Jump to content

K/d Ratio Not A Good Stat?


73 replies to this topic

#21 Khobai

    Member

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

Posted 19 March 2013 - 01:03 AM

Quote

There is no "kill stealing" in a team game. Otherwise I agree with the first and last sentence.



You are 100% wrong. There is kill stealing in a team game as long as KDR is a stat. Because only 1 person on a team gets credit for the kill. The whole team doesnt.

If youre going to have KDR it needs to reward all players for contributing towards a kill... like if player A does 95% of the damage to kill a mech and player B does 5% of the damage to kill a mech then player A should get .95 kills and player B should get .05 kills. Player B should not get 1.0 kills for only doing 5% of the damage while Player A gets nothing despite doing 95% of the work to get the kill.

Edited by Khobai, 19 March 2013 - 01:07 AM.


#22 SchwarzerPeter

    Member

  • PipPipPipPipPipPip
  • FP Veteran - Beta 1
  • 202 posts

Posted 19 March 2013 - 01:06 AM

Even in Counterstrike the K/D ratio pretty means nothing. You need to carry your team rather than doing for kills (although these can be the same :unsure: ).

Generally you have a roll to fulfil. If i'm piloting my Jenner and pop behind 2 LRM boats i usually wont get the kills, I just annoy them so that they wont shoot at my team. They get overrun by my Team later anyway.

And i would not go for kills, unless im sure that other team is not waiting behind the corner. The most useful pilot is the one, who stays alive.

View PostKhobai, on 19 March 2013 - 01:03 AM, said:



You are 100% wrong. There is kill stealing in a team game as long as KDR is a stat. Because only 1 person on a team gets credit for the kill. The whole team doesnt.


Of course you get credit for the assist. You even get more CBills for an assist than for an kill. How is this no credit? Sure it wont pop up in your K/D but wanye?

Edited by SchwarzerPeter, 19 March 2013 - 01:08 AM.


#23 Khobai

    Member

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

Posted 19 March 2013 - 01:11 AM

Quote

Of course you get credit for the assist. You even get more CBills for an assist than for an kill. How is this no credit? Sure it wont pop up in your K/D but wanye?


Assists arnt a stat and are in no way factored into KDR, so no you dont get credit.

Quote

Even in Counterstrike the K/D ratio pretty means nothing.


If it means nothing, then why do we need it? Please dont bring fail stats from other games into this game. GIve us stats that actually mean something instead.

If KDR remains a stat, KDR should award fractional kills based on percentile damage you did to the mech that was killed. Otherwise just replace it with a stat like damage done divided by damage received. Either way, kill stealing should not be encouraged or rewarded.

Edited by Khobai, 19 March 2013 - 01:15 AM.


#24 Arete

    Member

  • PipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 390 posts

Posted 19 March 2013 - 01:18 AM

As usual, stats are not equal to skill. Stats can indicate skill, but is no real proof.

Think of your stats as a way of seeing how you perceive you're doing in a mech compared to what the stats say.

Drawing from my own experiences, I've always felt that I've done better in the HBK-4G (AC20) as compared to the HBK-4P (mlas boat). With stats, I can see it's the other way around, which was surprising but interesting (higher win rate and higher KDR on the 4P). When I thought about why, I came to the conclusion that the 4G doesn't have the flexibility of the 4P. On the 4P I've put an LL in the left arm, so when the hunch goes I still got some firepower. On the 4G, this is hard to get without sacrificing speed. Also, the playstyle of the 4G simply doesn't suit me as well as the 4P.

The most interesting stat for now is win ratio, especially per mech. If you have a low win ratio and a high KDR, chances are you're not playing that mech in a way that's good for your team. You might be the one that comes in undamaged at the end of a match and can pick off 2-3 badly damaged opponents, but your team still lost because you were not participating in the fight where they needed you.

If you got low win rate and low KDR (often coupled with low damage), this usually indicates that you're getting picked off way too soon or has a build that's not working out.

High win rate and low KDR (but relatively high damage) is generally good, this means you're helping your team win.

Of course, high win rate, high damage and high KDR often indicates that this mech works well for you.

Edited by Arete, 19 March 2013 - 01:18 AM.


#25 Khobai

    Member

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

Posted 19 March 2013 - 01:20 AM

The point is KDR can be done better, in a way that actually conveys meaningful information.

#26 Naja

    Member

  • PipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 187 posts
  • LocationAustralia

Posted 19 March 2013 - 01:24 AM

I'd like to see (Damage done / No. Mechs damaged) or in layman's average damage per mech. Only taking to account enemy mech damage of course. Still not a super useful stat, but I'd wager it to be more useful than K/DR

edit: possibly even damage done as a percentage of mechs total hp, thus not penalizing hitting lower weight classes

Edited by Naja, 19 March 2013 - 01:25 AM.


#27 Hungus

    Member

  • PipPip
  • 43 posts

Posted 19 March 2013 - 01:36 AM

Damage per match is what should matter the most.

I also think assists should be rewarded more based on contribution. I remember in Battlefield 3 (crap game, I know) it was actually possible to get more points for a good assist then a plain kill.

Actual contribution needs to be credited more than your ability to last hit. Might even stop me over heating on an alpha strike to killsteal just so I can grind faster.

#28 xhrit

    Member

  • PipPipPipPipPipPipPip
  • Mercenary
  • Mercenary
  • 976 posts
  • LocationClan Occupation Zone

Posted 19 March 2013 - 01:49 AM

View PostRocdocta, on 18 March 2013 - 10:47 PM, said:

I hear people saying that if your K/D ratio is not positive then you are a bad pilot. But then as I am piloting jenners soley in PUGs i just cant do enough damage to rack up the kills. Even in an assault mech i dont get many kills. i find i do alot of damage but then die horribly after being focused upon. What would be a better stat to guage effectiveness? it doesnt bother me as its a game that is great fun. Just wondering how seriously other people take their K/D ratio.


K/D ratio is the best measure of personal effectiveness. I am a dedicated raven 4X pilot, who pilots almost exclusively in pugs, and I score highest in both kdr and damage per match in most of the matches I play in.



View PostHungus, on 19 March 2013 - 01:36 AM, said:

Damage per match is what should matter the most.


Inflicting massive damage is not the goal, destroying mechs is the goal. If you have a high damage score but low kill score then you need to work on two things.

First improve your aim. Where you land your shots is more important then actually landing shots. Imagine you have the biggest gun in the game, and you shoot a mech 10 times, in 10 different places. No mech ever made, not even stock trials, will die from that. Now imagine that same gun, except this time every shot hits in the same place. Even the biggest most upgraded custom mech in the game will be dead before you fire 10 shots.

Second, improve your knowledge of mechs. You should know exactly how much armor every mech in the game has, where the hardpoints are, and exactly how much damage your weapons do. Then you can calculate on the fly how many attacks you must inflict on each opponent to disable them, and exactly where to hit the mech to take it out the quickest. Math is OP.

Edited by xhrit, 19 March 2013 - 02:14 AM.


#29 SchwarzerPeter

    Member

  • PipPipPipPipPipPip
  • FP Veteran - Beta 1
  • 202 posts

Posted 19 March 2013 - 02:22 AM

The K/D ratio is mainly for our own ego. It is there because every other game has it and its one way to indicate your performance.

I don't see the problem if someone is complaining about kill "stealing". He was probably just saving you from an desperate alpha. And even if the other mech was shut down, you got some extra seconds to focus on the next enemy.

Don't be so foolish to just look on your K/D ratio, help you team. You can play a CoD Compain if you want a sky-high K/D ratio. If its that what makes you feel better.

#30 Sifright

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,218 posts
  • LocationUnited Kingdom, High Wycombe

Posted 19 March 2013 - 02:45 AM

View PostRocdocta, on 18 March 2013 - 10:47 PM, said:

I hear people saying that if your K/D ratio is not positive then you are a bad pilot. But then as I am piloting jenners soley in PUGs i just cant do enough damage to rack up the kills. Even in an assault mech i dont get many kills. i find i do alot of damage but then die horribly after being focused upon. What would be a better stat to guage effectiveness? it doesnt bother me as its a game that is great fun. Just wondering how seriously other people take their K/D ratio.


I pilot jenners whilst solo pugging and have 2.86 KDR with the jenner.

KDR might not be the only or even best way to judge skill but it certainly is a factor.

Edit:

I average 450 damage a match in my Jenner and 1.8 kills a game. Usually I go for large assaults and try and smash their rears, and bug out if i think a 3L is coming any where near me because a 3L will murder me dead with out much effort on the 3l pilots behalf.

Edited by Sifright, 19 March 2013 - 02:48 AM.


#31 KinLuu

    Member

  • PipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 1,917 posts

Posted 19 March 2013 - 02:49 AM

K/D
W/L
XP/Game

These three stats show you how you are doing, and if you are getting better or worse.
You can not look at one stat alone.

K/D is important, because it tells you how often you die, as the most efficient way to up your K/D is not dying.
W/L is important, because it tells you how often you carry your team to victory. (Or if you are a burden to your team, if it is below 1)
XP/Game is important, because it is a general measure of how well you are doing.

#32 The Amazing Atomic Spaniel

    Member

  • PipPipPipPipPipPipPip
  • 932 posts
  • LocationBath, UK

Posted 19 March 2013 - 03:00 AM

None of the stats actually mean very much when you start to think about them.

In conquest, a light mech can deliver a win for the team by being smart about capping without ever firing its weapons - but will get much less in the way of a match score than everyone else on the team.

If you pug solo, then pre-Elo it was almost impossible to get a W/L better than 1.0 because of those games where randoms faced teams and got steam-rollered.

If you have a high W/L, then Elo should now be slowly dragging you down towards 1.0.

Damage done often just reflects the size of mech you drive or how many LRMs you fire.

A high K/D can be gained by being either very good at the game, or by staying out of the fight and then moving in once the rest of the team has almost finished beating the opponents down.

If you drive the mechs that have some balance and model issues (cough *Raven* cough) then you will get a much better K/D than if you drive something less gifted.

... and so on and so on.

#33 DaisuSaikoro Nagasawa

    Member

  • PipPipPipPipPipPipPip
  • Mercenary
  • Mercenary
  • 973 posts
  • Google+: Link
  • LocationTaipei, Taiwan

Posted 19 March 2013 - 03:12 AM

View PostKhobai, on 19 March 2013 - 12:46 AM, said:


Uh for example if I do 164 damage to an Atlas, but someone runs in front of me and does 1 damage to the Atlas and kills him, thats a kill steal.

You have idiots out there that value KDR so much theyll run in front of you and block your shots just to get the kill. KDR rewards that kind of idiocy. Competing with teammates to get the killshot should not be rewarded.



A better anology would be michael jordan scoring 50 baskets in one game, then scottie pippen stealing the ball from michael jordan, and scoring 1 basket, which happens to be the winning basket in the same game, and then scottie pippen winning MVP even though he only scored 1 basket.

The player who does all the work is not the player being rewarded. That is how ******** KDR is. KDR needs to just be removed as a stat because not only is it meaningless but it also encourages player behavior that is entirely contrary to teamplay.



Hmm, I see where you are coming from. I agree that 1. People jumping in front of others to get kills is the worst and probably one of the worst aspects of focusing on KDR.

I appreciate your analogy and it's apt, I can see flaws in it, but your point is taken (no pun intended). Mind you I wouldn't take KDR away. What I wish were that PGI made it so that team play was emphasized (which I don't think that it is in MWO current iteration). Why aren't assists tracked?

Edited by Daisu Saikoro, 19 March 2013 - 03:15 AM.


#34 Moromillas

    Member

  • PipPipPipPipPipPipPip
  • 943 posts
  • LocationSecret **** moon base

Posted 19 March 2013 - 03:27 AM

It a metric of how many times you got the final shot in, and how many times you ended up being one of the last players the other team shot at.

It's entirely useless.

#35 DaisuSaikoro Nagasawa

    Member

  • PipPipPipPipPipPipPip
  • Mercenary
  • Mercenary
  • 973 posts
  • Google+: Link
  • LocationTaipei, Taiwan

Posted 19 March 2013 - 03:27 AM

There are some really good points made in this thread and its made me change my stance a bit.

KDR matters, but not alone. KDR is a component of how a warrior operates, but WLR is good to take into consideration when thinking of KDR. Accuracy also matters as well as AVG XP per match, and Cbills per match. I mean it all relates as a picture. I would hope that the warrior that has a low KDR but a quality WLR is appreciated more than the other way around, but until we get away from the individual drive that the game seems to be operated on (one of the areas I do wish PGI would improve).

The worst behaviors displayed were during the weekend when individual kills mattered (the tournament). It was painful to play during those times. And why aren't assists counted? Was that an oversight? I fully thought they would be included with the stats upgrade :)

#36 Commander Kobold

    Member

  • PipPipPipPipPipPipPipPip
  • The Territorial
  • 1,428 posts
  • LocationCanada

Posted 19 March 2013 - 03:39 AM

stats don't count for assists,K/D isn't that relevant to how usefull you may or may not be to your team.

#37 LordDante

    Member

  • PipPipPipPipPipPipPip
  • IS Exemplar
  • IS Exemplar
  • 782 posts
  • Locationmy Wang is aiming at ur rear... torso

Posted 19 March 2013 - 03:48 AM

if i kill People im usefull to my Team ! thats it ! if i get 4 kills than i defenatly did something right.
kills are mor importand than this ****** damage done thing
damge done with no kills isnt usefull but taking aim and hitting ur oponet in the face is usefull.

#38 John MatriX82

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 2,398 posts
  • LocationItaly

Posted 19 March 2013 - 04:01 AM

KDR doesn't mean much, you may have got it high since you snipe and rarely get bothered when doing it.

I had once a quite high KDR (and for being a brawler constantly in the teeth of the enemy it was way high I was nearing 11), but then I decided to exp cicadas and then Awesomes and my KDR went down by 2 and a half points.

Then ELO came, and I keep constantly getting paired up against teams with tonnage advantages ranging from 100 to 250 tons. Therefore it's quite natural to have a constantly decreasing KDr, especially when you don't want to keep playing Atlases or Assaults to counter this nice and uneven ELO matchmaking..

Edited by John MatriX82, 19 March 2013 - 04:03 AM.


#39 Lyteros

    Member

  • PipPipPipPipPipPip
  • 456 posts
  • LocationGermany

Posted 19 March 2013 - 04:05 AM

TL;DR:
KRr, WLr, DMG are all useless. The only stats that are right now actually a measure of skill are hit % and the state of your paperdoll when you die, showing you how much damage you have absorbed (or not) before dying, thus showing your piloting skill in manouvering and shielding.

Full version:
A single stat is always flawed, especially out of context. I would Ignore Kill/Death ratio altogether. Win / Loss ratio is not better, you're with a team of 8 people you dont know and fight against 8 other people who want your head. ELO is an improvement to the complete random matchmaking, but it is still flawed and misses out on a lot of information.

It's much easier to die in a light then in a assault and its way easier to get kills with an assault then with a light. If you're going all warboss and lead the charge in your assault, you're very likely to die, but if you're good you can still do a lot of damage and soak even more damage while the rest of your team advances relatively clean. Often leads to only the lead assault and maybe 1-2 allies dying, while the enemy team is ripped to pieces by the rest.

If you want some stats that say something about you:
Unguided Long range weapon hits (like gauss and ppc) so you see how good your aim at high ranges is, SRM hit rates, so you see how good your preaim and firing discipline is (firing on 200m will yield you less then half of the possible hits) laser weapons to see how good you're at tracking a moving target.
In addition the rate of your remaining armor divived by your full armor, to see how much damage you sokaed up through turning and proper positioning before you died (so if you're cleanly cored out its bad, if you're all internal before you die it's good since you spread the damage). Sadly we have no stats for this, but a somewhat imprecise measurement for this is the paperdoll after you died.

Take these stats together and see how good your piloting and aiming is. For anything else we do not have stats / measurepoints.

Damage done is also imho not a good stat for skill comparison, if you do nothing but boat LRM you can achive a high damage value with pretty much no skill needed, or a little more LRM damage with a little skill - but the damage in both cases will be spread all over a mech and not pinpointet into a targetet location. Be it the weapon filled side torso (Huch, Atlas), the head of a cat, or generally the center torso.
I'd always say 300 damage that cleanly core out 3 mechs are much more skilled 1200 damage that kill 3 mechs and take more then double the time.
The problem here is when someone is already severely damaged or a pilot actually wants to steal kills, it will inflate and blur the stats.

#40 Exoth3rmic

    Member

  • PipPipPipPipPipPip
  • Rage
  • Rage
  • 434 posts

Posted 19 March 2013 - 04:14 AM

Whatever your K/D, if it is low already then look to your win/loss and maps played etc.

If your win loss is considerably below 1.0 (i.e you win considerably less than 50% of the time) then obviously the problem begins to crystalise in to you simply aren't doing enough for the team to justify your position in that mech.

Reasoning being, if you die a lot and get few kills but your team contribution is enough for your team to benefit from whatever activity you are doing then your team will win and the MM will note that and eventually you will approach some semblance of sanity in the ratio of wins to losses.

If, however, you die a lot get few kills and your activity is not conferring an advantage to your team mates you will lose, the MM will note that and eventually you'll end up at the bottom of the pile where you simply are the worst at the game in that mech and you will likely always need carrying as you are holding any team you are on back. Still, in most cases such a player would have quit the game by that point and someone else can take the crown as king nothing.





8 user(s) are reading this topic

0 members, 8 guests, 0 anonymous users