76 replies to this topic

[Malcolm Vordermark]

K/D can be a useful stat but it isn't conclusive evidence of anything. A RVN-3L with ER Large Lasers can clean up at the end of a match and contribute nothing to the team otherwise. An AS7-D-DC can break a firing line with his push, winning the game for his team, and get zero kills. An entire lance can wander off and get killed leaving you and the rest of your team with a huge disadvantage that leads to a 12-0 loss.

If you can maintain a positive K/D in an environment such as this then I think that speaks well of your abilities. If your K/D is not positive I do not necessarily think it is because you are bad.

[Dunning Kruger Effect]

Quote

I think measuring K/d is useless because kills are last hits. So say you are a monster and your doing 80% of the damage to 10 to 20 mechs and each time someone else gets the last hit does this mean since your k/d ratio blows your not the one truly doing the work. I agree with majority of posts that damage and teamwork is the measure of a good player. Someone with a high K/d usually just means he has something that has rapid fire or gets lucky with last hits not that he is better than another player with less k/d. This isnt counterstrike K/d went out with old style fps where whoever got the shot off got the kill, now its damage, tactics teamwork.

Teamwork, sadly, is not measurable as a stat.

Kills DO have value, as they are the way you win (by killing the enemy team whilst not losing your own mechs).
Damage, if anything, has LESS relevance than kills, due to the fact that you can do lots of damage by "spreading" inaccurately all over the mech.

If you are a "monster" who does 80% damage to 10-20 mechs, and you have no kills, you are VERY poor at "finishing" enemy mechs. You need to learn to identify and shoot at damage components on an enemy mech. A player who can consistently do this (and get kills for his team) is not "lucky" but skilled.

High damage and low kills is a sign a player is (a) not targeting injured components ( a poor shot (c) using "spread" weapons like LRMs - neither of which is an indicator of "good" play.

In a game like WoT, where each tank has a single "pool" of damage, damage stats are more relevant. But in MW:O where a good shot can blow the head of a mech whilst only doing minimal damage, or cleverly shoot out the side torso on a XL mech, - components mean the "damage" stats have less relevance.

Quote

Assits are better.

Absolutely not. I can get 12 assists a match by brushing a laser over each enemy mech. In fact, the C-bill bonus makes it profitable to do so. Doing tiny amounts of damage to multiple mechs has no effect on the enemy team, and does not mean I am a good player - I am just "gaming" the system for C-bills and means I am a SELFISH player who does not effectively focus fire, and secure KILLS.

The reason "assists" have value is because they indicate someone, somewhere, actually KILLED an enemy mech, thus making your team more likely to win. Do you get an assist if the mech doesn't die? Nope.

I have a very ordinary .7ish KDR, but the lack of logic in the "KDR doesn't matter" camp annoys me.

[Galenit]

Dunning Kruger Effect, on 19 March 2014 - 01:45 PM, said:

Is KDR useful?
A low KDR tends to indicate you are dead more often then average, leaving your team shorthanded more often. I.e. a ECM Spider with a 2:1 kdr would suggest he is likely more useful to the team then one with a 1:2 kdr.
...
"I have a low KDR therefore I am a good team player?"
A low KDR is just as (if not more) indicative of someone who make bad choices for staying alive or killing enemy mechs - i.e.a low KDR is more likely to be someone who will simply banzai charge into the enemy and die stupidly at the start of the match than a selfless heroic flag capper.

A spider getting 2 nearly cored mechs with its mgs, lived in the end with 200 damage is more usefull then a 10 assist and 30+ spotting/narc/tag/uav/counter ecm assists raven with 0 kills that died with 58 damage?


Sure, the spider has killed 2 mechs, that were near dead.
But the raven has brought the rain of death to 8 enemys and has prevented flanking with its uav/narc.
Do you really think, that that spider is more usefull for a team then that raven?


The only stats we have that say a little bit about your share on winning matches and your usefullness for the team is win/lose and average xp.

Edited by Galenit, 27 March 2014 - 03:53 AM.

[Macksheen]

Appogee, on 21 March 2014 - 03:33 AM, said:

I created an Excel spreadsheet to analyse mine. I can copy and paste the Mech stats into it, and it will show me the average damage per match per Mech.

PGI could calculate and display this information very easily on the Mech stats screen if they wanted too. It's a simple calculation of the existing total Mech damage divided by the total number of matches played in that Mech ... both datapoints are already captured and displayed separately.

I must confess to doing the same - though I use Google docs. I tend to look at more than k/d ... k/match ... damage/kill ... damage/match ... xp/match .... for each chassis. One reason I look at XP is that assists (ECM, spot, whatever) aren't tracked, so XP is about the only way to guage non-kill, non-damage help. Then, there are the intangibles ... had a match yesterday with an awesomely done pincer. Six of us stayed and pretended to be the whole group, got shot up while the others circled around. it was rough and scary for a bit with those who stayed taking a beating, but we ended up 12-4 with the win. Some of that may show up in the win/loss, damage, kills - but otherwise, that is hidden from view on the stats.

Since the stat reset I have some wonky numbers that, while they amuse me, don't tell any real long-term story. One of my favorite mechs which I tend to rack up assists, etc. has had zero kills since the restart, making my kills/death look pretty abysmal. My Misery on the other hand was at a 15 kills/death ratio after a dozen matches ... which I screenshot for long-term chuckle value ... but one match later and I'm at a 9 k/d. My overall, while positive, is no where near that - it's just a matching outlier to my other flipped ones and completely meaningless on its own (except maybe to say that I find Misery to be a righteous beast).

Also, for some reason I identify most mechs I have as male. Despite the phallic shape, i can't help of thinking of Misery as a female. SHE is a monster.

That doesn't show up in the stats and k/d either ;-)

[worm4981]

I think you cant use the stats to compare yourself to other players, there are too many unknown factors (ELO!).


You can, however, use them to compare your own performance with different builds/mech types.

My dragons K/D surpass nearly all my other mechs, second only to the HBK-4P. Thats because they were build to "steal kills" as some would call it. This is not the correct term imho. I focus on exposed parts while the enemy is busy slugging it out with our AC meta boats. While they spread the damage all over the enemy, I get behind them and take out their weapons or side torsi when I suspect a XL engine. I wont immediately engage any enemy I see, I try to pick wounded targets, always with the thought in mind to reduce the enemys fighting strenght as fast as possible.

This is the best way to remove firepower from the other team. Mechs dying in the process is an added bonus but not the goal here. My 4P does the same by the way.

Now, my damage/match is usually quite low with these mechs because I do pick my shots very carefully. Shooting around just to get some more damage tends to alert them to my presence and keeps me from lining up a GOOD shot later on.
But I tend to have a good W/L rate on these mechs because they are effective in helping the team by reducing enemy fire power early in the match.

My Heavys/Assaults on the other hand have a bad W/L (exept for 2 or 3 that I could wrap my head around) because they do not fit my playstyle at all. I may deal good damage with them but I fail to help my team enough to secure a win.

Note: I almost always PUG, so the W/L is really just my own (bad) performance.

[C E Dwyer]

K/D means squat, any one who says otherwise hasn't a clue what this games about.

Take it from some one who started playing this game had a really crap K/d under .4 a match. built on it had it over 1.2 at the time of reset.

Now its close to 3 since reset

Has this made me a better player no

The only thing that makes a person a good player is the damage output the rest just falls in place. (if you must use stats as a measure of how good you are then use the first collum as it factors in, damage, assists, kills, part destruction, and shows how useful you were in that match.

This changes during PGI tournaments because PGI use a dumb scoring system and you score 20 points per kill and 10 per assist (they should be the same as the kill shot was only there because of the damage that other team players did in nearly every case) thinking back I had six noted moments in my banshee making a CT red only to have some ankle biter light, alpha into it after having not fired for three medium laser cool down cycles usually a Firestarter or a Raven once was a Spider.

But other than for tournament scoring Kills mean nothing, but for some sad stat warrior, to say you know nothing because my K/d is higher than yours so there, in forum wars

Edited by Cathy, 27 March 2014 - 05:07 AM.

[C E Dwyer]

Rouken, on 26 March 2014 - 07:39 AM, said:

K/D can be a useful stat but it isn't conclusive evidence of anything. A RVN-3L with ER Large Lasers can clean up at the end of a match and contribute nothing to the team otherwise. An AS7-D-DC can break a firing line with his push, winning the game for his team, and get zero kills. An entire lance can wander off and get killed leaving you and the rest of your team with a huge disadvantage that leads to a 12-0 loss.

If you can maintain a positive K/D in an environment such as this then I think that speaks well of your abilities. If your K/D is not positive I do not necessarily think it is because you are bad.

Amen brother

[Dark Jackal]

I posted up a suggestion for Kills + Assists / Deaths (K+A/D) for a new stat:
http://mwomercs.com/...4-new-stat-kad/

Though we should probably have most of the flat statistics the game tracks when it awards you cbillies.

[meteorol]

For me, the stat reset was... relieving.

I swore to myself never to care for my k/d again, and the game is much more fun now.
TBH, i'm a somewhat competitive guy, and i can't stand losing. Can't even stand not being good at something.
If i'm trying something for an appropriate time and don't get good, i just stop doing whatever it is.

So making this step, not caring about my k/d or any stat in particular was not easy. But i'm glad i did it.
Before the stat reset, i would check my global k/d after every day (sometimes even after each match)
3.20 and raising with each match, with above 4 on my "good" mechs. (ironing out the fails of my earlier days).
Even lowering it by 0.01 on a day left a sour taste.

K/D is by no means an accurate stat on how well you did, but it was the only true indicator i could check without calculating stuff like average damage per match or average damage per kill.

Scoring high, many kills, many assists, efficient damage, winning, winning, winning was the only thing i cared for.
At times, playing the game was more exhausting than fun. The results were ok (3.20 k/d, 1.28 W/L) but sometimes the need to do better was almost oppressive. Maintaining good numbers in a game like MWO where your personal influence on the match outcome is rather limited can be a buttload of work at times.

So i used the stat reset as a cut.
Now, i'm doing what i like. I used to play only the best mechs with the best builds, now i'm using all mechs (even giving trail mechs a spin). I don't care if me or my team does well. The game is much for fun right now. Honestly, i wouldn't mind if they would remove stats all together.
.

My global k/d dropped down to like 2.6 (don't know exactly), w/l dropped to around 1.0, average cbills took a hit of like 10k.

But i don't care, and i'm having more fun than i ever had with this game. Screw those stats.
I have very limited time to play each day (maybe a hour at max). No need to spend my precious time getting angry.

Edited by meteorol, 27 March 2014 - 06:01 AM.

[Enigmos]

Playing 'better' is what you should evaluate. K/D ratio is a measurement independent of context. Now you have to figure out what 'better' looks like.

Example: You're on Alpine and your team is pinned in trench/sniper warfare. Attrition will place your team at a disadvantage. They have the high ground and better weaponry for the terrain.

You notice most of the survivors with you are brawlers loaded for medium range or closer. They've survived so far because they have nothing with the range, so they didn't expose themselves.

If you have an opportunity and can break the enemy line you can let your brawlers gain optimal range against a team that is hopefully then in disorder.

Then, even if it will mean certain destruction for you personally, your choice to be the bowling ball is good play. You chose to die breaking their sniping war into a brawling war that your side has a chance to win.

Being a good player has more to do with making the best decision possible, ready and willing to act. It is better to kill the other guy and survive, yes. Sometimes you're going to die anyway no matter what you do, so watch for opportunity and capitalize on it without worrying about the dumb K/D.

[Dunning Kruger Effect]

Quote

A spider getting 2 nearly cored mechs with its mgs, lived in the end with 200 damage is more usefull then a 10 assist and 30+ spotting/narc/tag/uav/counter ecm assists raven with 0 kills that died with 58 damage?

Sure, the spider has killed 2 mechs, that were near dead.
But the raven has brought the rain of death to 8 enemys and has prevented flanking with its uav/narc.
Do you really think, that that spider is more usefull for a team then that raven?

The 10 assists by the ECM Spider shows, someone, somewhere was getting the kills needed for victory for his team. If no one did this vital task (killing enemy mechs)... then your team could not win (except maybe on conquest). This is why kills and deaths do matter.

If his team were all people who did "great damage" but got no kills - you'd be left with 12 enemy mechs that are still at least partially functional. Meanwhile the other team (which shoots out heads, and XL side torsos) does a fraction of the damage but kills the entire enemy team.

Anyway, I'd like to stay "on topic":

Let me repeat my arguments:

#1. KDR is NOT meaningless (it's not the only, the best or even a very good metric for "skill" but it is certainly not "worthless" in a game where destroying enemy mechs (kills) and staying alive (deaths) is the way to win)

#2. Assists mean nothing (I can brush a ERLL over every member of the opposing team quite easily, doing nothing for my team's chances of victory)

#3. Damage is not any better than kills (accurately blowing off a head or coring a damaged mech, killing it and doing 50 damage is far more useful than a player who sprays fire all over the mech, spreading 300 damage and leaving the enemy functional and intact). Huge damage stats but no kills is often a sign of someone who cannot shoot accurately or uses weapons like LRMs.

KDR is not a overly important stat. But to say it is meaningless or worthless is silly
(In any case, the damage and assists stats loved by low KDR players are even more "meaningless").

[Galenit]

Dunning Kruger Effect, on 27 March 2014 - 01:57 PM, said:

Anyway, I'd like to stay "on topic":

Let me repeat my arguments:

Maybe my bad english made this, if you dont understand what i wrote, just ask. Sorry for the confusion.


But i stayed on topic.
Let me repeat my example:

Spider finishing 2 near cored mechs with 200 damage and some assists.
No ecm involved, because the example was about a spider with machineguns.
Kd 2/0

Raven narcing 8 enemys that die to missiles,
hitting 10 enemys, but doing nearly no damage
Preventing a enemy flank from 4 enemy mechs under ecmcover with his uav.
Dieing in the end.
Kd 0/1

Did the spider really more for the teamvictory then the raven as your arguments suggests?
What did the kd from this match say about the stat "k/d"?
Are assists really that unimportend?

In the example, the 3 missileboats share 8 kills, but without narc, tag and counterecm it would be a lot less.
In the end the ravens share of this would be 2 kills, but they are not reflected in the stats.



Lets talk about focusfire.

Every player with some skill know what it means and how importand it is.

If 4 mechs fire at one target (lets say they are all able to hit), there is no skill involved in getting the killhit, its just luck.

Or holding fire and trying to push the own kd with timing the killing blow, but letting the enemy doing more damage to you or an ally, weakening the team.

Edited by Galenit, 27 March 2014 - 06:37 PM.

[Dunning Kruger Effect]

Well your English is excellent - probably better than most English speakers! (It's also a billion times better than my German)
I'll try to match up your examples to my arguments.

Quote

What did the kd from this match say about the stat "k/d"?

So you're saying (as per #1) "KDR IS meaningless" with the Raven/Spider example?
A. What I got from the example was "there are a lot of things that happen in an individual game, that are not properly recognized/cannot be statistically measured" which I agree with.

However, your team overall, by winning say 12-6, got a 2:1 kdr. Most of your team got a positive KDR for the win. Over time, players who win consistently should gain a better KDR than habitual losers.

Remember, statistics are only meaningful over time, with repeated examples. One-off examples (or games) are not indicative of ability. I might have a terrible round, or a great round, with a particular mech. It's like flipping a coin once. If it lands "heads" it does not follow the other flips will be always heads. Flip the coin 1000 times, and you will quickly figure out the chances are 50/50.

If you say "KDR is meaningless within a single game" than I agree with you. You might do 900 damage AND be aiming carefully at crippled mechs but have idiot team mates jump in front to steal the kill. (I also agree that a over-focus on KDR leads to poor play)

However you can track your stats over many, many games, which means they are more valid.

If your Raven and Spider K each play 100 games, and the Spider gets 200 kills and 50 deaths (4.00), and the Raven gets 50 kills and 200 deaths (0:25), who is "better?" It's not completely clear, but I'd say a Raven pilot with that KDR needs to improve both his ability to make killshots and survive the game, and that the Spider pilot is very good at "finishing" opponents (removing an enemy threat is ALWAYS a good thing).

But lets compare two mechs that are the same (because we have individual mech stats, we can do this)
If one Raven pilot has a 4:1 KDR and one Raven pilot has a 1:4 KDR, the other MAY be a genius NARC-recon-scouting-ECM marvel, but we know FOR SURE the first guy is better at killing enemies and staying alive.

KDR has got "some" context- important for using stats - as we can track individual mech types, over many games - if I am getting a 0.8kdr in 50 games in my Jager I might be unhappy, but I might be quite happy if that is my KDR in a Locust.

But "kill death ratio" is NOT meaningless or useless (although it can be VERY misleading within a small sample size or without context).

Quote

Are assists really that unimportent?

B. No, but they are so easily "manipulated" as to be meaningless. I.e. I use my Spider 5K to brush every enemy mech with a ERLL and collect 12 assists. I guarantee you I could "farm" 6-10 assists a game this way, each game, every game, without actually contributing to my team. At least if I made a "kill" I remove an enemy threat and move our team 1 kill closer to victory.

Maybe if an "assist" was doing say 20% damage of the total damage points of a mech, or removing a component, then "assists" would be more meaningful. Landing a single point of damage on a mech is not necessarily a meaningful contribution. If it killed or crippled the mech, or removed a component or weapon, then yes, that single point is meaningful.

C. In your "focus fire" example I agree that the "killing blow" can be luck. I also agree that people who time their firing to farm kills can hurt their team. In fact, people who obsess over KDR are often bad team mates.

However I'm not sure what point you are addressing? Are you saying people who play well can be "unlucky" so it modifies their KDR? In a single game, sure. But if you "focus fire" with your friends 100 times, you SHOULD get your "share" of kills - and if you are focus firing with your team, you should be getting more kills per game than someone who runs off lone-wolfing to die. Your KDR should be solid because you are killing more and dying less (as your effective focus fire "kills" enemy threats).

I don't see how focus firing makes KDR meaningless.

Edited by Dunning Kruger Effect, 27 March 2014 - 11:01 PM.

[Galenit]

Dunning Kruger Effect, on 27 March 2014 - 10:56 PM, said:

But lets compare two mechs that are the same (because we have individual mech stats, we can do this)
If one Raven pilot has a 4:1 KDR and one Raven pilot has a 1:4 KDR, the other MAY be a genius NARC-recon-scouting-ECM marvel, but we know FOR SURE the first guy is better at killing enemies and staying alive.

But "kill death ratio" is NOT meaningless or useless (although it can be VERY misleading within a small sample size or without context).
......
B. No, but they are so easily "manipulated" as to be meaningless. I.e. I use my Spider 5K to brush every enemy mech with a ERLL and collect 12 assists. I guarantee you I could "farm" 6-10 assists a game this way, each game, every game, without actually contributing to my team. At least if I made a "kill" I remove an enemy threat and move our team 1 kill closer to victory.
...

I don't see how focus firing makes KDR meaningless.

Kills are easily manupulated too.
Look what happens at the tournaments....

In your raven example, the first raven maybe tries to push his kd, looking for bad damaged mechs doing the killingblow or he can do well and do real kills that help the team.
We dont know.
He can be good at killing other mechs or at killstealing, he can be good at piloting or hiding in the end of a lost match. We dont know, kd does not say anything about it.

The second raven, does not have any killingblows, but without him, his 3 teammates would not be able to get their 8 kills. His share of the kills is 2, making his kd for 1:4 up to 2:1, but thats not reflected in the stats.
Maybe he sucks and narcs every tree he passes by, maybe he is the good scout and increses the chance of winning, we dont know, kd does not tell the story.

In the end, the kd of both ravens say nothing about their skills, because we dont know how it was archived.

Without the context and all other information kd in this example is meaningless.


Blowing parts off / focus fire:
i have seen more then one assault removing the biggest guns from an enemy, letting his allys finishing him off, during that he turns to the next enemy with a big gun to remove it ...
This is not reflected in kd.

Blowing parts off 2:
If i go in my light or medium and meet a single mech with short range weapons, i remove a leg and the go on. In the time he needs to snail to the fight, we have finished his friends off.
This is taking a enemy out of a fight without killing him in the first . Is this reflected in kd?

Staying alive:
Some try to stay alive, some hide, some sacrifiece them self in the end.
If my stalker has no weapons left, i can hide or i can go to the enemy, block his shooting. This gives him the kill on me, but it saves my ally from the damage, let him cool down his mech and gives the enemy some heat, increasing the chance that my ally will kill the enemy and we will win the match.
This reduces my kd, but it helps my w/l. Does kd tell this story?


Average xp says more, because it reflects not only getting the killingblow and surviving, it also reflects assists, counterecms, tag, spotting, blowing parts of, ...


After talking that much about stats, i have looked at my:

My raven went down in its kd by more then 50% with the new stats, because i reduced its weapons to some mls and have taken the whole e-war package.
But the average xp a match increased by nearly 50% (getting more then my stalker now) and the wl increased too .

My stalker has around 1.75kd and 850 axp pre and after patch.

If kd would mean anything, it would say that i have become a bad player with my raven.
But why do i get more wins and 50% more xp then?

Edit:
I like playing with numbers, but stats mean nothing for me.
I evaluate my games by the fun i have.

Most fun matches are when only 1 or 2 stay on its legs (or 1 leg) when its done.
Mostly that are the matches where the most players have 6-8 assists and the kills are shared between 6-8 players.
After 13 minutes of fun and tension it does not matter who wins.

Edited by Galenit, 28 March 2014 - 03:23 AM.

[Funkadelic Mayhem]

To each their own.

[Steinar Bergstol]

Meh. Bollocks to stats. If I'm having fun then I'm having fun. Anything else I'm getting too damn old to worry about

[Dawnstealer]

My playing style doesn't really lend itself to a lot of kills. I don't mind being the door-breaker or the skirmisher that gets behind enemy lines and gets them all to turn around. My life-expectancy doing that is real short, but my team will tend to win more - the enemy's all out of position trying to kill me when the push comes and then they get slaughtered.

Some guys REALLY care about that K/D - I just want to win.

[Danghen Woolf]

I applaud people with high K/D. I do not focus on it though, I play to have fun. I run mechs in their roles, CPLT-A1, fire support, HBK-4G, brawler, COM-2D, recon/harassment. I do not subscribe to meta-builds, I enjoy tweaking the mechs, DHS, Endo, etc. My K/D was 0.27 before stat reset and is 0.32 now. Do I die alot, oh yeah, do I sit around waiting to finish off a crippled mech, nope, I play the objectives, I flank, I TAG, I brawl, I spot for teammates. I have no problem getting my HBK in the mix and dumping AC20 rounds at anything I can. I play for fun. Count the K/D if you want but you should be counting teamwork. An AS7 pilot who wanders off by himself and gets killed by a light lance, even if he does get a kill or two, is worse to me than someone who uses their brainpan for more than glory chasing. Does it make sense for an assault lance to be drawn off by one or two light mechs to chase a kill when there are bigger nasties out there for them to take on? What if chasing those light mechs leaves the enemy assault lance free reign to take out a couple mediums or fire support guys while the AS7s and HGNs chase a couple SDRs through the terrain... just my opinion, but K/D is not a true indication of battlefield skill. Willingness to work with others is much more important.

[Barkem Squirrel]

At first I though, god I need a high KDR, then I played lights for months at low frames per second. I was having fun, but kills were few and far between. I had to clean up my computer and still run below 20 fps, but the difference is huge. I can leg a raven with the Jager at 1200 m and call it in to the lights in chat. Pick that damage side torso to focus on and even decide when to stop firing LRMs at a target so I do not waist LRMs on a target that is dead.

The last two days I have been running Locusts completing the Basics. I was having 300 damage games with the little ankle biters. I even almost got the bridesmaid title, with 12 assists, but I had to fire one last time and kill someone for a kill and 11 assists.

One big thing to look at is how many deaths to matches with a mech. Running a Locust with TAG and 4 SL/SPL's is fun. Or the MG's. I look at it as, with the locust I am just securing the kills in a timely manor. I am more important to watch, TAG or be that LRM launching bug on the wall from the flanks or behind.


the longer you damage someone the more damage that they can do to you and some other teammate.

I do not care if I get the kill or someone else, just that they go down fast. LRMs I can open them up, but change targets before the mech is dead and start on the next mech.

[AllOuttaBubbleGum]

I like KD to track performance of my mechs. But when playing with a team the KD is not as strong as a positive indicator as working with team mates, sacrificing one self for the win, allowing others to get a kill while you rescue a teammate who is getting gang banged..etc..