Jump to content

Kdr Has No Value, We Need Better


42 replies to this topic

#1 Dread Render

    Member

  • PipPipPipPipPipPipPip
  • Philanthropist
  • 847 posts
  • LocationSouth River NJ

Posted 31 December 2015 - 09:42 AM

We need a better way to match people than KDR
http://imagizer.imag.../695/QQpc8D.jpg

I have 12 assists, 2 Most damage done and a nice chunk of damage.
Did this improve my Kill Death Ratio? ...No.
Although this is an extreme example I would say Most of the time this holds true for me personally.
I almost always have several assists and a few Kill most damage.
so... it is a real bummer when I look at my KDR.

then I think... is this what the MM is using? I hope not.
If so we need something better.

#2 Mcgral18

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • CS 2019 Top 8 Qualifier
  • CS 2019 Top 8 Qualifier
  • 17,987 posts
  • LocationSnow

Posted 31 December 2015 - 09:44 AM

PSR uses Match Score (which is half your Damage and everything else has a point value.).

Edited by Mcgral18, 31 December 2015 - 10:01 AM.


#3 Champion of Khorne Lord of Blood

    Member

  • PipPipPipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 4,806 posts

Posted 31 December 2015 - 09:48 AM

One issue I find is that no stat really holds any value depending on the situation.

If you got high damage you could have been just spreading damage too much and hitting worthless components, if you got high kills they could have been all kill steals, if you got high assists you could have just sprayed a large laser over the enemy team one meter within max range. Solo kills could be awarded to you when you shot up the arms and did a little spread damage with some streaks or lrms but someone else just goes and headshots them with 6 CSPL.

I can't trust any stats anymore!

#4 TercieI

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 8,189 posts
  • LocationThe Far Country

Posted 31 December 2015 - 09:54 AM

Yeah, MM is based on Tier, which is based on match score (and volume, sadly). KDR has nothing direct to do with it.

Honestly, it might be better than what we have now.

#5 Captain Stiffy

    Member

  • PipPipPipPipPipPipPipPipPip
  • Liquid Metal
  • Liquid Metal
  • 2,234 posts

Posted 31 December 2015 - 10:02 AM

I want to see a system that penalizes player score for those who refuse to provide cover fire and just hang around waiting to die.

#6 Kubernetes

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Blazing
  • The Blazing
  • 2,369 posts

Posted 31 December 2015 - 10:10 AM

View PostDakota1000, on 31 December 2015 - 09:48 AM, said:

One issue I find is that no stat really holds any value depending on the situation.

Solo kills could be awarded to you when you shot up the arms and did a little spread damage with some streaks or lrms but someone else just goes and headshots them with 6 CSPL.

I can't trust any stats anymore!


Actually, solo kills require both KMDD and the killing blow. Solo kills and KMDD are pretty solid stats.



#7 Hexenhammer

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 3,729 posts
  • LocationKAETETôã

Posted 31 December 2015 - 10:21 AM

View PostCaptain Stiffy, on 31 December 2015 - 10:02 AM, said:

I want to see a system that penalizes player score for those who refuse to provide cover fire and just hang around waiting to die.



The question is how do you quantify that?

"Ding his score! He wasn't supporting me!"
"Dude. I'm a spider with MGs and medium lasers. They're shooting you with gauss at 800 meters."

Sorry. Way to subjective of a requirement to be met. Not even an AI would be able to do it.

Edited by Hexenhammer, 31 December 2015 - 10:22 AM.


#8 Champion of Khorne Lord of Blood

    Member

  • PipPipPipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 4,806 posts

Posted 31 December 2015 - 10:22 AM

View PostKubernetes, on 31 December 2015 - 10:10 AM, said:

Actually, solo kills require both KMDD and the killing blow. Solo kills and KMDD are pretty solid stats.


Oh, could have sworn I got them on matches were I had no kills, guess not then.

#9 Void Angel

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 7,782 posts
  • LocationParanoiaville

Posted 31 December 2015 - 10:25 AM

We do have a system designed to penalize people who don't provide covering fire and just wait around to die - it's called match score. Not really sure what Stiffy's point was...

#10 Deathlike

    Member

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

Posted 31 December 2015 - 10:26 AM

Winning still ultimately matters. The question is whether you were actually a factor in the wins.

Most of the time, excessive damage qualifies, but something like stopping the cap in Assault is enough (even if you die terribly) and goes totally unrewarded (you know, this Lostech concept).

Numbers don't always tell the entire story.

Edited by Deathlike, 31 December 2015 - 10:26 AM.


#11 Monkey Lover

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 7,918 posts
  • LocationWazan

Posted 31 December 2015 - 10:45 AM

No real players care about kdr.

#12 adamts01

    Member

  • PipPipPipPipPipPipPipPipPip
  • Death Star
  • 3,417 posts
  • LocationPhilippines

Posted 31 December 2015 - 11:02 AM

I love the emphasis on winning. No matter how your contribute (damage, killing, scouting, drop commanding, distracting, providing ECM, good looking paint), many of these not measurable, if you consistently win you should move up.

The three problems are:
  • It's too easy to move up
  • It's impossible to move down
  • Tiers are completely mixed up in matches anyway


#13 xTrident

    Member

  • PipPipPipPipPipPipPip
  • Bad Company
  • 655 posts
  • LocationWork or Home

Posted 31 December 2015 - 11:02 AM

View PostDeathlike, on 31 December 2015 - 10:26 AM, said:

Winning still ultimately matters. The question is whether you were actually a factor in the wins.

Most of the time, excessive damage qualifies, but something like stopping the cap in Assault is enough (even if you die terribly) and goes totally unrewarded (you know, this Lostech concept).

Numbers don't always tell the entire story.


Even if you weren't a factor in the victory (talking strictly PUGs) eventually that kind of play style would even out if the MM was based on winning and losing. Said PUG hangs back, doesn't do much in the way of support or damage but wins. Their player rating for the win goes up and said PUG goes against tougher opponents. Tougher opponents who are less forgiving and require a fully supported opposing team. Said PUG plays his normal way and winds up losing bringing their rating back down.

Of course this type of MM would be a constant balancing act with the player rating... But that's just exactly what a MM does. And as it stands my wins/losses ratio is nearly one to one right now. I think I might have between two or four more wins than losses. I'd like to chalk it up to the MM working properly but I have way to many ROFLMAO stomp losses for me the believe that.

I just wonder what people think would make the best "match maker"? What stats would be the best? I can tell everyone this from my experience - While the games are completely different there is something to be said for players KDR. On a FPS that shows other players KDR the ones that run 1.5 and well up from that you can absolutely tell they're better in game. I understand it's really hard to compare a FPS to MWO... But between damage done and KDR I bet we could get a pretty good indication at who's got skills in MWO.

#14 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 31 December 2015 - 12:36 PM

View PostRender, on 31 December 2015 - 09:42 AM, said:

We need a better way to match people than KDR


Actually, we need none. Randomize everything!

And yes, I am serious.

#15 Khobai

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 23,969 posts

Posted 31 December 2015 - 12:44 PM

Quote

I want to see a system that penalizes player score for those who refuse to provide cover fire and just hang around waiting to die.


There already is a penalty for that. Its called losing.

#16 5LeafClover

    Member

  • PipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 317 posts

Posted 31 December 2015 - 01:00 PM

If games lasted less than a minute and we racked up 100s per week, then just going off wins/losses would be perfect.

At MWOs game rate of 5ish per hour (and mixed modes), win/loss is not quite statistically reliable enough, unless you take a loong time window. In which case it isnt sensitive enough (a sudden improvement in skill wont be reflected in tier for a long time)

To overcome this, match score is used. Theres a good mix of good team behaviours that it captures, but it is still clumsy. IMO it's weighted far too heavily towards DMG (1pt for 2dmg, normally contributing to over half my match score). My progress in the tiers runs at different rates if I play heavies rather than lights. Levelling panthers and ravens has been killer for me lately as i just can't achieve the same damage.

Edited by 5LeafClover, 31 December 2015 - 01:05 PM.


#17 Hawk_eye

    Member

  • PipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 325 posts
  • LocationGermany

Posted 31 December 2015 - 01:26 PM

AFAIKT, PSR is a flawed system as in, eventually, everyone reaches tier-1.

Instead of moving up or down in your PSR depending on your match-score, use the match-score itself (yes, match-score is also kinda flawed, but it´s the best we have atm).

Add all scores and divide them by the number of matches played, which gives you your average match score.

Now just give each tier a number of slots (percentage of total player base or something)

So let´s say there are 5000 accounts (completely arbitrary number)

Tier-5 35% of player-base --> 1750 players with the lowest rating go there
Tier-4 30% of player-base --> 1500 player
Tier-3 20% of player-base --> 1000 player
Tier-2 10% of player-base --> 500 player
Tier-1 5% of player-base --> 250 player with the highest rating go here.

As a player improves his score enough, he replaces a higher tier player, who goes down one tier to take his place.
That way, tier-1 would truly represent the top 5% of the player-base and matching similar tiers vs. each other could actually work.

Just my 2 cent

#18 Metus regem

    Member

  • PipPipPipPipPipPipPipPipPipPipPip
  • The Sureshot
  • The Sureshot
  • 10,282 posts
  • LocationNAIS College of Military Science OCS courses

Posted 31 December 2015 - 02:38 PM

View PostMcgral18, on 31 December 2015 - 09:44 AM, said:

PSR uses Match Score (which is half your Damage and everything else has a point value.).


And since damage counts for so much, I get penalized for my PSR, as I try to keep my kills below 150 damage per kill... So even if I get 4 kills, I'm looking at less than 600 damage or 300 match score for that, usually closer to 260-ish match score from damage, with another 100 or so from doing the other things that should count more....

#19 Ex Atlas Overlord

    Member

  • PipPipPipPipPipPipPipPip
  • Survivor
  • 1,018 posts

Posted 31 December 2015 - 03:06 PM

View Postadamts01, on 31 December 2015 - 11:02 AM, said:

The three problems are:
  • It's too easy to move up
  • It's impossible to move down
  • Tiers are completely mixed up in matches anyway


1) If you're good.
2) No it's not.
3) Sort of.

Aka, you're full of it.

#20 VinJade

    Member

  • PipPipPipPipPipPipPipPip
  • 1,211 posts

Posted 31 December 2015 - 03:43 PM

I always felt bad about kill steals and always wished that I didn't get the 'kill' bonus which was clearly earned by someone else before my rain of hell fell upon the enemy mech destroying it and taking the win from whoever should have gotten that.

I will always feel that way and would be happy with just the asst kills.

but maybe that's just me.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users