Jump to content

Is Vs Clans, With Science! New Data - 17/07/14

Balance General

312 replies to this topic

#141 Ted Wayz

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 2,913 posts
  • LocationTea with Romano

Posted 16 July 2014 - 07:58 PM

A large contributor to the disparity in damage is the CERML would be my guess based on my own experience.

#142 Asmudius Heng

    Member

  • PipPipPipPipPipPipPipPipPip
  • Survivor
  • 2,429 posts
  • Twitter: Link
  • Twitch: Link
  • LocationSydney, Australia

Posted 16 July 2014 - 08:01 PM

View PostTezcatli, on 05 July 2014 - 08:13 PM, said:

I think people can agree the clans have a higher DPS. But the IS meta builds are dangerous due to their pin point damage. But unless you're using a hit and run build. You're kinda in for a hard time. I know my LRM Treb feels pretty weak because of how easily it gets shredded by Clan builds. And it can't compete with Clan LRM builds, even if they supposedly suffer more to AMS, since not surprisingly a lot of people play without AMS.


This is often not taken into account when people are talking about Clan Vs IS.

Everyone points to the high end meta of the IS to compare but never to any other mechs.

If the way to compete with clan mechs is to push the FLD PP meta as hard as you can then this speaks volumes to the power of clan mechs.

The thing is that Clans can do it as well with ERPPCs and Gauss and i see more and more of those on the battlefield as time goes by.

Clans are not outrageously OP but lets compare across a greater range of builds than focusing on the bleeding edge of competative meta.

#143 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 08:07 PM

View Postmirrimon, on 16 July 2014 - 07:57 PM, said:

One thing to consider is that removing (C) and 0 damage mechs by ignoring their end game stats still leaves their 'damage taken' and deaths added to the non (C) mechs. Further since Clan mechs can not be (C) mechs yet, this means their damage/kills are skewed higher. Still it is really nice to get some preliminary numbers. I wish PGI would release stats like this (or better yet the raw data).


I actually removed all of that - I added stats to the raw data, so I could exclude it in formula:

Posted Image

So any match that has even 1 (C) or 0 dmg mech has ALL it's data excluded from the appropriate counts:

Posted Image

This means that the only figures going into (C)/NOT(C) etc are whole matches that fit the criteria, so no figures with even one (C) mech will make it into a non-c count. If that makes sense. The numbers are kind of swimming in front of me atm, heh.

#144 Thunderfrog

    Member

  • PipPipPip
  • CS 2021 Participant
  • 53 posts

Posted 16 July 2014 - 08:43 PM

What about players who are piloting (c) mechs they own? I've bought a few on sale. PS: Sent about 10 screens. I've been capping more since you asked.

Edited by Thunderfrog, 16 July 2014 - 08:44 PM.


#145 Jman5

    Member

  • PipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 4,914 posts

Posted 16 July 2014 - 09:12 PM

My poor hunchback. One of the worst mechs in the game. :)

#146 mirrimon

    Member

  • Pip
  • Talon
  • Talon
  • 17 posts

Posted 16 July 2014 - 09:23 PM

I see. That's good thinking. Thanks!

Edited by mirrimon, 16 July 2014 - 09:24 PM.


#147 Jman5

    Member

  • PipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 4,914 posts

Posted 16 July 2014 - 09:26 PM

View Postvondano, on 01 July 2014 - 03:48 AM, said:

looking at those chart...

i see only 4 clan mechs on the top ten dmg/match...

and OMG only 3 clan mechs on the top 10 kills/match...

buff clan pls! IS OP!

5 in the first chart, 4 in the second chat.

Consider the fact that there are only 8 freaking clan mechs and half are already in the top 10 versus 29 IS mechs. Hypothetically speaking you made everything perfectly balanced, you would expect to have 8 IS mechs and only 2 Clan mechs in the top 10.

#148 Lefty Lucy

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 3,924 posts
  • LocationFree Tikonov Republic

Posted 16 July 2014 - 09:28 PM

View PostJman5, on 16 July 2014 - 09:26 PM, said:

5 in the first chart, 4 in the second chat.

Consider the fact that there are only 8 freaking clan mechs and half are already in the top 10 versus 29 IS mechs. Hypothetically speaking you made everything perfectly balanced, you would expect to have 8 IS mechs and only 2 Clan mechs in the top 10.


Yep, I see some data that points to the clans being overpowered compared to IS, even if not to a *huge* degree, and it seems like people are reading that it proves that they're equal in balance. Mind-blowing.

#149 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 09:47 PM

View PostJman5, on 16 July 2014 - 09:26 PM, said:

5 in the first chart, 4 in the second chat.

Consider the fact that there are only 8 freaking clan mechs and half are already in the top 10 versus 29 IS mechs. Hypothetically speaking you made everything perfectly balanced, you would expect to have 8 IS mechs and only 2 Clan mechs in the top 10.


This is true. The thing i'm really missing is group results. I'm a firm believer that the IS meta is on a reasonable footing with the Clan meta, but you just don't see that many MetaSlayers in Pug data. MetaCats, however... because there's an over-saturation of Clan Cats, i think it's skewing the results. Maybe the old-guard IS meta pilots are testing the Cat?

Also, the few clan mechs we have seem to have a -very- viable medium->top end, and not a whole lot of choice. I'm guessing when they add new mechs, there might be an under-performer or two that drags things back down.

The MadCat and StormCrow though, man, they really show that a lot of forethought was put into making them powerful.

#150 Praehotec8

    Member

  • PipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 851 posts

Posted 16 July 2014 - 09:52 PM

Where are your confidence intervals and P-values? Where is the statistical significance? Really just kidding, excellent work!

I was just discussing with my significant other how nerdy we are that we were discussing whether or not your data is powered enough to show statistical significance! =P

I think these are at least a foundation to demonstrate that the balance between the factions is actually not too bad. I certainly don't see the overpoweredness of any mech except maybe that crazy locust!!

#151 Lynx7725

    Member

  • PipPipPipPipPipPipPipPip
  • 1,710 posts

Posted 16 July 2014 - 09:58 PM

View PostKiiyor, on 16 July 2014 - 09:47 PM, said:

This is true. The thing i'm really missing is group results. I'm a firm believer that the IS meta is on a reasonable footing with the Clan meta, but you just don't see that many MetaSlayers in Pug data. MetaCats, however... because there's an over-saturation of Clan Cats, i think it's skewing the results. Maybe the old-guard IS meta pilots are testing the Cat?

Hell, I pumped nearly 90 Timber Wolf matches into the overall pool. That's what, 20% of the pool? Even removing the non-close games, that's gotta shift the balance a bit.

#152 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 10:01 PM

View PostPraehotec8, on 16 July 2014 - 09:52 PM, said:

Where are your confidence intervals and P-values? Where is the statistical significance? Really just kidding, excellent work!

I was just discussing with my significant other how nerdy we are that we were discussing whether or not your data is powered enough to show statistical significance! =P

I think these are at least a foundation to demonstrate that the balance between the factions is actually not too bad. I certainly don't see the overpoweredness of any mech except maybe that crazy locust!!


LOL!

Posted Image

Posted Image

I STARTED digging around in proper variance tests, but being inebriated when I started, i very quickly hit the point where I had NO IDEA WHAT THE HELL I WAS LOOKING AT. "It's a job for sober me", I thought.

In the harsh light of the morning sun, I viewed my work and realised that I had the same problem as so many great rockstars and poets; that the temporary genius I had when under the influence was just that... temporary.

#153 Too Much Love

    Member

  • PipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 787 posts

Posted 16 July 2014 - 10:04 PM

PGI should hire you. No, not PGI but UN or White House:) I'm really impressed. Good job!

#154 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 10:05 PM

View PostLynx7725, on 16 July 2014 - 09:58 PM, said:

Hell, I pumped nearly 90 Timber Wolf matches into the overall pool. That's what, 20% of the pool? Even removing the non-close games, that's gotta shift the balance a bit.


I agree. Hopefully, as data keeps coming in, things will stabilize. I've pretty much got this set up as a template now, so I just need to add to the raw data.

#155 Aresye

    Member

  • PipPipPipPipPipPipPipPipPip
  • Heavy Lifter
  • Heavy Lifter
  • 3,462 posts

Posted 16 July 2014 - 10:06 PM

View PostLefty Lucy, on 16 July 2014 - 09:28 PM, said:


Yep, I see some data that points to the clans being overpowered compared to IS, even if not to a *huge* degree, and it seems like people are reading that it proves that they're equal in balance. Mind-blowing.


The devs never intended for Clans to be fully balanced with IS. They did however, want Clans to require more effort to realize those advantages.

As it stands right now we can assume a good majority (not all) of people who bought Clan mechs are more experienced, as evidenced by the majority of competitive players already owning them. Once the Clan mechs are available for CBills and accessible by all we'll be able to see where the balance truly lies.

#156 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 16 July 2014 - 10:21 PM

View PostKiiyor, on 16 July 2014 - 09:47 PM, said:


This is true. The thing i'm really missing is group results. I'm a firm believer that the IS meta is on a reasonable footing with the Clan meta, but you just don't see that many MetaSlayers in Pug data. MetaCats, however... because there's an over-saturation of Clan Cats, i think it's skewing the results. Maybe the old-guard IS meta pilots are testing the Cat?

Also, the few clan mechs we have seem to have a -very- viable medium->top end, and not a whole lot of choice. I'm guessing when they add new mechs, there might be an under-performer or two that drags things back down.

The MadCat and StormCrow though, man, they really show that a lot of forethought was put into making them powerful.


You've got a big enough sample size, there's only a few thousand total MW:O players, if that. You're literally looking at data for almost a full 1% total population segment. Pew would laugh at someone who said they needed a sample size that big. 40 is technically enough to settle your Elo.

Which brings me to a factor in the subject that I just can't get around - Elo.

Fundamentally, Clan pilots will skew towards veterans. All Clan mechs currently are cash purchase. That's going to skew into experienced players. When I looked at just the top 3 scores/killers/assists/damage in each match you start to see an even steeper skew towards Clan mechs too, but is that the mechs or the players?

I've got 40 matches post 3x4 if you're still collecting data. Overall, amazing work. You mined a crap ton of stuff out of it, angles I hadn't even considered.

One thing I did consider though and you just might have the gumption to help manage it (I'm working 60-70 hour weeks for the next while and it's just not a viable concept) would be.....

Private matches with IS on one side and Clan on the other. No VOIP allowed, take volunteers on a forum thread - two groups. One with premium time folks willing to pugboss 12man teams and help facilitate the matches, one with volunteers willing to take their mechs.

Even better... people submitting to the pugbosses what their build is (even just a simple '325 XL, 2PPC, 1UAC5 CTF 3D') and end of round screenshots.

Telemetry on performance not just of chassis but builds and weapons.... Mmmm, delicious.

How about people submitting screenprints of their weapon stats pages? Not sure how useful that is though without the associated data.

You're a boss though man. Sexy work. Many due props.

#157 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 10:28 PM

View PostMischiefSC, on 16 July 2014 - 10:21 PM, said:


You've got a big enough sample size, there's only a few thousand total MW:O players, if that. You're literally looking at data for almost a full 1% total population segment. Pew would laugh at someone who said they needed a sample size that big. 40 is technically enough to settle your Elo.

Which brings me to a factor in the subject that I just can't get around - Elo.

Fundamentally, Clan pilots will skew towards veterans. All Clan mechs currently are cash purchase. That's going to skew into experienced players. When I looked at just the top 3 scores/killers/assists/damage in each match you start to see an even steeper skew towards Clan mechs too, but is that the mechs or the players?

I've got 40 matches post 3x4 if you're still collecting data. Overall, amazing work. You mined a crap ton of stuff out of it, angles I hadn't even considered.

One thing I did consider though and you just might have the gumption to help manage it (I'm working 60-70 hour weeks for the next while and it's just not a viable concept) would be.....

Private matches with IS on one side and Clan on the other. No VOIP allowed, take volunteers on a forum thread - two groups. One with premium time folks willing to pugboss 12man teams and help facilitate the matches, one with volunteers willing to take their mechs.

Even better... people submitting to the pugbosses what their build is (even just a simple '325 XL, 2PPC, 1UAC5 CTF 3D') and end of round screenshots.

Telemetry on performance not just of chassis but builds and weapons.... Mmmm, delicious.

How about people submitting screenprints of their weapon stats pages? Not sure how useful that is though without the associated data.

You're a boss though man. Sexy work. Many due props.


I'd love to be able to differentiate ELO somehow, but i'm assuming the MM will do that for me.

Also, as far as the 1v1's go, I've thought about propositioning Sandpit for this, I think he has something awesome brewing:

http://mwomercs.com/...n-with-sandpit/

And hell yes i'd like your data - i'm convinced that I need to remove outliers to get proper results, and i'd like to eventually have a sample that didn't have a 4 match locust appearing as one of the most deadly mechs in the universe.

My gut feeling now is that things will (hopefully) become more interesting with enough 3/3/3/3 matches. The IS seem to have an edge with Lights and Assaults, the Clans are dominating with Mediums and Heavies.

And builds? That would be glorious. I could get averages of LRM damage from LRM boats if there's no AMS, and then compare those damage figures to matches with lots of AMS. Hmmm.

#158 Lynx7725

    Member

  • PipPipPipPipPipPipPipPip
  • 1,710 posts

Posted 16 July 2014 - 10:34 PM

View PostKiiyor, on 16 July 2014 - 10:28 PM, said:

And hell yes i'd like your data - i'm convinced that I need to remove outliers to get proper results, and i'd like to eventually have a sample that didn't have a 4 match locust appearing as one of the most deadly mechs in the universe.

Hey. No dissing of the Dreaded Locust.


:)

#159 Aim64C

    Member

  • PipPipPipPipPipPipPip
  • 967 posts

Posted 16 July 2014 - 10:39 PM

As I've been playing over the past week or so, I've noticed a considerable shift in how clan mechs play. While the meta should be expected to shift, to some degree, I am wondering if it is going to stagnate on PPCs and LRM/SRM builds.

Part of the 'problem' with clans is that many of them have a 'free catapult' attached to them which disproportionately skews their ability to contribute to total indirect fire.

I've come into the discussion a bit late - but it might be more useful to study the blowout matches in terms of discovering imbalancing factors.

I've seen more 'blowouts' the past couple nights than I have in the previous week and a half (or two - haven't been tracking that closely). Pretty much all of the games I played were a near shut-out victory of <=3 losses for the victor.

Which is a distinct shift from my previous experiences (which were ending on a much more even keel and had far more engaging battles).

Of course, when the winning team has 3 direwolves - it's a nail-biter in the rare event you do win - and you usually only get there with a good scout paired with a competent pilot behind a hefty set of LRM racks.... and your team is good at not dying while the enemy is not confident enough to rush.

Sometimes a balance of math does not indicate a balance of gaming experience. As others have noted - my IS builds are largely some variety of min-maxed build taken to ridiculous and/or compromising extremes. As compared to the Clan mechs that get to play with a variety of toys (and it's only a matter of time before they start concocting frankenbuilds).

Now - this wouldn't bother me if we had an actual battlefield with battlefield objectives and what-not/where-for (which would also imply some aspect of community warfare aside from "choose your banner.") If we had an actual clan invasion - it would truly set the atmosphere to have your "creme of the crop" custom ride with the latest and greatest out of the Helm Memory Core revelations... and all we can do is -barely- compete with the Clans in narrow regions of specialization.... that's neat if we're dropping with imbalanced teams and differing objectives based around some kind of immersion into the lore.

But, back to reality - I feel that where the clans will move is that they will be the more 'fun' way to play the game, spare for a few eccentric individuals, because they can more universally react/dominate the battlefield than IS mechs can. There are fewer times where you get caught "helpless" to the degree that you can't even spit in the face of the enemy.

But I digress.

#160 Kiiyor

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 5,565 posts
  • LocationSCIENCE.

Posted 16 July 2014 - 10:40 PM

View PostLynx7725, on 16 July 2014 - 10:34 PM, said:

Hey. No dissing of the Dreaded Locust.


:)


Posted Image





17 user(s) are reading this topic

0 members, 17 guests, 0 anonymous users