15 replies to this topic

[Tykelau]

So every now and then I have a match where I do 900 or more damage. I usually sit back and think about how awesome I did.

Then I start to think about it. 900 damage aimed well would be enough to core 9 atlases, 12 cataphracts, mediums suck so who cares (I almost kid), and 25 ravens/jenners. Not even beginning to touch headshots.

There isn't a way for stats to track this though, even though I think that this would be a much better metric of skill than any we have so far.

This is the problem, ELO is only as good as the skill metrics, and all of our skill metrics suck so far. I think (rarely) that if we could get some smart fellers to think about this it could make a big difference in the MWO experience.

So, good idea?

[multiplesanta34]

Are you suggesting a metric that reflects the number of kills shots vs. non kill shots to better reflect skill? Not every non kill shot is a bad shot though, and doing alot of damage is always a good thing because if you weren't able to kill your target you at least weakened him for your teammates. I don't think you could do this without encouraging people to hold their fire unless a CT were visible, and that would run counter to team play. Unless I misunderstood you.

[OneEyed Jack]

Nobody hits on the exact location every time. That wouldn't be aimed well, it would be aimed perfectly. If you're going to go there, might as well say that you should have soloed all 8 of them for less than 320 damage (8 head kills, rounded up to 40 each to allow for over-kill from larger weapons).

The most important question is "Did your team win." After that, if you want to measure your contribution you can ask yourself:

• Was it won on kills?
  
• How many of those kills were yours?
  
• How many assists?
  
• Were you in the thick or off playing Rambo?
  
• We you fighting with your team and helping keep heat off or safely sniping from a distance while your team died for your kills?

None of which truly define anything, but I think they're good points to consider for self-evaluation in this particular case.

Actual skill can never be measured by stats. Stats can only measure a level of success, from which skill might be inferred.

[Kiiyor]

COUNTERPOINT.

I've noticed that when you are facing tougher opponents, it is exponentially tougher to make them stand still enough to put all your damage in one place.

Even with high alpha builds, skilled players know how to spread the love all over themselves, rather than letting you get straight to their gooey cores.

Those pesky opponents. If only they would stand still.

EDIT: Correction of monkey like typing skills.

Edited by Kiiyor, 18 April 2013 - 08:57 PM.

[Nonsense]

This topic has been around many times and it always lacks real critical thinking. There are PLENTY of ways you can end up with high damage and still be putting shots where you want them.

1. As your opponents get better they'll torso twist more and/or use cover, sometimes presenting you with only non-CT shots.

2. Sometimes it's better to disarm a mech because of the way it's already damaged by teammates and the weapon/engine configuration.

3. If you've only got a flank shot, should you simply not fire just because it's not going to hit the CT? Similarly, should you fire at the CT of a fresh atlas when your teammate is on its flank? What if you both take out the side torso you can both hit?

There are a ton more examples I can think of. High damage simply means you're firing a lot and you're positioning yourself well so that you can be hitting enemies a lot. If you're also consistently getting kills and assists, you're definitely contributing more to the team on average than someone with lower damage (unless of course you're a pre-splash-damage-bug-nerf-thing LRM boat).

Low damage is not better. Efficiency doesn't matter in the way you're talking about. Yes, not wasting heat/ammo/opportunities does matter, but as a metric, accuracy/efficiency is not a good measurement of skill or team contribution IMO.

[Tykelau]

multiplesanta34, on 18 April 2013 - 08:50 PM, said:

Are you suggesting a metric that reflects the number of kills shots vs. non kill shots to better reflect skill?

No, I'm still hashing out an idea, but the general concept I had was along the lines of tracking, per mech, how many components you hit, and for how much damage. Maybe even the components state before also, IE, did the successfully identify and hit the weakened component? Again, tracking frequency, over time should show a difference between the greats and the others.

There are always dangers of stat whores, but I can't help wonder if the idea might have some merit. Maybe don't show the info to the players?

OneEyed Jack, on 18 April 2013 - 08:54 PM, said:

Nobody hits on the exact location every time.

The most important question is "Did your team win." After that, if you want to measure your contribution you can ask yourself:

• Was it won on kills?
  
• How many of those kills were yours?
  
• How many assists?
  
• Were you in the thick or off playing Rambo?
  
• We you fighting with your team and helping keep heat off or safely sniping from a distance while your team died for your kills?

None of which truly define anything, but I think they're good points to consider for self-evaluation in this particular case.

Actual skill can never be measured by stats. Stats can only measure a level of success, from which skill might be inferred.

You went black and white, I'm not saying hit the same location every time. I am saying that a better player would have a higher frequency of hitting a specific component however. Not "always", but more consistent. The other points you make are great, but while I don't have any ideas on how to track them, I still think that this idea has merit. However it is possible that it would require more bandwidth and/or more crunching client side and is not a realistic idea with todays tech.

Kiiyor, on 18 April 2013 - 08:57 PM, said:

COUNTERPOINT.

I've noticed that when you are facing tougher opponents, it is exponentially tougher to make them stand still enough to put all your damage in one place.

Agreed, but again I'm not talking about just one match, I'm talking about trends that would show after many matches.

[Cik]

damage is just a function of time on target. high damage does not mean you are a bad shot, it simply means the enemy is proficient at using his armor correctly. better to shoot and hit, then to hold fire for only perfect CT shots.

[Tykelau]

I agree with your points on a case by case basis, however the more matches one plays the more trends appear. The idea that a skilled player faces skilled opponents is a problem with this idea only if this metric was the only one used. I am not saying that at all.

In fact all of the counterpoints seem to assume that I am saying either or when it comes to metrics. I am suggesting an additional metric, possible a weighted one. I'm not saying low damage is good, high damage is bad. I'm just saying tracking where you hit on a mech would be an aid in setting up ELO.

[Royalewithcheese]

IMO it's less that you want to shoot for low or high damage and more that damage is a function of your build. The best way to tell how much you contributed is to pay attention to how the match went. Did you hit nothing but CT on one enemy? Did you get a headshot or two? Etc.

[Levi Porphyrogenitus]

If you want to track effective damage, then Component Destruction results might be the thing to track (combined with Kills, probably). The idea then would be that your shots mattered, since you stripped something of value from the target rather than just lowered his damage absorption by a bit.

I'm not actually convinced that this is a good metric, myself. I'd rather see some kind of composite score that evaluates your win/loss, your kills, your assists, your deaths, your damage, your components destroyed, your objective completions (capture assists, etc.), and so on and so forth. PGI'd have to come up with some kind of equation and it'd likely wind up being complicated, but once it's in place it should be easy for their servers to do the math behind the scenes and figure out Elo ratings.

[Kiiyor]

Tykelau, on 18 April 2013 - 09:17 PM, said:

No, I'm still hashing out an idea, but the general concept I had was along the lines of tracking, per mech, how many components you hit, and for how much damage. Maybe even the components state before also, IE, did the successfully identify and hit the weakened component? Again, tracking frequency, over time should show a difference between the greats and the others.

There are always dangers of stat whores, but I can't help wonder if the idea might have some merit. Maybe don't show the info to the players?



You went black and white, I'm not saying hit the same location every time. I am saying that a better player would have a higher frequency of hitting a specific component however. Not "always", but more consistent. The other points you make are great, but while I don't have any ideas on how to track them, I still think that this idea has merit. However it is possible that it would require more bandwidth and/or more crunching client side and is not a realistic idea with todays tech.



Agreed, but again I'm not talking about just one match, I'm talking about trends that would show after many matches.

Same. I've been getting the feeling lately that i'm climbing the corporate ladder in the ELO foodchain. My opponents seem to be getting progressively tougher. In the social foodchain of yore, I believe I was something resembling plankton.

This is entirely anecdotal.... but i've noticed that my most extreme killstreaks seem to coincide with the matchmaker throwing it's hands up in frustration when trying to find a match. I've read that if the matchmaker starts to struggle to find you a match, it dumps you into whatever it can find. I cherish these moments.

"Hmm, there's an Atlas shooting at me a... wait.. he's standing still! GO FOR THE EYES!"

[Ralgas]

this again *sigh*

how much of that 900 dmg was ammo/gauss/engines going boom (random and inflated numbers depending on target mech load out)

although your thinking is correct it's still possible (unlikey)to get all those kills via 40 odd dmg per mech (headshots)

W/L ratio isn't perfect, but other than reassigning xp to everything equally (and giving lights more incentive to play capwarrior than they do now) and using match average xp as your modifier W/L is the best naturally balancing act of a bad bunch we have.

Edited by Ralgas, 18 April 2013 - 09:42 PM.

[ProtoformX]

Ralgas, on 18 April 2013 - 09:41 PM, said:

how much of that 900 dmg was ammo/gauss/engines going boom (random and inflated numbers depending on target mech load out)

If internal explosions counted towards your damage total, you'd be dealing some hundreds of damage per ammo explosion. I find it very very unlikely that they count towards your damage total at all... Please correct me if I'm wrong.

FYI 900 damage is nothing crazy when 7/8 mechs on the other team are assaults. Obviously it's good, but nothing to swoon over. You've gotta dig through a lot of armor to core 7 assault mechs. When you do insane damage, check the other team's weight and behold the vast amount of armor.

[MustrumRidcully]

Elo is not based on accuracy or damage done. It's just based on comparing your Elo rating to that of your enemy and adjusting your rating based on a win or loss against that rating. And I think that's the best method, really. It doesn' t matter if you're some kind of miracle worker that can squeeze 2,000 damage out of your mech and have poor aim, or if you're a deadly sniper that kills everyone with headshots. If you're a poor player, it's simply unlikely that you could keep winning in teams with good players against teams with good players, because your ineptitude would drag you down, and likeweise, if you're a good player, you're likely able to turn a battle in your favor.

[Tykelau]

Interesting, this is the last match that motivated my OP. Enemy team was 505 tons vs our 535, so not a fair matchup either.

MustrumRidcully, on 18 April 2013 - 10:18 PM, said:

Elo is not based on accuracy or damage done. It's just based on comparing your Elo rating to that of your enemy and adjusting your rating based on a win or loss against that rating. And I think that's the best method, really. It doesn't matter if you're some kind of miracle worker that can squeeze 2,000 damage out of your mech and have poor aim, or if you're a deadly sniper that kills everyone with headshots. If you're a poor player, it's simply unlikely that you could keep winning in teams with good players against teams with good players, because your ineptitude would drag you down, and likewise, if you're a good player, you're likely able to turn a battle in your favor.

Excellent point. My problem is that I believe matchmaking is the biggest deterrent to new players and even old players. An 8-0 stompfest is not fun for me. Yes I'm an immature neckbeard geezer sometimes, but I'd imagine that most people find getting stomped sucks.

What I would like to see but have no idea how to make happen, is a more accurate, faster method of determining the values for ELO and sought out help/opinions/ideas from the forums. IMO most topics that keep coming up in the forums come up for a reason.

Edited by Tykelau, 18 April 2013 - 10:30 PM.

[Zolaz]

Tykelau, on 18 April 2013 - 08:07 PM, said:

Then I start to think about it. 900 damage aimed well would be enough to core 9 atlases

Not any Atlases I pilot. 900 damage will get you through 3.75 Atlases outer armor and then you need to get through internals. Not counting the fact that some of that is going to splash over to my RT, LT or Head.