match points for could be not valued correctly. Are we going to have threads like "TAG assist should be worth 5 match score instead of 2 so that I get a better PSR rating"?

Win/loss is not subject to faults like this. If you're in a group that carries you, it's still very easy for you to get a high match score as you deathball over a routing enemy that doesn't put up good resistance, so it still has the same possible problem. In particular, light mechs tend to get much higher scores during stomps as they have the speed to take advantage of out of place/unsupported enemies.

With win/loss if you had a group of 11 good players and 1 bad, they would lose against teams of 12 good people, so their Elo would still be beneath a good group of 12. Then if you're that 1 bad, anytime you play outside of your group of 11 good friends your Elo would go down without them there to support you, so it corrects itself.

If you play with better people, you're going to get a better match score because you'll tend to win more, so you'll still get carried. That won't change.

I don't think this change is going to have the effect people desire and we've introduced potential balance problems with the amount of match score actions give.

Also there's generally a large disparity in match score between the weight classes as damage/fighting tends to offer higher rewards than anything else.

Edited by oneproduct, 17 August 2015 - 03:34 PM.


#46 Nothing Whatsoever

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 3,655 posts
  • LocationNowhere

Posted 17 August 2015 - 03:32 PM

View PostMister Blastman, on 17 August 2015 - 03:30 PM, said:


Right five tiers but separated by what? Just overall?

I mean, what if I suck donkey-crap in a BJ-1 and decide to take it for a spin. I don't want it to be mixed in with my TimberWhale match scores.

The score used should be independent for each chassis we bring so we don't get punished for taking an underhive non-GGnoscope420 'mech.


It's gonna be just one PSR score: http://mwomercs.com/...ost__p__4627627

#47 Mister Blastman

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Survivor
  • Survivor
  • 8,444 posts
  • LocationIn my Mech (Atlanta, GA)

Posted 17 August 2015 - 03:33 PM

View PostPraetor Knight, on 17 August 2015 - 03:32 PM, said:


It's gonna be just one PSR score: http://mwomercs.com/...ost__p__4627627


*sigh*

That sucks.

#48 Deathlike

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 29,240 posts
  • Location#NOToTaterBalance #BadBalanceOverlordIsBad

Posted 17 August 2015 - 03:35 PM

View PostGas Guzzler, on 17 August 2015 - 03:01 PM, said:


Its actually not because poptarting (gundam wing online) isn't the thing any more.

You would have to replace it with something like alpha vomit online and then it would apply again.


He mostly called mechs "gundams".

I don't think his terminology was strictly what it was... but more reflective of what was the case at the time.

#49 Nothing Whatsoever

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 3,655 posts
  • LocationNowhere

Posted 17 August 2015 - 03:36 PM

View PostMister Blastman, on 17 August 2015 - 03:33 PM, said:


*sigh*

That sucks.


Yeah, but on the flip side, I wonder if the devs are planning a synergy with the other system they are working on dealing with the Battle Value for mechs.

I hope, they are just rolling this phase out first to stress test it, as painful as it sounds being a guinea pug!

#50 ScarecrowES

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 2,812 posts
  • LocationDefending the Cordon, Arc-Royal

Posted 17 August 2015 - 03:37 PM

View Postoneproduct, on 17 August 2015 - 03:31 PM, said:

The problem with this PSR is that it introduces multiple points of failure, in that every action that you receive match points for could be not valued correctly. Are we going to have threads like "TAG assist should be worth 5 match score instead of 2 so that I get a better PSR rating"?

Win/loss is not subject to faults like this. If you're in a group that carries you, it's still very easy for you to get a high match score as you deathball over a routing enemy that doesn't put up good resistance, so it still has the same possible problem. In particular, light mechs tend to get much higher scores during stomps as they have the speed to take advantage of out of place/unsupported enemies.

With win/loss if you had a group of 11 good players and 1 bad, they would lose against teams of 12 good people, so their ELO would still be beneath a good group of 12. Then if you're that 1 bad, anytime you play outside of your group of 11 good friends your ELO would go down without them there to support you, so it corrects itself.

If you play with better people, you're going to get a better match score because you'll tend to win more, so you'll still get carried. That won't change.

I don't think this change is going to have the effect people desire and we've introduced potential balance problems with the amount of match score actions give.

Also there's generally a large disparity in match score between the weight classes as damage/fighting tends to offer higher rewards than anything else.


I've put in a lot of posts in other threads that describes exactly how Elo is set up for MWO and why it doesn't work. Yours is a common misconception that's easily proven wrong, but I'm not typing it all again right now. Suffice it to say, Elo doesn't work for skill ratings for individual players in team games, and as implemented in MWO, your Elo score actually stays pretty much exactly the same as it was in the beginning with little deviation, no matter how well or poorly you actually play. It's a flawed system implemented poorly.

The new system will much better reflect individual skill.

#51 Mawai

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,495 posts

Posted 17 August 2015 - 03:41 PM

View PostSpheroid, on 17 August 2015 - 02:13 PM, said:

I wonder about the match time waits. The current throws random collections of five tiers together. The new system will exclude Tier 4 and 5 from matching with Tier one. As a percentage isn't that a massive reduction in possible combinations?

I assume player skill is normally distributed with the mean being Tier 3.


We can only hope it works.

PGI is notoriously bad at wishful thinking rather than actual analysis. They originally thought that enforcing a 3/3/3/3 mech distribution would be fine without realizing that queue times for any mech with more than 25% would be over an hour within a very short period of time (this was trivially easy to show with a very simple analysis).

Their initial response when it went live on test and the queues were immense was that "people will adjust and play different mechs" which changed to "we need to pull and rewrite this matchmaker ASAP".

So, we will have to wait and see how it all works out. :)

#52 ScarecrowES

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 2,812 posts
  • LocationDefending the Cordon, Arc-Royal

Posted 17 August 2015 - 03:43 PM

View PostGas Guzzler, on 17 August 2015 - 03:35 PM, said:

Yeah that was a bit disappointing. But its no different from the situation where your medium class elo could be split between Vindicators and Stormcrows..


It's actually good for the player community as a whole, really - having your skill score balanced out between your great performances in good, properly-leveled and equipped mechs, and your poor performances in new, unleveled and poorly equipped mechs.

Let's say you're awesome with your mastered Timberwolves, but terrible with brand new Hunchbacks. Ultimately, your new skill score should drop a bit because of your play in the Hunchbacks. When you're piloting your Hunchies, you should find that you're not getting stomped as hard as you would if you were facing your Timberwolf-level opponents. And when you're in your Timbies, you should find you're able to turn out a better-than-average performance. For folks who play a variety of mechs, it's going to make leveling new mechs less frustrating, and more satisfying to take those fully upgraded mechs out to battle.

#53 Alwrathandabout42ninjas

    Member

  • PipPipPipPipPipPipPipPip
  • The Formidable
  • The Formidable
  • 1,098 posts

Posted 17 August 2015 - 03:47 PM

Its about time, now we dont have to deal with scrubs who think they are pro but are clearly not, if the system works properly.

#54 Mawai

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,495 posts

Posted 17 August 2015 - 03:48 PM

View Postbad arcade kitty, on 17 August 2015 - 02:28 PM, said:


well, who would doubt that pgi are [Redacted]

they introduce performance into the elo... i.e. you shouldn't care of the win, you should care of a high match score, it's better to do something which will more likely ensure a high match score for you than something which will ensure win but you don't get a good score

btw, it will hurt light pilots, especially with supportish builds


I think the point of the example was that you gain more by winning than maximizing your numbers over the team contribution to winning.

If their ranking system front loads assaults into tier 1 and lights into tier 5 because they get naturally higher and lower match scores on average then the entire system will be broken since the PSR won't be reflecting the player skill but the performance of the mech chosen.

I can just imagine a bunch of Tier 1 "bads" driving LRM assaults or meta Timberwolves who get over 1000 damage and over 100 match score. It will be hilarious to watch the tears on here if it works out that way :)

#55 Kira Onime

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Dragoon
  • The Dragoon
  • 2,486 posts
  • Twitch: Link
  • LocationMontréal, Québec.

Posted 17 August 2015 - 03:49 PM

So I'm assuming each tier will be matched with/against people of the same tier, 1 tier lower and possibly 1 tier higher than them......
I kinda hope it's not the same tier only.

#56 Deathlike

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 29,240 posts
  • Location#NOToTaterBalance #BadBalanceOverlordIsBad

Posted 17 August 2015 - 03:50 PM

View PostMawai, on 17 August 2015 - 03:48 PM, said:

I can just imagine a bunch of Tier 1 "bads" driving LRM assaults or meta Timberwolves who get over 1000 damage and over 100 match score. It will be hilarious to watch the tears on here if it works out that way :)


The tears would be glorious...

#57 KraftySOT

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • 3,617 posts

Posted 17 August 2015 - 03:52 PM

View PostGas Guzzler, on 17 August 2015 - 03:35 PM, said:

. Unless you are saying people will abuse it to artificially lower their tier, but I don't know who would want to do that,


Me.

No more aiming for legs. Now its coup de grace for minimum matchscore and maximum KDR.

And I dont care. It is an improvement, but its not a big improvement. Its not really a matchmaking 'system' persay so much as it is a new way to determine something that isnt really reflective of your actual skill.

Itll be better. But dont expect miracles.

PGI COULD just be like "Ok all systems are open to change, we've been wrong before, we're doing what we can here folks".

Which ill admit, has been the case for a while now. Its just annoying when two years ago, we said this, and got nothing but static in return.

#58 ScarecrowES

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 2,812 posts
  • LocationDefending the Cordon, Arc-Royal

Posted 17 August 2015 - 03:53 PM

View PostKira_Onime, on 17 August 2015 - 03:49 PM, said:

So I'm assuming each tier will be matched with/against people of the same tier, 1 tier lower and possibly 1 tier higher than them......
I kinda hope it's not the same tier only.


In all likelihood, there will have to be some minor relaxation of standards to accompany shorter matchmaking wait times. Possibly try to match you purely by tier first, then just by an acceptable skill spread, similar to how it works now.

As of now though, they say it will be by tiers. They also said 3-3-3-3 would be a thing. So grain of salt there.

#59 FupDup

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 26,888 posts
  • LocationThe Keeper of Memes

Posted 17 August 2015 - 03:54 PM

View PostKraftySOT, on 17 August 2015 - 03:52 PM, said:

...
No more aiming for legs. Now its coup de grace for minimum matchscore and maximum KDR.
...

Actually, legging people can often be one the fast/easiest ways to kill a target...it's just not as intuitive (or fun) as unloading everything you have into their center-mass. :P

Even assaults like the Stalker can be easier to kill by double-legging than trying to take out their juicy core, because torsos can be shielded (on most mechs) but legs can't.

#60 ScarecrowES

    Member

  • PipPipPipPipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 2,812 posts
  • LocationDefending the Cordon, Arc-Royal

Posted 17 August 2015 - 03:55 PM

View PostKraftySOT, on 17 August 2015 - 03:52 PM, said:


And I dont care. It is an improvement, but its not a big improvement. Its not really a matchmaking 'system' persay so much as it is a new way to determine something that isnt really reflective of your actual skill.



What is a better way of dertermining your actual skill than by recording your actual results and comparing that to others entirely objectively?





17 user(s) are reading this topic

0 members, 17 guests, 0 anonymous users