342 replies to this topic

[Ultimax]

REDACTED (accidental double post)

Edited by Ultimatum X, 29 April 2014 - 03:56 PM.

[Roland]

MischiefSC, on 29 April 2014 - 03:47 PM, said:

KDR is situational and subjective. I'd give more credence to damage/shot, which is a good balance for the differences between improved accuracy created by spread vs FLD damage from the hit/miss dynamic of the AC10 ballistic.

The question is, how much difference? Also what change, if any, to win/loss and KDR for the chassis.

You should include the KDR in the stats you track, because it is a primary yardstick by which to measure effectiveness. The point of the game is to kill mechs, and K/D is the most direct measurement of your killing power vs. your survivability. It is certainly not the only yardstick, but it is one which is important enough that it cannot be ignored.

Simply tracking damage per shot won't really give you particularly valuable data in this comparison, because even if you were to land the entirety of an LBX shot on a mech and do damage with all of the pellets, they would likely still be spread over multiple locations, which would tend to be inferior to an AC10 landing all of its damage on a single location.

This is why K/D will be important, because it will more directly measure killing power.. In many cases, it'll benefit the AC10, which will be able to focus its fire and likely kill targets faster.. but this could potentially be tempered by the LBX's ability to do large amounts of damage after armor is exposed, and possibly even get more kill shots.

Ultimately, in this type of analysis, the more information you gather, the better... Its why video recordings of matches is so valuable, since you can analyze more nuanced aspects of engagements and see things that may not be immediately apparent in the raw numerical data.

[Bishop Steiner]

Ultimatum X, on 29 April 2014 - 03:50 PM, said:

3 Salvos of the 2x LB 10-X =

20 x 3 = 60 damage
2.5s CD = 7.5s

+ 4x MG @ 1 DPS each x 7.5s = 30 damage


So if you are firing 3 volleys of 2x LB10s and 4x MGs all at the same time that's around 90 damage you should be doing in 7.5s.






Bishop, to be fair to Roland in this thread he's not saying that LB 10X is bad, or that people saying they are good are bad, or any of that.

He's just emphasizing that methodical testing demonstrates more than anecdotes.


Now with the private, especially premium, queues there are more options for dedicated testing.


I might do some testing of my own, maybe this week or this weekend, in testing grounds to see how many shots it takes of X, Y or Z weapon to down a target dummy for example.

It's actually not Roland so much this thread, but a few others who are being highly dismissive. But Roland and I have locked horns a few times, and while not a big deal, will do so here, too, lol. He has some very valid points, and then sometimes, I believe sometimes gets locked into a view that is not as encompassing as he thinks (as cna I, and most players).

The only time I really have an issue is when he, or Fup, or MWHighlander take a condescending and dismissive view of something because "that's not how 12 mans do it" (not always the case, but often comes across that way)..... and can sometimes seem oblivious to the fact that PUG matches and 12 mans are two inherently different environs, because of the difference in communication and coordination and that things that might not seem "ideal" for comp play can be absolutely lethal in the chaos of PUG play. and that TBH, not everyone really cares that much about 12 man.

Effective 12 man will always be about what is highest reward/lowest risk with shallowest skill curve. Not always the highest skill cap, though. Because that is the fastest most efficient way to roll. That doesn't always leave much room to really effectively give weapons that don't fit the "meta paradigm" a fair shake out.

[FupDup]

Bishop Steiner, on 29 April 2014 - 03:58 PM, said:

---
The only time I really have an issue is when he, or Fup, or MWHighlander take a condescending and dismissive view of something because "that's not how 12 mans do it" (not always the case, but often comes across that way)..... ---

And for the record, I've never even played 12 mans, and I generally drop without a premade. I just so happen to be clicking +1 on most of Roland's posts because he has good points much of the time.

[Bishop Steiner]

Roland, on 29 April 2014 - 03:56 PM, said:

You should include the KDR in the stats you track, because it is a primary yardstick by which to measure effectiveness. The point of the game is to kill mechs, and K/D is the most direct measurement of your killing power vs. your survivability. It is certainly not the only yardstick, but it is one which is important enough that it cannot be ignored.

Simply tracking damage per shot won't really give you particularly valuable data in this comparison, because even if you were to land the entirety of an LBX shot on a mech and do damage with all of the pellets, they would likely still be spread over multiple locations, which would tend to be inferior to an AC10 landing all of its damage on a single location.

This is why K/D will be important, because it will more directly measure killing power.. In many cases, it'll benefit the AC10, which will be able to focus its fire and likely kill targets faster.. but this could potentially be tempered by the LBX's ability to do large amounts of damage after armor is exposed, and possibly even get more kill shots.

Ultimately, in this type of analysis, the more information you gather, the better... Its why video recordings of matches is so valuable, since you can analyze more nuanced aspects of engagements and see things that may not be immediately apparent in the raw numerical data.

I agree, tracking the KDr is important. The mor einfo, period, the better. But I don't think the Ratio actually matters as much as the actual Kills and Deaths, total. I was running a 17/1 KDr. The next match, I got 3 Kills.... and died. So my KDr went down to 10/1. And its a very biased curve downhill. I get 4 kills and die next match.... my KDr goes to 8/1....despite the fact I actually just added 4 more kills to the 1 death. So I would say it important to track total kills vs total deaths, and discover the actual average kills per match, vs avg deaths.

[Roland]

Bishop Steiner, on 29 April 2014 - 03:50 PM, said:

the question I pose (as I am indeed recording my progress, since I was able to use the stat wipe as a clean break from the mechs previous stats) is are you really as open as you think, or do you let the LB-X not being what you WANT it to be, color your thinking (as I know from previous posts, that it not working like in TT or at least in the vicious MW4 shotgun manner, seems to be a big beef with you. If you recall, I too have spent months working on and proposing ways to "fix" it).

I don't really care about TT battletech.

I used the weapon extensively in MW4, because it have a very clear niche.. that is, it was weak at range, but was extremely powerful up close. It offered significant rewards to counter its limitations.

I didn't originally start out thinking the LBX was bad. But in all of my testing, it's always been inferior to other options. Perhaps that's changed, although I'm not sure what change could have resulted in it becoming better, since i'm not aware of any changes to that specific weapon since I last tested it.

The reasons I've tested it numerous times is because I want it to be good, but it just hasn't been thus far.

Quote

At no point should the LBX be BETTER than the ac10, as a total combat system. What the testing needs to ask, is it a legitimate viable alternative, that might fulfil a different role from the ac10.

That's generally why I've always proposed upping its damage per pellet. The result would be a weapon which was weaker than the AC10 at range, while being stronger up close. Thus, both weapons would have utility.

Bishop Steiner, on 29 April 2014 - 03:50 PM, said:

We will see. But until then, I think the Peanut Gallery (and they know who they are) would also be advised to put up or shut up, scientifically, and stop attacking those who are willing to think outside the precious Meta-Box.

I don't believe that anyone in this thread has attacked anyone, other than perhaps your own somewhat antagonistic posts.

[Bishop Steiner]

FupDup, on 29 April 2014 - 04:01 PM, said:

And for the record, I've never even played 12 mans, and I generally drop without a premade. I just so happen to be clicking +1 on most of Roland's posts because he has good points much of the time.

Might be so Fup, but you can get on a very "Pro" tangent at times. So, sorry for calling you an evil premade, but don't really retract where we can vehemently disagree as to what constitutes effective in MWO.

[Vlad Ward]

Roland, on 29 April 2014 - 03:31 PM, said:

All the talk in the world won't replace well structured empirical testing.

Record stats, including k/d ratio with lbx, then do the same with a comparable build using other weapons, and compare them. Ideally, keep video records for analysis.

Everything else is just smoke.

stahp. stahp that. Empirical testing? Playing a dynamic, 24-player first person shooter a few times with different weapons is empirical testing?

In-game weapon performance is a wash statistic. It doesn't mean jack. Too many variables, impossible to control for, all that fun stuff.

That's why anyone who cares evaluates weapon potential performance. That's easy to calculate, and can be done without so much as installing the game. The behavior of these weapons are all well documented and quantified. We know exactly how much damage each pellet does, both to armor and internal structure. We know (or can easily derive) the solid angle of the projectile cone. We can find simple, concise numbers to demonstrate exactly how much damage a weapon is capable of doing.

Now, the underlying assumption is that even though players will probably never reach that maximum potential damage, they will fall short by a similar amount on each weapon they use. By that I mean a player reaching 70% of damage potential of the AC/10 will probably also reach 70% of the damage potential for the LB-10X or AC/5.

Unfortunately, that assumption starts to degrade when player behavior becomes more erratic. For example, when player accuracy drops off, there is a significant increase in in-game performance with cone of fire weapons like SRMs and the LB-10x. Partial credit and all that.

Of course, it's a lot of work to go around saying "If you have carpal tunnel, the LB-10X is better than the AC/10, but if you have 20/10 vision you may prefer the Gauss Rifle and PPC. Oh, but if you get motion sickness, stay away from short range weapons." It's easier, and more generally accurate to categorize weapons based on their potential damage and let players make up the gap.

Edited by Vlad Ward, 29 April 2014 - 04:05 PM.

[Roland]

Bishop Steiner, on 29 April 2014 - 04:01 PM, said:

I agree, tracking the KDr is important. The mor einfo, period, the better. But I don't think the Ratio actually matters as much as the actual Kills and Deaths, total. I was running a 17/1 KDr. The next match, I got 3 Kills.... and died. So my KDr went down to 10/1. And its a very biased curve downhill. I get 4 kills and die next match.... my KDr goes to 8/1....despite the fact I actually just added 4 more kills to the 1 death. So I would say it important to track total kills vs total deaths, and discover the actual average kills per match, vs avg deaths.

Well, that's really only the case for very small datasets.

Over time, your K/D will stabilize.

[Bishop Steiner]

Roland, on 29 April 2014 - 04:03 PM, said:

I don't really care about TT battletech.

I used the weapon extensively in MW4, because it have a very clear niche.. that is, it was weak at range, but was extremely powerful up close. It offered significant rewards to counter its limitations.

I didn't originally start out thinking the LBX was bad. But in all of my testing, it's always been inferior to other options. Perhaps that's changed, although I'm not sure what change could have resulted in it becoming better, since i'm not aware of any changes to that specific weapon since I last tested it.

The reasons I've tested it numerous times is because I want it to be good, but it just hasn't been thus far.


That's generally why I've always proposed upping its damage per pellet. The result would be a weapon which was weaker than the AC10 at range, while being stronger up close. Thus, both weapons would have utility.


I don't believe that anyone in this thread has attacked anyone, other than perhaps your own somewhat antagonistic posts.

you don't have to name and shame to attack people Roland. A person takes a snide and dismissive tone toward someone, or group of someones daring to question the HiveMind, and it is the same thing. So if I am antagonistic, it is in response to (perceived) antagonistic posts.

[Roland]

Vlad Ward, on 29 April 2014 - 04:04 PM, said:

stahp. stahp that. Empirical testing? Playing a dynamic, 24-player first person shooter a few times with different weapons is empirical testing?

No, that wouldn't really be empirical testing.

[Bishop Steiner]

Roland, on 29 April 2014 - 04:04 PM, said:

Well, that's really only the case for very small datasets.

Over time, your K/D will stabilize.

Stabilize is still not as specific as discovering how effective it is per match. I have 2 mechs with a 4/1 KDr, for instance. One just happens to be I don't die a whole lot, but I also don't get a lot of kills (lots of damage though, it's an LRM boat Battlemaster), and I actually have LESS kills than wins in it. (thus, effective at support, but not my all that fun.... cuz everyone else gets the kills, darn it!). The other one I do good damage and get a LOT more kills, but die a lot more often, because as a brawler I have to be a lot more aggressive.

Both have "stabilized" stats, but tell two very different stories. Hence while KDr is useful, i don't think it near as useful as finding out what the mech's actual per match average is. I know I have mechs that I think, "man I get 700-800 damage matches all the time..... but when you average it out with those matches you catch a full flaming poptart to the face 10 seconds into the match, it actually only averages 392 a match, which I would contest as solid performance, as an average, for anyone, but it is a very accurate tell of it's overall performance, over peaks and valleys, over KDr)

[Vlad Ward]

Roland, on 29 April 2014 - 04:05 PM, said:

No, that wouldn't really be empirical testing.

You could play and Youtube 10,000 games and it would still represent very little useful data.

Nutshell version: How well you use a particular weapon (in a team with 11 other people, against 12 other people) is a poor indicator of the objective value of the weapon.

Edited by Vlad Ward, 29 April 2014 - 04:12 PM.

[SirLANsalot]

The LBX as a good weapon is nothing new at all. Its been a secret weapon for a lot of high skilled players when they roll a brawler mech. Its the scrubs and the one who THINK there good, that have no idea how well that gun can preform in pairs (and must be pairs). Solo the gun is a little lack luster, and the AC10 is the better pick for a solo (non AC20 slotted) ballistic or the Gauss if you have the spare tonnage.


Two LBX on an Atlas, or a Protector (yes mini atlas) is just devastation up close.

[BigFatGator]

I think the relevant comparison isn't AC10 vs LBX 10, but 2AC10 vs. 2LBX10 + 2MG+ 1 ton MG ammo (wt. for wt. comparison). If you have the ballistic slots for it, I think the LB10X+MG is better than AC10, IF you are otherwise ready to brawl at close range.

In a Jaeger S with STD250 engine and 4ML, over 50-100 matches with each in the last 3 months, the 2LB10X+2MG did better (2.75 K/D) vs. either 3AC5 (2.2) or 2AC10 (1.5) builds. These stats also probably say I can't aim for crap, and DPS weapons are more my thing. Also the fact that the 4ML with lack of extra heatsinks needs cool-running weapons paired with it.

[Hillslam]

NNNNNNNNNNNNNNOOOOOOOOOOOO!!!

MAKE THIS THREAD GO AWAY!

HUSH ALL OF YOU!

NO DISCUSSING THE LB-10X

NOTHING TO SEE HERE!

LEAVE WELL ENOUGH ALONE!

Move Along. Move along.

[Roland]

Vlad Ward, on 29 April 2014 - 04:11 PM, said:

You could play and Youtube 10,000 games and it would still represent very little useful data.

Nutshell version: How well you use a particular weapon (in a team with 11 other people, against 12 other people) is a poor indicator of the objective value of the weapon.

I would tend to agree that there are certain mathematical analysis that can be conducted outside of a combat environment. Most of them tend to suggest the LBX is a poor weapon.

However, in a combat environment, there are other additional metrics which can be measured.. and honestly, over 10k games with two different builds, you almost certainly would get a great deal of useful data.

Now, of course, you're correct in that said data would be tailored to YOU specifically, and may not be broadly applicable to the entire piloting community. However, if you are a competent pilot, that data could likely be applied in a general sense to other competent pilots.

[Bishop Steiner]

Roland, on 29 April 2014 - 04:05 PM, said:

No, that wouldn't really be empirical testing.

No it's not "Scientific Testing", which tends to miss the big picture anyhow (empirical research is data gained through direct, indirect observation and experience, which is actually exactly what we are talking about), hence the reason things get filed tested instead of lab tested then deployed in the military. Because there are a ton of variables that often make something "lab superior" perform in an inferior manner in real world conditions. The M-16 of the early Vietnam War was "lab better" which is what Vlad is referring to, but in "empirical" testing the AK-47 was superior..... despite being slower firing and less accurate.

Outside factors, such as combat conditions, ranges, and even what you are pairing a weapon with can magnify or decrease it's effectiveness.
(edited to quote right post, lol)

Edited by Bishop Steiner, 29 April 2014 - 04:20 PM.

[Daekar]

On a lark I dropped two LBXs and 4 medium lasers in my K2 today. I've always done badly in the build, and I only had time for one match, but I did much better than I expected. Took out three mechs almost solo before my team got pushed back and I got focused down. Hardly evidence, but I was surprised nevertheless. I really hate not having PPCs on my K2, so I tore out the LBXs... but I might put them on another mech just for kicks. Perhaps my Protector would make a good candidate, with a std engine.

[MischiefSC]

Roland, on 29 April 2014 - 03:56 PM, said:

You should include the KDR in the stats you track, because it is a primary yardstick by which to measure effectiveness. The point of the game is to kill mechs, and K/D is the most direct measurement of your killing power vs. your survivability. It is certainly not the only yardstick, but it is one which is important enough that it cannot be ignored.

Simply tracking damage per shot won't really give you particularly valuable data in this comparison, because even if you were to land the entirety of an LBX shot on a mech and do damage with all of the pellets, they would likely still be spread over multiple locations, which would tend to be inferior to an AC10 landing all of its damage on a single location.

This is why K/D will be important, because it will more directly measure killing power.. In many cases, it'll benefit the AC10, which will be able to focus its fire and likely kill targets faster.. but this could potentially be tempered by the LBX's ability to do large amounts of damage after armor is exposed, and possibly even get more kill shots.

Ultimately, in this type of analysis, the more information you gather, the better... Its why video recordings of matches is so valuable, since you can analyze more nuanced aspects of engagements and see things that may not be immediately apparent in the raw numerical data.

Totally agree and am tracking KDR. It's just a statistic that because of the variance in matches takes a lot to shake out effectively. Then again I've pinned myself to 80 drops. Statistically that should shake out outliers, my stats on this chassis haven't changed much since about 80 drops and this should give a reasonable comparison sample.

Really need to have some other people who don't actually normally roll LB10X and have rolled with AC5s and 10s plenty try this, or it is simply too subjective to be of a lot of use.