36 replies to this topic

[Enigmos]

Many players appear to focus most on the K/D ratio as their measure of worth in MechWarrior Online. The ELO appears to focus more on Win/Loss. Yet when I build my loadouts I tend to focus on how much damage I can do given the hardpoints available on a chassis, and the chasses I choose are selected based on the hardpoints that will be most useful for my intended role.

When chatting and reading posts commanders tend to express some frustration with the way the current stats are presented because K/D does not seem to factor for support roles, especially scouting/spotting/Tagging, and W/L alone doesn't tell much about individual contribution. A high W/L may hide poor contribution.

K/D may be dismal for even highly skilled pilots. For example if the pilot is using a high damage-over-time weapon like your missiles, AC2, LB X-10 the enemy is seriously weakened but there isn't substantial kill-power and the final shot often yields the kill to a teammate in what is commonly called a 'kill steal'.

Yet effective teamwork should not have such a concept as 'kill-steal'. Part of the effectiveness of a coordinated drop using good voice communications is designated targeting where everyone who has a shot on a specific mech will fire on that target to burn him down faster. Coordinated fire is indispensable for a lance.

Yet, overwhelmingly, most pilots that I have discussed these issues with list the kill/death ratio as the most significant measure of a pilot's worth.

I think, if it is a problem, part of this metric evaluation stems from the way tourneys are set up as individual competitions. Even though the formula does factor team wins as more valuable than losses, competitors must drop ungrouped for the match to be counted.

So I have a couple of questions I'd like to hear about, and there ar surely other issues you may feel are important besides.

1. What do you evaluate yourself as a pilot by?
2. Should we have this concept of 'kill stealing' in a team-centric game?
3. Do you have any suggestions that might improve the metrics we commonly use that will better promote teamwork over individual performance?

[Clownwarlord]

Both, why not have a good K/D and a good W/L? If you can kill 3 or more mechs then obviously you are helping your team out which means you are more likely to get wins (objective of every game is to remove mechs from the field and you can get a win ... mind you not only way to get a win but it is a way in every match).

I try to have a K/D over 1.5 and a W/L over 2 ... but hey with stat wipe I got a lot of work to get it there and then keep it there (just seems like a lot of pugs have fallen down in their skill level though recently).

[PhoenixFire55]

KDR and Elo are meaningless.

1. By being/not being able to perform well in games with regular top 10 tourney players.
2. No. But as long as another guy is also shooting the enemy mech the entire time not holding his alpha till the very last moment I don't mind if he 'steals' my kill.
3. Difficult, but something like they do in sports maybe. Average kills/assists/damage per game. This is a truly complicated matter, as its always very hard to properly judge just how much one individual contributes towards team success.

[Bilbo]

OriginalTibs, on 21 March 2014 - 04:46 AM, said:

...
K/D may be dismal for even highly skilled pilots. For example if the pilot is using a high damage-over-time weapon like your missiles, AC2, LB X-10 the enemy is seriously weakened but there isn't substantial kill-power and the final shot often yields the kill to a teammate in what is commonly called a 'kill steal'.
...

This is why K\D ratio is meaningless to me, though I wouldn't call the guy who got the killing shot in a kill stealer. If my Atlas is actually supported during a push or flanking movement, I rarely get a kill shot. I bring the pain and the supporting mechs usually end their misery.

[Marmon Rzohr]

Bilbo, on 21 March 2014 - 05:08 AM, said:

This is why K\D ratio is meaningless to me, though I wouldn't call the guy who got the killing shot in a kill stealer. If my Atlas is actually supported during a push or flanking movement, I rarely get a kill shot. I bring the pain and the supporting mechs usually end their misery.

KD isn't meaningless. While damage done is pehaps the most important stat (after win rate, IMHO), when accompanied with a low KDR it could tell you one of 2 things:
1) You die when you could have survived by being a bit more careful
2) You are not efficient in taking down enemy mechs

Both are quite important:

1) In a close game staying alive longer to fire a just a couple more shots can mean the difference between a win and a loss. There's no point in taking a lot of damage to take down a mech which has a much smaller impact on the game then you do, for instance.
2) Efficiency on the other hand is essential when you need to carry a game for you team. You need to eliminate as many targets as you can with the least amount of wasted effort.

Number 2 is a big part of why a lot of people perfer PPCs over lasers and larger AC5s over AC2s and why the LBX-10 is inferior to the AC10.

DISCLAIMER: there are some exceptions, like LRM boats and light mechs who usually get less kills per damage done and more kills per damage done respectively.

Edited by Marmon Rzohr, 21 March 2014 - 05:33 AM.

[b1ackadder]

I think I have an idea there:
Usually the so-called kill-steals appear when you're in a 1on1-fight, got the other guy down and some "mate" comes around to deliver the final blow.

For supporting teamwork, I suggest the following:
By pressing "R" you lock a target and everyone is able to see that on the minimap. Okay, fine. Nice "teamwork" there.
BUT: What about kind of a "primary lock", marking a specific target for your complete lance as primary? This might be a way to tell your lance "Oh good lord help me out here!" and/or "Focus fire on that ONE guys and noone else".

Advantages:
- Improved communication even for PUGs!
- No more chat required that no one will read during a fight in the first place
- Quick and simple way to act as a silent team: Point out -> Shoot. No discussion

Disadvatages:
- Should be limited on lances, otherwise during a hot fight you'd have 7+ "primary" targets (in 12vs12 on the other hand, it could be restriced to the lance commanders, should suffice)


According to that, there could be bonuses for a group kill, for giving a primary target, and so on.

I know this is not an exact answer to the posed questions, but I think it is a way to improve teamplay in a good way, even in matches were you don't know Jack (who's that anyway...).

Back to the questions:
1. What do you evaluate yourself as a pilot by?

As PhoenixFire said before: "By being/not being able to perform well in games with regular top 10 tourney players." Being up to #19 in the last tournament was a great experience.

2. Should we have this concept of 'kill stealing' in a team-centric game?

To be honest: In a fight, you shoot at the enemy. I don't wait for him to shoot my teammate, and while turning around to their fight I won't have time to check on my mates health - I will shoot. That's teamplay for me. I do not intend to killsteal, but I don't wait until my team dies to farm the kills. Some will see that as killstealing, I don't - and I guess the definition of "killstealing" is varying by who you ask.
Actively blocking shooting teammates to get to the kill is douchey, still.


3. Do you have any suggestions that might improve the metrics we commonly use that will better promote teamwork over individual performance?

See above.


I'd like to hear some feedback on my idea there, so feel free to quote me.

Edited by b1ackadder, 21 March 2014 - 05:37 AM.

[Bilbo]

Marmon Rzohr, on 21 March 2014 - 05:28 AM, said:

KD isn't meaningless. While damage done is pehaps the most important stat (after win rate, IMHO), when accompanied with a low KDR it means one of 2 things:
1) You die when you could have survived by being a bit more careful
2) You are not efficient in taking down enemy mechs
...

If ignoring mechs that no longer pose a threat over engaging the next actual threat is inefficient, so be it. I like getting the kill shot as much as the next guy, but tunnel vision doesn't necessarily get a win for the team.

I can't argue with #2. I stick my nose out far too often, I'd much rather be the one to make a flanking maneuver than sit and wait.

[b1ackadder]

Bilbo, on 21 March 2014 - 05:43 AM, said:

I stick my nose out far too often,

Having that username you literally HAVE to....^^

[AlmightyAeng]

Honestly, it's a combination, and you kinda gotta rate yourself and ignore what anyone else says. Here's why those values are all problematic.

1. Damage. If you "bring the pain" with a 900 damage game, did you do well? If that 900 damage is spread between 2-4 kills and multiple assists...I'd say absolutely! If it's all on a single Atlas that you eventually kill? Well, sounds like your fire discipline was shoddy...you hit it EVERYWHERE instead of going for the CT for the killshot...and so ended up using WAY more time that could have gone to helping the rest of your team. If you were in a Spider that did it? You picked apart a mech 4 times your size and distracted it from being in a brawl. There, you WERE playing smart. It's about context.

2. Kills. Are you running around looking for damaged 'mechs JUST trying to steal kills? There's nothing wrong with giving it all you've got, helping kill things...but if you're the tiny guy getting that last shot in all the time...well...would you say you're more valuable than the guys doing all the damage BEFORE you show up for that last shot? If I'm wracking up multiple kills in a tiny guy like that, I expect to see a certain level of damage to make sure I'm not just mooching off of others...and I try to help out teammates that are OUTnumbered...not get into the middle of a 5 on 1 and steal the kill.

3. Win/Loss. Well, PGI has stated that ELO is set up to intentionally make your wins and losses as close to 50/50 as possible. When the system is throwing you into matches on purpose that it thinks you'll lose to even that score out, I find W/L ratio to be pretty useless. For the individuals who manage to have an excellent Win/Loss ratio...it either means they are "SUPER LEET" or have a helluva 4-man they drop with. Depending on their loadouts, I am more or less impressed.

So, after each game, review your performance, and think about those things, and compare your numbers to the REST of the people in that match. That'll help you judge whether or not you did well. But just the numbers, without the context of the games? Doesn't mean a whole lot in my book.

[Magna Canus]

1. There are already some good answers here, but personally I measure myself by the "feeling" of my performance. You should usually know yourself and your performance well enough to say you did good, or you sucked. So many factors influence the numbers on the screen, especially in a team game, that the numbers alone can't give you the whole picture. In general I "like to have" 2 kills and over 300 damage, then I am happy.

2. Kill-stealing; Yes, there is actually a thing like kill stealing. It should not be one, as others have said. A slain enemy is a good enemy, but there are enough people out there that purposefully shy away from combat until they spot a wounded mech engaged with a team mate. They watch the battle, wait until the "time is ripe" and Alpha. I don't mind it if a team mate gets the kill when we were both working on the same target, but the above sort are the ones who brough the term "kill steal" to MWO.

3. Metrics to improve team work; We "kinda sorta" do, at least in the direction of C-Bill/XP rewards for things like components and assists, but none of that really factors into any stat in a meaningful way. It factors into W/L in a vague manner, but I am sure at least a few people reading this have carried hard for the under 100 damage crowd and know that a win can happen even if you sit on your thumbs. I can't really forsee any stat that could noticably encourage team play.

[The Basilisk]

Since win/ loose is next to random and highly dependable on the Mechs and loadouts/ Pilots you are thrown in with, W/L ratio doesn't say a thing about you personaly, since you cant realy influence it.
Sadly the D/K ratio on the other hand is the only measurement of how well you can survive and be nevertheless deadly/usefull at the same time.
Even low dmg + high K/D does't mean you are a killstealer.
Its just the result of the fact no mech survives 3 LPULSE to the Head.
Jokes aside even a maxarmored atlas gives you just around 180 dmg if you core him mano a mano from front through CT. ( Yes it depends on his armor distribution its a number derived from experience ) But basicaly manny kills with low dmg show you are a good shoot or at least able to read the enemys dmg diagram.

Elo does not depend on W/L but on the ELO of the enemys you defeated.
To simplify it you gain ELO when you defeat a higher ranking player and loose if be defeated by lower ranking.

Edited by The Basilisk, 21 March 2014 - 06:06 AM.

[Furiel]

OriginalTibs, on 21 March 2014 - 04:46 AM, said:

So I have a couple of questions I'd like to hear about, and there ar surely other issues you may feel are important besides.

1. What do you evaluate yourself as a pilot by?
2. Should we have this concept of 'kill stealing' in a team-centric game?
3. Do you have any suggestions that might improve the metrics we commonly use that will better promote teamwork over individual performance?

I'm still pretty new, but I evaluate my performance as a pilot on how much damage I did, how long I lived, how many assists I had and did we win. The first mech I mastered was a catapult C4, so my KDR was pretty terrible because A I was still learning and B I play the part of LRM support as it is supposed to be IMO, which is to rain LRMs on targets from a long ways away to soften them up and keep them under cover so the brawlers and direct fire support can advance to finish them off. That does not yield alot of kills like high damage pinpoint weapons do.

As such I don't subscribe to the idea of kill-stealing in this team game because it is a team game. We aren't lining up 12v12 to have 12 individual duels. It's Red vs. Blue may the best team win.

As for what stats to use, KVD and W/L ratio are used because they are available. To me I think avg damage and assist v death would be more valuable and easily doable since those stats are already tracked by PGI's systems. Ultimately though I don't think there is a good metric or set of metrics that can be used to really quantify what a good pilot is because there are too many variables that have to be accounted for the pure numbers don't show. Like if someone is scouting and calling out opposing force composition and location. Incredibly valuable but doesn't show in the numbers. Base/resource captures also don't show but in assault and conquest they can or are incredibly important.

[Modo44]

Compare to your team. Those are people reasonably close to your skill level.

Check your win/loss ratio. It is the closest thing to Elo, i.e. your overall skill level -- better than any single stat or limited stat set at showing how you are doing.

[Edustaja]

Play 12-mans. Usually the guys that can perform there are the better pilots overall. Good damage is good, good damage with kills is better. Watch for trends and compare performance among their peers and roles.

I'd especially recommend the MRBC league at the moment. It requires a lot of versatility and adaptability from both the pilots and the teams. The best pilots usually are those that can perform on a short notice and can do the roles they're assigned to.


In general good K/D at a high Elo usually is a sign of a good pilot. You know who the high Elo players are by dropping with them in 4-mans. Having a teamplay oriented attitude can compensate for an amount of personal skill as it's easier to organize stuff with team players.

I mostly don't appreciate players that are relatively low or average in skill yet have some personal quirk that they must bring to each match. Like using flamers or lrms on every mech or charging in mindlessly each time.

[Hillslam]

KDR is no indicator of how good or bad you are at efficiently killing mechs.

All it shows is how good you are at swooping in on a torn up mech your teammates have been working over and stealing the killshot.

Hurray for you.

Damage is so much more important. Also more lucrative.

I regularly end up, alive, at the end of matches with the highest win score in my Fatlas or BM or Vic, while my SPD teammates have like 200 damage and more kills.

They didn't "help me out" - I was in the process of killing those mechs anyway, in the most painful and lucrative fashion available. AKA all of those dead mechs were one or two salvos away from me dropping them. They were dead anyway. Congrats - you swooped in and MG'd a critical. Oh THANK YOU, oh high KDR saviour! All Hail the wizard.

Also - if you die before the round ends but have torn the enemy team a new a-hole in damage to the tune of 700 or 800 damage, don't tell me that was stupid. High damage dealers helped the team assure the win, and actually made it possible for those "super snapshot deadeye killers" to get their high KDR and come on here and assume they did it all themselves.

Edited by Hillslam, 21 March 2014 - 06:32 AM.

[Josef Nader]

1) Average XP per match -would- be a much better metric, but I believe it's influenced by premium time and hero mechs. K/D and W/L aren't useful at all at determining how much that player actually contributes. A light mech pilot who spends the whole match risking his tail to NARC or TAG targets in the open and keep good locks for his boats is going to have a much lower K/D than his teammates, and if he's pugging his contribution to the W/L is based on the number of LRMs on the field and how attentive they are.

tl;dr KDR and WLR is a lot less useful of a metric than a lot of people would have you think.

2) No. Kill secured. I'm just glad that we're working together and focusing targets.

3) We need more in depth stats that track more things and support roles beyond "do tons of damage and kill all the robots".

Edited by Josef Nader, 21 March 2014 - 06:25 AM.

[The Basilisk]

Hillslam, on 21 March 2014 - 06:20 AM, said:

KDR is no indicator of how good or bad you are at efficiently killing mechs.

All it shows is how good you are at swooping in on a torn up mech your teammates have been working over and stealing the killshot.

Hurray for you.

Damage is so much more important. Also more lucrative.

I regularly end up, alive, at the end of matches with the highest win score in my Fatlas or BM or Vic, while my SPD teammates have like 200 damage and more kills.

They didn't "help me out" - I was in the process of killing those mechs anyway, in the most painful and lucrative fashion available. AKA all of those dead mechs were one or two salvos away from me dropping them. They were dead anyway. Congrats - you swooped in and MG'd a critical. Oh THANK YOU, oh high KDR saviour! All Hail the wizard.

Also - if you die before the round ends but have torn the enemy team a new a-hole in damage to the tune of 700 or 800 damage, don't tell me that was stupid. High damage dealers helped the team assure the win, and actually made it possible for those "super snapshot deadeye killers" to get their high KDR and come on here and assume they did it all themselves.

That sounds a bit .... fustrated maybe even bitter.
A headshot delivers.... what ? 24 dmg and a dead enemy mech that can't do anything to your team mates.
An Atlas with shredded front armor will yield 300dmg+ and.... yea thats it... nothing for your team because hes still there posing a thread.
So its only logical NOT TO inflate your dmg but to get a lot dead mechs for the win.
Don't get me wrong I'm no friend of the high pinpoint dmg meta.
In my opinion weapons shouldn't be aimed directly at all in this game.
But fact is a dead oponent can't hurt you and your team.
Dealing out dmg is fine but pointles when it comms to winning.
I m blazing away at an Atlas with my Banshee and am unable to reach his torn backarmor but spider ran behind him and gets the kill ? Well yes its frustrating but damaging his armor and getting damaged is completely useless when I'm taking too much dmg and too much time to get to the kill.
The LRMers shoot of the headarmor of three enemys but I took the liberty to set them on an airstrike... booom three kills for me.
The Point is when you can do something in 20 secends your team mates failed to do for the last five minutes youve earned the kill and the points.

Long story short:
Its all about the balance young padawan
-- Lot of dmg (700+) no kills and few assists = bad shooting or LRM camper
-- 3+/ Kills/match but below 200dmg = killstealer ( or damn best headhunter around or maybe a light hunter you never know)

~ 500+ dmg in a heavy or medium with 2+ kills and 3 to 4 assists and you are a very serious buisiness man

Since I have a personal grudge against lights and am still thinking piloting lights is nothing more than easy mode and bugexploiting I realy don't know what to say to them.

Edited by The Basilisk, 21 March 2014 - 07:47 AM.

[Amsro]

Damage Per Match Average. Your ability to deal reliable damage every match gives your team the best chance to win.

This stat requires people to do their own math. But it's a far better indicator of mech efficiency for your playing style.

If you are under 200 damage per match and aren't in a capping/scouting/tag/narc light mech you are doing it wrong. IMO.

[MattEdge]

Good questions...

1)
I would have to say that I enjoy the K/D stat, but I don't focus on it. However, I wouldn't say that damage is the thing to focus on either... efficiency (damage per kill) is generally what I try and focus on personally for an overall stat...

On a match by match basis, I weigh a lot of different things in how I rate my performance. I start with the scoreboard, but I also factor in non-trackable things like if I think I did good spotting or scouting, and also if I felt that I consistently hit the boxes I was intending to hit, and whether or not I was able to exert some control over the enemy (distractions, turning a lance of mechs, etc.).

2)
I don't think we should have the kill-steal concept... the faster the enemy mech dies the better in my perspective. There are rare times when I come up on two lights circle fighting and if I see the enemy is red core and the friendly is pretty healthy, I'll hold fire. But if I was in that situation (which I wouldn't be 'cause circle fights are nasty), I would want the other mech to be dead, regardless of who shoots it. My opinion is shoot at the weakest mech in the weakest place for the quickest kill... then the team wins. And when the team wins, the rewards for everyone are larger.

3)
I actually think all the assist achievements are pretty cool and team oriented... not that it makes one try for an assist, but it does give a nice feeling to racking them up.

[DONTOR]

Ill throw in my 2 cents on kill stealing.
If Im in an obscure corner of the map deulling a spider in my comando and i have it legged, and I myself have taken 20% damage. If a streak Kintaro comes around the corner and fires a salvo at the spder and kills it, it IS stealing.
I did 95% of the work, wasnt in any immediate danger, and the the final blow is taken from me.
(damn those kintaro streak boat pilots!)

Edited by DONTOR, 21 March 2014 - 07:22 AM.