259 replies to this topic

[Pope RW]

Nothing like patting yourself on the back with a limited data set, telling us how great a job you did.

Time for the nerf hammer again for the clans.

Edited by Pope RW, 29 May 2017 - 02:49 PM.

[SlippnGriff]

>underperforming night gyrs
what

KGC tho needs a good bit of work, even before the patch it was slightly lacking

Also check in on the Dire wolfs, they are hurting bad and have been for a while!

[CK16]

Sight, things I have learned about light pilots from their QQing...

They want their lights as tanky as a heavy, as much fire power as a heavy, but still want to maintain their 130+kph and turn on a dime ability...sounds fair to me...

Edited by CK16, 29 May 2017 - 02:55 PM.

[Click]

Chris Lowrey, on 29 May 2017 - 10:01 AM, said:

*snip*

So just to be clear, you consider the NTG overperforming only now or even more than it already was?

Either way if it is overperforming I figure its because of the skill tree or more specifically because of the firepower/defense trees and if I'm correct you're mostly concerned about the dakka build.

So what's the plan, decrease skill tree node values further? I mean the UAC and the gauss specific nodes are crap already and I don't see you touching the defense tree when you're so dependant of it to increase TTK all around. The only thing left to cut is in the mech itself aka putting negaquirks on omnipods or cutting down baseline armor and agility. So you put negaquirks on the most popular pods or cut down on the armor and all NTG owners will rage at you with the force of a thousand suns..or you cut the agility and hope that in due time it'll impact whatever metrics you pull. Which I don't think it will unless you give it assault-like agility values.

Either way I don't think it's enough. You really should buff other mechs so they can keep up instead. Leave the mech as is and improve alternatives through those energy/general weapons rebalance you have coming up, or if you really insist on using the nerf bat use it on overall dakka builds through there. Otherwise it'll be just another sh*tshow.

Edited by Click, 29 May 2017 - 03:03 PM.

[Burke IV]

I have just spent about an hour messing about with the skill tree planner when i should have been playing. As i look at my little collection of hexagons i have no idea if i have put them in the right places or not and wtf im not playing trial and error with it. Honestly im feeling a little annoyed, this has spoiled my fun and in the end i know im not gonna play. I wonder how many others have filled the tree in with no idea where stuff should be, or even just slightly sub optimal. Dont all those people make the data invalid?

Edited by Burke IV, 29 May 2017 - 03:04 PM.

[FupDup]

Chris Lowrey, on 29 May 2017 - 12:56 PM, said:

This is the queue for people attempting to drop into a match, not the queue for what 'Mech's are being taken into the matches. All this image shows is that if you are entering the Match Making Queue, you will drop into matches faster if you take a Light 'Mech. It is not an accurate depiction of the tonnage distribution for the matches themselves.

Nor am I saying that I'm satisfied with the current state. Which is why we are sharing our current findings with everyone to show the current trajectory from the past few weeks and planning further actions through upcoming patches.

The queue breakdown is still useful for telling what type of mechs that people want to play the most, because a small queue with fast search time means that not many people want to play those mechs while a large queue with a long wait time means that a lot of people want to play them.

The actual in-game tonnage breakdown isn't as meaningful of a metric because the matchmaker tries to both limit the numbers of each class as well as try to match classes on both teams. If the matchmaker was "anything goes" like it was years ago, I suspect that we'd see a much larger in-game weight class distribution imbalance similar to the queue size imbalance.

The MM trying to match classes means that it's a lot harder to determine if a certain class is too weak or too strong since you'll probably see roughly equal numbers of that class on both teams, thereby cancelling out the positive or negative effects of having that class on your team. If we could have something like a team of 7 heavies and 5 assaults against a perfect 3/3/3/3 team, it's fairly easy to guess who would win that match.

[Chris Lowrey]

Tsar Bomba, on 29 May 2017 - 02:40 PM, said:

So you're nerfing the high end and boosting the low end so they all perform similarly?
...interesting.
So when are you going to increase the price of the cheap ones and lower the price of the expensive ones?
...yea, thought so.

You seem to be operating under the axiom that there has to be some sort of balance between all mechs, and yet we still pay premium for higher tonnage mechs and by your own design work detailed here they aren't meant to be any better than the lower tonnage. How do you justify that? I'd really like to hear some thoughts on that from PGI in this new era of 'openness' you pretend you got going right now.

I have nothing to do with anything that relates to the game play economy or 'Mech sales. My primary domain is in-match game and 'Mech balance. Costs and price points has zero part of any game-play balance decision that I or anyone else who works on game play balance makes. Our sole focus is on ensuring that balance continues in a trajectory that sees better balance between all available 'Mechs within the game.

You are free to form your own opinion of things as you see fit, but any lack of comment or response to anything regarding 'Mech cost or price points by myself or anyone else who works on balance has nothing to do with us being deaf to or ignoring any kind of criticism so much that we have zero say in that aspect of MWO's development.

[Tarogato]

I hope Chris isn't being too horribly mislead by the effect of double coolshot 27's on the game. Every single mech he listed that improved with the exception of the Commando (but maybe not necessarily, because Death's Knell) is a laservomit mech. Even the Dragon (which is the DRG-1C, while surely the staple DRG-1N is performing worse than always now).

Remove double coolshot 27 from the game and we'll achieve better parity.

[Mcgral18]

I am wary about certain metrics being used...metrics which found the SadCat to be overperforming



Though, Gyr and M2C certainly do qualify for that role, Spheroid Lights performing may be...questionable.
They rarely Carry matches, they get carried. Hence the Win/Loss being a...less than desirable quality when referring to their improvement. Jenner F and Wolfie did improve, at least.

[Pope RW]

Why all these stats based on win/loss ratio of individual mechs?

A mech is 1/12th of a team, how a single mech performs win/loss wise has less to do with that mech and more to do with the team it was on.

[Bud Crue]

Chris Lowrey, on 29 May 2017 - 10:01 AM, said:

Hello MechWarriors,


As with every patch, we closely observe both the feedback and match metrics. This assists us in observing the overall state of the game. Which allows me, as one of the designers who works on game balance, to both determine the overall state of balance, and target changes to further improve it. While we do not often share the results of our findings publicly, given the scope of the changes introduced with the latest patch, and the many discussions that have cropped up from those changes, I wanted to take the time to share with you guys some of the takeaways we have had through observing the current metrics since the release of the Skill Tree.

• Inner Sphere Lights are currently seeing the biggest overall improvement from the Skill Tree system over the previous system. Their overall Win / Loss ratio jumped up 22% over pre-skill tree Win / Loss numbers.
  
• The introduction of the Defensive Tree has resulted in an overall boost to resilience. This has resulted in an increase to overall damage output scores across the board. Globally averaging at a +30 damage increase over the previous system.
  
• Inner Sphere to Clan balance in Win / Loss ratios between 'Mechs is the closest it has ever been within the game. With a global average performance difference of 6% between the factions. With the widest individual performance difference between 'Mech chassis' being 8%. While these numbers are the best that I have observed while part of the team, we are still of the opinion that these can be improved further. More on that below.
  
• Low performing 'Mechs got a huge bump with the Skill Tree / Engine Desync. The separation between top performing 'Mechs and lowest performing 'Mechs has greatly diminished over previous patches.
  
• Additionally, a much wider variety of variant load-outs are performing well under the current system. We are seeing a much higher variety of different load-outs not only deployed, but preforming well within their matches.

Here are the most improved 'Mech chassis since the introduction of the skill tree based on data since the May 16th patch:

• Dragon
  
  • 
    
    
    4th Greatest improvement in Win / Loss ratio. (30% increase)
    
  • 
    
    
    3rd Highest Avg. Damage Improvement over all 'Mechs. (+69.35 average damage)
    
  • 
    
    
    2nd Highest improvement in Kill / Death Ratio. (38% increase)
• 
  
  
  Commando
  
  • 
    
    
    3rd Greatest improvement in Win / Loss ratio. (33% Increase)
    
  • 
    
    
    Highest improvement in Avg. Damage for Light 'Mechs. (+56.95 average damage)
• 
  
  
  Panther
  
  • 
    
    
    Greatest improvement in Win / Loss ratio. (63% increase)
• 
  
  
  Gargoyle
  
  • 
    
    
    Highest improvement in Kill / Death Ratio (42% increase)
• 
  
  
  Warhawk
  
  • 
    
    
    Highest Avg. Damage Improvement over all 'Mechs. (+70.34 average damage)
• 
  
  
  Raven
  
  • 
    
    
    2nd Greatest improvement in Win / Loss ratio. (58% increase)
• 
  
  
  Nova
  
  • 
    
    
    2nd Highest Avg. Damage Improvement over all 'Mechs (+69.58 average damage)
• 
  
  
  Grasshopper
  
  • 
    
    
    3rd Highest improvement in Kill /Death Ratio ( 20% increase)
• 
  
  
  Wolverine
  
  • 
    
    
    4th Highest Avg. Damage Improvement over every 'Mech. (+65.79 average damage)

These numbers are from a fairly limited data pool given the limited amount of time that the Skill Tree has been released. So we are currently not viewing this data as "absolute" at this point in time. With time, things will settle as we get more data and these numbers may change. But we wanted to share with everyone the latest look at the current state of the game under the skill tree system based off the previous few weeks. With this said though, I want to make crystal clear that this does not mean we feel that balance is as good as it could be. This data has also highlighted a number of rough edges that we are fully aware about and we want to assure everyone reading that we will be addressing these points in future patches as we identify them. Among these points are a number of 'Mechs that have swung wide into both over performing, and under performing territory. The most extreme of them will be addressed in the June patch. With more to be addressed in future patches past June.



Among these 'Mechs being targeted for June adjustments include:

• Night Gyr
  
• Marauder IIC
  
• King Crab
  
• Firestarter
  
• Black Knight

Beyond this, baseline Inner Sphere / Clan balance remains a high priority as we move forward. June will see a pass at re-balancing the Energy weapons, and we will continue to refine the balance between Inner Sphere and Clan weapons and equipment as we move forward into future patches.



We will continue to observe and make adjustments based on our findings to ensure that we continue to refine balance within the Skill Tree system.

Chris,

Does your statistical analysis take into account tier and time?
What I mean is that several competitive players and units studied the skills tree and given their propensities to "play to the quirks" or to play what mathematically appears to be superior...then that is what they will play. For example the Panther and Dragon results: There has been several posts, both here and elsewhere from competitive players noting the superiority of these mechs given their remaining quirks and how to min max these mechs using the skills tree. It should thus be no surprise that when some of the best players in the game report that a mech and build is superior that more folks start playing that mech more than they otherwise would and thus get superior results given the benefits that those mechs now provide.

Does your analysis consider the past meta? Example your Grasshopper notation. This was one of the top IS heavies for countering clan PPFLD. It should be no surprise that meta players are bringing this mech out to see how it performs. It was a superior mech for a reason (the 5P and 5H...apropos does your data suggest that the N and J are that much better? Or just the 5P or just the H or what?) and it probably is not a surprise that it is being disproportionately represented in play by superior players trying to make sure that the mech is still superior as well, no?

I'd be interested in seeing data for mechs that had no quirks before, and no appreciable change va the Skills Tree. Mechs like the Ebon Jag for example. Did the skills tree make its stats improve or are they relatively stable?

[Wyattorc]

It would be interesting to see all the stats for the top 5 and bottom 5 mechs for each weight class.

Thanks for sharing the stats.

[Tarogato]

Chris Lowrey, on 29 May 2017 - 12:56 PM, said:

This is the queue for people attempting to drop into a match, not the queue for what 'Mech's are being taken into the matches. All this image shows is that if you are entering the Match Making Queue, you will drop into matches faster if you take a Light 'Mech. It is not an accurate depiction of the tonnage distribution for the matches themselves.

Nor am I saying that I'm satisfied with the current state. Which is why we are sharing our current findings with everyone to show the current trajectory from the past few weeks and planning further actions through upcoming patches.

Hrmmm, in my experience, the queue distribution shown here correlates exactly with what I see in my solo queue matches. Heavy queue at 50%? Expect to see 4-5 heavies per team in each match. Lights at 10% or less? Except to see 1-2 lights per side, instead of the ideal 3. New mech released and everybody is playing it? The waiting times for that weight class goes up.

Isn't the queue distribution showing overflow? So if everybody (50%+) is playing heavy mechs, the matchmaker is having to try to cram them all into matches, and making them wait in queue before triggering release values and allowing more then 3 heavies per team.

[Carl Vickers]

I hope their stats separate out the massive amount of damage being done by air strikes from mech damage as well.

[Half Ear]

Do not scare Chris away!!!

Win/Lose leans mostly towards TEAM play, communication, AGGRESSION and positioning. Any thing else is a wipe on the deurerre. This is not based people running through the same course, aka benchmarks of when systems are run using similar programs.....

But, thank you for the feedback!!!!

Edited by Half Ear, 29 May 2017 - 03:54 PM.

[Ultimax]

I have serious doubts about how you are interpreting your data, especially as I really doubt you've taken things like double coolshots and double arty into effect.

You just made a bunch of sweeping changes, some of which made a few mechs upper tier/meta - and some which took powerful mechs like the KDK-3 and decimated the amount of them being fielded.

Let your skill tree shake out for a few months, let the new tech enter the game - and don't touch anything yet.


The only thing you should be looking to tweak is the absolutely atrocious agility nerfs you doled out to 100Ts - they are awful to pilot now. (and removing negative quirks from TBR)

I think you tried to create a different agility bracket for each 5 ton mech increment - but there is not enough room in the spectrum.

Instead of different profiles at
1) 20
2) 25
3) 30
4) 35
5) 40
6) 45
7) 50
8) 55
9) 60
10) 65
11) 70
12) 75

etc.,

It would be more manageable like this:

1) 20/25
2) 30/35
3) 40/45
4) 50/55
5) 60/65
6) 70/75

and so on.

Edited by Ultimax, 29 May 2017 - 04:40 PM.

[KekistanWillRiseAgain]

Pope RW, on 29 May 2017 - 02:47 PM, said:

Nothing like patting yourself on the back with a limited data set, telling us how great a job you did.

Time for the nerf hammer again for the clans.

I wish they would just Buff IS and not Nerf Clan... but since they refuse with Clan Crocodile Tears fervent support to address XL<cXL issue, then these are the band-aids that have to be done. Help us fight Clan Crocodile Tears so that Engine Parity can be addressed and most of everything will fall away

[Tarogato]

Ultimax, on 29 May 2017 - 04:39 PM, said:

I have serious doubts about how you are interpreting your data, especially as I really doubt you've taken things like double coolshots and double arty into effect.

On the one hand, you can't ignore the effects of double coolshots and arty. They are a part of the game, and they are a part of the balance as they strongly affect which mechs are viable.


On the other hand, you also can't ignore that double coolshots and arty is just stupid. Absolutely ridiculous, and has no business being in this game. If I could name two things that swinged balance the most this patch, it would be the coolshots and arty, in order. No question, no doubt.

[Carl Vickers]

Tarogato, on 29 May 2017 - 04:44 PM, said:

On the one hand, you can't ignore the effects of double coolshots and arty. They are a part of the game, and they are a part of the balance as they strongly affect which mechs are viable.


On the other hand, you also can't ignore that double coolshots and arty is just stupid. Absolutely ridiculous, and has no business being in this game. If I could name two things that swinged balance the most this patch, it would be the coolshots and arty, in order. No question, no doubt.

It is my suspicion that it is very hard to afford said double cool shots and arties/air strikes without premium time. Most players will be losing cash per match very quickly so PGI made them more expensive to make premium time more valuable.

Just to add, removing the premium time from mech packs helped with this too.

Edited by Carl Vickers, 29 May 2017 - 04:52 PM.

[Liveish]

Hello Chris,

Thanks for the info, are you able to break it up between solo and group play.

The reason I ask, in solo i can bring bad IS Lights mechs and get away with murder, in group not so much.