42 replies to this topic

[Artgathan]

Often when discussing match performance, players will often refer to their "average damage per match" as a benchmark statistic that can be used to gauge their performance. However, there are a few issues with this stat:

• It cannot account for non-combat contribution to a battle
  
• It can be easily artificially inflated
  
• High Damage output can be a result of poor marksmanship
  
• It is biased toward larger mechs

Therefore, I propose that players consider looking at average XP earnings per match. Admittedly there are problems with this as well:

• 2X XP Bonus can inflate values
  
• Active Premium time will inflate values

The first issue, of 2X XP, can be self-correcting. Presuming that a player has a large pool of matches played (and they play many matches without the 2X Bonus), the additional earnings from the 2X Bonus will be relatively small compared to the rest of the XP Earnings.

The second issue, of Active Premium Time, can be solved by adjusting XP by the amount of time you've had Premium Time active while playing that mech (IE: If I have Premium Time active 50% of the time I play a mech, I can subtract 25% of that mech's XP to see it's "real" XP earnings/match). However this relies on players knowing how much Premium Time they've had active on their mechs.

With all of that considered, using Average XP Earnings as an indicator of performance has the following advantages:

• It is not biased toward any weight class
  
• It is more difficult to inflate than damage
  
• XP Earnings require the application of multiple skills

What do you think?

[Mercules]

Artgathan, on 21 July 2014 - 06:56 AM, said:

• It is not biased toward any weight class

This is untrue. For the most part shooting a mech earns you XP. Capping a point earns you 0 XP so lights who are called upon to cap in Conquest often leave those matches with a win(which everyone on the team gets) and very little other XP. If the light pilot is off to the side at 600 meters holding a target with direct LoS but a friendly assault is 200m from the same mech locking him through the light pilot's lock, guess who gets the Spotting XP.

Lights earn less XP on average than heavier chassis for the same matches.

[Bhael Fire]

Honestly the only real measure of player skill depends on whether or not the player plays mostly solo or in groups.

If they play mostly solo, the most indicative stat for player skill is KDR. This stat shows how effective the player is at removing enemies from the field, regardless if they are kill stealing, dueling, or otherwise — a high KDR shows that the player is an effective killer, which contributes greatly to the overall team effort when in a PUG match. Which is all you can ask of a solo player.

If they mostly play in groups, it's W/L ratio. A high win to loss ratio shows that the player works well with their group and has become an effective team player...which is all you can ask of a grouped player.

That said, a high value in either KDR or W/L speaks volumes about player skill, regardless if solo or grouped. Damage and XP are also supporting stats, but speak less of skill than KDR and W/L do.

[Mercules]

Bhael Fire, on 21 July 2014 - 08:09 AM, said:

If they play mostly solo, the most indicative stat for player skill is KDR.

Unless they play lights. My Medium/Heavy KDR is much higher than my light and I play lights more frequently. You just don't have as much leeway in Light before the "Death" part of the KDR.

[Roadkill]

Bhael Fire, on 21 July 2014 - 08:09 AM, said:

If they play mostly solo, the most indicative stat for player skill is KDR.

Huh? No it isn't. KDR does not reflect organizational skill, scouting skill, spotting skill, or a myriad of other valuable combat skills.

KDR really only reflects twitch skills, and even then it is subject to play style. Brawlers by their very nature are more likely to die during combat than snipers, so they're virtually guaranteed to have lower KDR. Yet without the brawlers, the snipers would get rapidly overrun.

KDR is a very poor reflection of actual overall skill in MWO.

[Bhael Fire]

Roadkill, on 21 July 2014 - 09:01 AM, said:

KDR is a very poor reflection of actual overall skill in MWO.

I completely disagree. Killing the enemy is the whole point of deathmatch-style games — which is pretty much the only game mode we currently have in MWO. Sure, you can help other players get kills or capture bases, but ultimately it's about the kills and how fast you can remove an enemy threat from the battlefield.

Scouting, spotting, NARCing, fire support, etc. are great ways to help out, but ultimately you'll be doing your team a greater service in a PUG match by simply killing the enemy before they can kill one of your teammates.

I'm not discounting other skills that are useful in winning, because ultimately ALL of these skills define a player's overall skill. I'm just saying that the most skilled solo players have high KDRs. However, this isn't always the case with players that play in groups more frequently where those other team-focused skills are used more frequently to achieve victory.

[Mercules]

Bhael Fire, on 21 July 2014 - 09:40 AM, said:

Scouting, spotting, NARCing, fire support, etc. are great ways to help out, but ultimately you'll be doing your team a greater service in a PUG match by simply killing the enemy before they can kill one of your teammates.

Which doesn't happen if the enemy hits your flank unexpectedly because no one was a able or willing to "scout". Which doesn't happen if your team gets suppressed by pop-tarting because no one distracted them from a flank or spotted them so that LRMs could fall in among them long enough for the brawlers to get in close.

Saying, "Killing your enemy before they can kill you." is simple, doing it requires all the OTHER skills, usually being utilized by someone not focused on KDR and sacrificing their "kills" to offer information or distraction that lets other pilots get the kills.


Me rushing in and unloading on an Atlas' back then getting him to (stupidly) spin to try and spot and fire on me and exposing himself to my teammates coming around the building doesn't net me a single kill, but swings the outcome of the battle.

[Deathlike]

Average Damage doesn't properly quantify much... you could have a win where you do under 200 damage in a Light, or a loss when you put out 300+ points of damage in the same mech.

XP isn't a great indicator either, and while double XP will influence your overall average over time, you need to consider what that value is before the doubling... even then, Savior Kills tend to put that number out of whack (which is good and bad - good because you are trying to finish of crit mechs as a group, bad because it pads stats like noone's business).

Consistency through winning and losing is a lot more telling than anything else.

[Bhael Fire]

Mercules, on 21 July 2014 - 09:59 AM, said:

Saying, "Killing your enemy before they can kill you." is simple, doing it requires all the OTHER skills, usually being utilized by someone not focused on KDR and sacrificing their "kills" to offer information or distraction that lets other pilots get the kills.

That's why I said those other skills are important and paint a better picture of overall player skill. Ultimately, a good player will have all of these skills honed which will lead them to having a higher KDR in solo matches. The less the solo player has to rely on his other teammates to be a self-sufficient killer, the better they will do in solo matches.

Helping your teammates get kills is great, but ultimately a consistent record of getting your own kills is very indicative of self-sufficiency, piloting, gunnery, and combat awareness.

[Mercules]

Bhael Fire, on 21 July 2014 - 10:18 AM, said:

That's why I said those other skills are important and paint a better picture of overall player skill. Ultimately, a good player will have all of these skills honed which will lead them to having a higher KDR in solo matches. The less the solo player has to rely on his other teammates to be a self-sufficient killer, the better they will do in solo matches.

Helping your teammates get kills is great, but ultimately a consistent record of getting your own kills is very indicative of self-sufficiency, piloting, gunnery, and combat awareness.

On which I call BS. I am running lights. Someone literally has to these days. The damage I can put out in 1 minutes isn't even half what a Direwolf can output. I can either bring 1-2 long ranges weapon and opportunity snipe, or a lot of closer ranged weapons but then have to wait until a mech is out of position to attack. I don't have the armor to trade shots so my damage and thus my chances for the kill shots will ALWAYS be less than when I am in a Medium/Heavy/Assault.
W/L is a nice positive number in my lights, my K/D typically is not even though in some matches my Kitfox might wrack up 3 kills. Not enough armor to get a "Kill" in before I die every time I drop. Now... I could be a total selfish jerk and just sit back and then steal the kills and I could have an amazing K/D but doing so will actively hurt my team's chances of winning and thus make me a "bad player".

[Roadkill]

Bhael Fire, on 21 July 2014 - 09:40 AM, said:

I'm not discounting other skills that are useful in winning, because ultimately ALL of these skills define a player's overall skill. I'm just saying that the most skilled solo players have high KDRs.

No, that isn't what you said:

Bhael Fire, on 21 July 2014 - 08:09 AM, said:

If they play mostly solo, the most indicative stat for player skill is KDR.

Skilled solo players having a high KDR is not at all the same as KDR being the most indicative stat for solo player skill.

Furthermore, even your new stance isn't necessarily true as has already been explained. Scouts aren't going to have high KDR. Brawlers are going to have lower KDR than snipers. And a good commander is far more valuable to a team than one person who can shoot straight.

KDR is not a good indicator of overall player skill regardless of queue. It's an indicator of just one type of skill for certain kinds of players, but even there it can be manipulated and so really isn't a good measure of actual skill. Use your teammates as meat shields and save your massive alpha to steal the kill and you'll have a good KDR, but it won't mean you're skilled at MWO.

[Bhael Fire]

Mercules, on 21 July 2014 - 10:34 AM, said:

Now... I could be a total selfish jerk and just sit back and then steal the kills and I could have an amazing K/D but doing so will actively hurt my team's chances of winning and thus make me a "bad player".

No, that would make you a bad team player. You'd still be an efficient killer, and therefore a skilled player.

Roadkill, on 21 July 2014 - 10:47 AM, said:

Skilled solo players having a high KDR is not at all the same as KDR being the most indicative stat for solo player skill.

Yes it is, maybe not in syntax...but it means the same thing. If the player has all those other skills, they will tend to have a higher KDR than a player without those skills. Just like those skills help a player in a group achieve a higher W/L ratio.

Therefore, the most indicative stats to look at when determining player skill are KDR and W/L.

[Roadkill]

Bhael Fire, on 21 July 2014 - 10:56 AM, said:

maybe not in syntax...but it means the same thing.

No, it doesn't. Causality is reversed in the two statements.

I don't really have a problem with the statement that "If the player has all those other skills, they will tend to have a higher KDR than a player without those skills." Causality is in the correct direction in that statement.

I have a big problem saying that a high KDR is the best indicator of high player skill. It is one possible indicator of one type of skill, but it is sorely lacking as a general indicator of overall player skill regardless of queue. You're implying that getting kills is the only (or best) way to help your team win, which just isn't true. In fact the best way to help your team win isn't really measureable in MWO - and that's being a good commander.

The saying is that Teamwork is OP, right? Well if you're good at getting your team to work together, especially in the PUG/solo queue, you're going to be far more valuable to your team than someone who just has a high KDR. Herding pugs is by far the most valuable skill you can possibly have in the solo queue, and it's not captured by KDR.

Bhael Fire, on 21 July 2014 - 10:56 AM, said:

Therefore, the most indicative stats to look at when determining player skill are KDR and W/L.

Another new position?

[Angel of Annihilation]

Damage is actually very useful stat for gauging performance though everything you say about it is true.

What damage done does tell you it two very important things.

1 - That a player is actively contributing to the fight.

2 - How long he survives in a fight.

The straight truth is you can't get high damage scores unless you are actively shooting the enemy and shooting them often. A camper isn't typically going to have high damage for example. Nether will a vulture (player with high kill count but low damage associated with them only firing at badly damaged mechs).

Also if you suck, you generally get killed early which of course leads to a low damage count. Basically you have to be shooting often and long to have a high damage score which means you must survive long enough to do that damage.

Also any mech can have a high damage count. I ran my Firestarter a couple times last night and the very first match I had 2 kills and almost 700 damage. Got like a 92 match score out of that match and even managed 10 spots showing that I was actually doing my job as a scout. However you do have to adjust the scores a bit.

To judge pilot skill by damage I think the following is a good gauge:

Lights: 400+ damage
Mediums: 500+ damage
Heavies and Assaults: 600+ damage

Also one more thing to add. Damage is damage. Each point of damage inflicted upon the enemy is 1 point closer to victory so even an LRM boat sitting in the backfield contributes to a win.

[Joseph Mallan]

Mercules, on 21 July 2014 - 08:02 AM, said:

This is untrue. For the most part shooting a mech earns you XP. Capping a point earns you 0 XP so lights who are called upon to cap in Conquest often leave those matches with a win(which everyone on the team gets) and very little other XP. If the light pilot is off to the side at 600 meters holding a target with direct LoS but a friendly assault is 200m from the same mech locking him through the light pilot's lock, guess who gets the Spotting XP.

Lights earn less XP on average than heavier chassis for the same matches.

As rarely as I get to Cap in an Atlas I had no idea there was no XP for Capping. It seems rather silly that there is no XP reward for performing the main objective for the conquest mission.

[Joseph Mallan]

Bhael Fire, on 21 July 2014 - 10:18 AM, said:

That's why I said those other skills are important and paint a better picture of overall player skill. Ultimately, a good player will have all of these skills honed which will lead them to having a higher KDR in solo matches. The less the solo player has to rely on his other teammates to be a self-sufficient killer, the better they will do in solo matches.

Helping your teammates get kills is great, but ultimately a consistent record of getting your own kills is very indicative of self-sufficiency, piloting, gunnery, and combat awareness.

And what if you are a better support player than head on? A good team will have its brawlers, support, snipers and recon... Teamwork wins a match over individual effort.

[Bhael Fire]

Roadkill, on 21 July 2014 - 11:17 AM, said:

Herding pugs is by far the most valuable skill you can possibly have in the solo queue, and it's not captured by KDR.

If you want to win, yes. I suppose it comes down to what's important to you. When I'm playing solo I could care less about winning or losing as long as I'm alive at the end of the match and I have at least 1 or 2 kills.

If that means using teamwork then that's the route I will take to achieve that goal. However, if I'm with a bunch of derps that have the hardest time following the simplest of plans, I start looking out for myself to ensure I take out as many of the enemies that I can before they get me.

Roadkill, on 21 July 2014 - 11:17 AM, said:

Another new position?

It's the same position as my previous statements; those two stats are the best indicators of player skill because they encompass a set of other minor skills that are too varied to list individually or too difficult to gauge as a tangible stat within the current stat system.

Joseph Mallan, on 21 July 2014 - 11:28 AM, said:

And what if you are a better support player than head on?

Then you are a good team player!

[Mercules]

Bhael Fire, on 21 July 2014 - 10:56 AM, said:

No, that would make you a bad team player. You'd still be an efficient killer, and therefore a skilled player.

It's not a solo game, ergo makes you BAD. Means you actually suck worse than the guy who did half your damage and got no kills but held up 4 mechs flanking the team so that the team could win.

Bhael Fire, on 21 July 2014 - 11:37 AM, said:

If you want to win, yes. I suppose it comes down to what's important to you. When I'm playing solo I could care less about winning or losing as long as I'm alive at the end of the match and I have at least 1 or 2 kills.

Which basically screws everyone else and makes you one of those bad players I don't want on my TEAM in my TEAM game.

[Bhael Fire]

Mercules, on 21 July 2014 - 12:12 PM, said:

It's not a solo game, ergo makes you BAD. Means you actually suck worse than the guy who did half your damage and got no kills but held up 4 mechs flanking the team so that the team could win.

What is better; "holding up" 4 mechs or killing 4 mechs so your team could win? In the solo queue, killing those 4 mechs is much more valuable.

Again, I need to reiterate; teamwork is a good thing. Nobody is arguing that. But when it comes to measuring player skill on what makes for a valuable player in a PUG match, the players with the higher KDRs are going to carry the rest of the team.

[Screech]

To truly know what your skills are at only look at stats from loses, that will really give you an idea of what you are about.