21 replies to this topic

[ShadowedR]

WARNING: This is going to be one of those long, stats heavy posts that try to analyse certain aspects of the game. If you were looking for a place to list your grievances against AC/5’s or LL boating, this ain’t the place for you.
YES. I do know the forums are full complaints against this and that there are similar posts out there. I don’t care, I’m still posting this.
This is a follow-up to my previous “analysis post”, http://mwomercs.com/...age__p__2898402
Kay, here we go.


Introduction
I’m going to try to compare different weapons and different weapon groups to one another on a hack and slash maths + graph basis. The goal here is to discuss what we see and what the game feels like. There is a lot of info to crunch and a lot of graphs to draw so it’s a bit daunting. Anything that I might have “missed” is accidental and not meant to skew the numbers to a result that I wanted.




These 6 graphs illustrate the amount of damage dealt, spaced over 60 seconds. It is an accurate reflection of the REAL damage dealt to a single target, rather than just evaluating DPS.
Conversely, to the right of each DPS chart is the amount of heat generated by said weapons.
Overall, everything seems OK and balanced. Just a few special few things to note:

1. AC/20 starts and stays the top DPS Ballistic Weapon. But, it only generates the 2^nd most heat.
2. AC/2 deals a lot of consistent damage that is nearly on par with a Gauss.
3. AC/5 is your worst performer in terms of damage over time, but has excellent heat ratings. Good swarm weapon.

4. PPC’s perform poorly when compared to other energy weapons.
5. Large Pulse Lasers are extremely efficient in dealing damage.
6. ER PPC generates the most heat -> nearly double the heat from a Large Laser.

7. LRM6’s perform better than LRM10’s at less heat generation over time
8. SSRM2’s are down below in both heat and damage. This makes them excellent swarm weapons.
9. LRM5’s generate low amount of heat over time. Good swarm weapon.

OK, we know all of this already. What is new?
People load AC5 / SSRM2 / LRM5 into every hole they can and start blasting away at you.
These weapons all benefit from “Mediocre” damage output, with low heat generation.
Except you, Machine Gun. Nobody likes a crit seeker.


One Step Forward

First thing that pops into your head is the issue of Ammunition. Yeah, AC’s and M’s get a benefit in damage but are limited to x amount of rounds. Energy is allowed to shoot until the end of time.

*Interesting idea I read on the forums, consider Heat Sinks as the ammo required for Energy Weapons to function. Without them, you are one alpha strike and then nothing for 30s until you cool down. I’m not going to explore this idea, but it is interesting.

Right, so in order for us to take “ammo” into account, we have to bring weight into the equation.
It is right here that we try and optimise the amount of damage we can deal, keeping the amount of heat generated as low as possible, per ton we load on our mechs.



Both the missile and ballistic charts shows the amount of DPS / (Weapon Weight + Ammo Weight).

You can see that diminishing returns in DPS as you load more and more ammunition.
Note the logical sequence in Damage/T: AC/2 at the top, Gauss at the bottom. All is well in ballistic land.

Missiles are a bit of a mess: SRM4’s are at the top of the chart for low amounts of ammo, giving way to SRM6’s when loading more than 3 tons of ammo. Surprising is the fact that LRM20’s get out done by SRMS. No, not really. I mean, 270m range VS 1000m range. We’d expect to see that.

Another special note is the LRM10/15 -> both start off at near exactly the same DPS/T ratio.
For Energy based weapons we have the Small Laser coming out on top as the best DPS/T in the game. It beats everything. Secondly, is the SPL and ML -> Very efficient weapons to load.
PPC’s get outdone by all the lighter, more efficient guns. We want to see this, yes. Whats interesting to see, is that the actual data for:

AC5 : 0.302727
AC20 : 0.294118
GAUSS : 0.208333

PPC : 0.357143
LPL : 0.392857
LL : 0.424

LRM5 : 0.338
LRM20 : 0.356154
SRM4 : 0.492

These values are for DPS / (Gun Weight + 3T Ammo).
I’ve found that 3 Ton of Ammo will allow you to spam nearly as much as a laser and still allow you to fight for 70% of an extended battle. From here on in, any calculation that assumes ammo weight, they weight will be 3 Ton.

Ballistics get ruined by all the other weapon groups in terms of DPS per ton.
Look at how efficient LL’s are compared to AC20 and LRM20’s.
This just proves that Energy Weapons are good in being precise, low weight, high DPS / weight weapons (with horrendous heat generation as a drawback).

Another way to look at it would be, for every ton Ammo I load, I could’ve loaded 1 Medium Laser.
This is known as the Opportunity Cost. Again, something I’d like to analyse, but not today.


Nothing New

So, we’ve noted that AC weapons aren’t all that, are they? They are HEAVY to fit and don’t do as much Ton for Ton as other weapons. At this point, the stats become less hard en more voodoo in nature. I ‘created’ a ratio stat consisting of the following:
Alpha: (Damage / Heat) / Weight
DPS: (DPS / HPS) / Weight
This reads as “The amount of Damage dealt in relation to the heat generated, per unit weight”.
We get a ratio that factor in Damage, Heat and Weight to fit.


Wow, Small Lasers are super-effective.
Streaks are doing super well for themselves. And LRM5’s perform great, compared to other LRM’s.

Surprising enough, Gauss is a super cool ballistic to fit. And AC5’s aren’t looking bad either.
What I do want to point out is, (ER)PPC is the worst weapon one could fit, according to the stats shown. (ER)LL / LPL’s outperform the PPC variants completely.
I don’t agree with AC20’s place on the charts. It feels like AC20 is too buffed when compared to PPC, AC10 and AC2. Either the other guns need a boost, or this one needs a nerf.
A note on UAC. UAC is a bit of a conundrum. Because it has a % chance to jam, a UAC’s DPS varies as one keeps on firing. I’ll do a chart on a UAC’s potential damage output one day and post it, but all in all, UAC has the potential to be the best ballistic weapon in the game.
From this, I’d rate people should spam Medium Lasers, LRM5, any SRM weapons, Gauss and AC5’s in their mechs.

Wait, they are currently doing just that, aren’t they? With the exception of the gauss which has a different firing mechanic, the other weapons are employed where possible.
BUT WAIT! LRM’s and SRM’s deal spread damage. Same with the LB-10. You can’t compare them to weapons that deal pin point damage!
Well, yes. You are right, in a manner of speaking. Let us group them up, ok?

Group 1: Pin Point – AC’s and L’s
Group 2: Spread Damage – SRM’s and LB-10
Group 3: Fire and Forget – LRM’s and SSRM

Group 1 has the benefit of being precise damage. But it relies on your skill as a marksman VS the piloting skills of your opponent. AKA, I can AC/5 all I want, if my opponent is going to stand still and allow me to focus all of my shots on his centre torso, then he is going to die within seconds.
The downsides here are pretty small: It’s purely skill driven. Right, the two sub groups:
Ballistic and Lasers. Ballistics are held back by Ammo while Energy weapons are held back by less DPS and more heat Generation. A plus side to Energy weapons are, even if you miss your initial shot, you still have a chance to correct your aim and deal *some* damage. With AC, miss is miss.

Group 2 suffers in that, not only do you to aim accurately, the damage you deal is randomly distributed across your enemy. To make up for this, the amount of damage you can deal is significantly higher than other weapons of the same type (SRM / LRM).

Group 3: LRM’s have the benefit of doing all of the hard work for the pilot: You just have to maintain a lock and off they go. It takes a lot less skill to fire LRM’s and conversely, you’ll never see your LRM’s performing better than more skill based weapons.
So we need to assign some arbitrary measure of skill / accuracy / system / spread in order for us to get a clearer understanding of weapon systems.
I’ve chosen to go with some Accuracy Stats I data mined from our group a while back. Additionally, I applied the following principle:

The amount of shot you actually hit * damage per shot = the amount of damage you should have dealt on optimal range. I compared that to the actual amount of damage you dealt to factor in for shots fired out of optimal range, or where laser attacks wasn’t focussed on the target for the entire duration of the beam.


The adjustments reflect a more “real” approach to how effective each weapon really is. Just take note that the number used to turn the “optimal” into “real” are very random and based on average accuracy and damage across a few players. And those numbers take into account prior weapon mechanics, such as the way Gauss used to work etc.
With more data, separated by Patch, we can estimate more real numbers for the weapon analysis.


Conclusion

As far as conclusions go, this will be a very short one. Certain guns perform better when a lot of parameters are taken into account. We can clearly see why Medium Lasers, AC/5's and SSRM's are fitted in masses - they rock overall.

Other weapons sort off take a backseat to the top "few" weapons.

But, hey, discuss this with me

Edited by ShadowedR, 14 November 2013 - 06:31 AM.

[Furball42]

By Kerensky, why do I have this sudden urge to stuff as many SPLs into a mech as I can fit?

On the UAC front, we had a nice discussion on how to change the jam mechanic. The current one is just, meh. Maybe something like a progression in the jam % upwards the longer you fire, instead of just jamming on the very first shot (am I the only one?)

Anyways, good post. you have my beans.

[The pessimistic optimist]

I have to say you make many good points.

Edited by SirSmokes, 14 November 2013 - 06:54 AM.

[Why Run]

Are you missing a critical variable; Effective range? You kind of factor it, but by taking the values at the most effective range, you ignore what this variable really means. This is the real flaw with the AC/20 arguments, and a huge downside on the SLs that does not necessarily manifest in the pure statistics of damage and heat.

Personally, the AC/2 is the scariest weapon when mounted 3+. They tear off armor faster than nearly anything else.

Edited by Why Run, 14 November 2013 - 07:07 AM.

[Mechteric]

Quote

4. PPC’s perform poorly when compared to other energy weapons.

Unfortunately I think this is false, because the one problem with all lasers is duration, and in many cases its just not possible to keep the duration of the laser on the same part, let alone entirely on the mech in some cases. Looking only at heat vs. damage skews that representation without considering this factor.

[ArmageddonKnight]

An interesting post, im about to pop out right now so i cant go on to much of a rant, but one thing that did pop into my head when u mention LRM's is that, while they dont require to much skill to aim, they are a PITA to use when u keep lossing lock. out in the wide open where ur the one in view of the target the u are ok, but if ur locking onto targets u cant see and are rellying on ur teammates lock, they are very hit and miss.

I will point out im not defending how hard or easy they are to use, i dont use them, i never like using F&F weapons in a shooter, i prefer the more hands on skill bassed direct fire feel, its just that wheni have tried them i can understand why some peopel say they do requir skill, and others say they dont, its very ether "im in view, i lock, i fire, easymode" or "im not in view, i lock, i fire, i loose lock, i miss, bad day"

props for trying to make some sense of the games weapons, there will always be other ways to change the figures bassed on ones view of how the player affects the figures, but i give u props for trying non the less.

[Shadey99]

Err... you mention multiple times that lasers are 'pint point' rather than 'spread' damage, however more so than missiles lasers are quite likely to spread damage as most require they stay on target for as much as 1 second and there damage is actually incurred in small pulses or cycles during the beam duration.

Normal ACs, Gauss, and PPCs are 'pinpoint'. Lasers are 'variable' (Flamers can probably be in here as well). MGs are 'Random Cone of Fire'. LB-10X are 'Distributed Cone of Fire'. SRMs and LRMS are 'spread in groups'. And lastly SSRMs are 'random target' weapons. MWO loves to give differing firing methods to different weapons.

Beam Duration is a factor that seriously needs added to these figures (It also effects heat and firing cycle).

[Nothing Whatsoever]

With the SLs, it reminded me of a video I remember watching of an awesome boating them, couldn't find the video, but the build could have been something like this: http://mwo.smurfy-ne...9b0fc7c28656ca2

Edit: found it, I was wrong it was SPLs: http://mwomercs.com/...oat-with-video/

Edited by Praetor Shepard, 14 November 2013 - 07:57 AM.

[Almond Brown]

Sadly, a energy build with the equivalent of 20 damage (2 x LL) requires 19 DHS's (92% HE) (8 engine based) versus the same chassis with an AC20 and 3 tons of ammo (125% HE). So AC20 = 19t total (gun + 3t ammo + 2 DHS's to make 10), versus the 2 x LL = 24t (2 guns + DHS's to capacity). With minor armor tweaks, starting from maximum, also provides the AC20 mech room for 4 more tons of ammo. The Energy mech is both crit space and weight maxed after said tweaks.

Is this the way it should be?

Energy:

http://mwo.smurfy-ne...45462e12a85f0c5

Ballistic:

http://mwo.smurfy-ne...64e023650fa1e31

Edited by Almond Brown, 14 November 2013 - 12:30 PM.

[Shadey99]

Almond Brown, on 14 November 2013 - 12:27 PM, said:

Sadly, a energy build with the equivalent of 20 damage (2 x LL) requires 19 DHS's (92% HE)

Actually 2 Large Lasers is 18 damage (2x9 spread over a 1 second beam duration) or about 7/8ths of the 20 points. A better match are 4xML for 2 more heat generated (2 less if we had TT values) and can actually do 20 damage (again spread over a 1 second beam duration).

Edit: Yes, I did forget MWO made them 9 rather than TT 8. It's fixed now.

Edited by Shadey99, 14 November 2013 - 01:08 PM.

[Vodrin Thales]

Shadey99, on 14 November 2013 - 12:53 PM, said:

Actually 2 Large Lasers is 16 damage (2x8 spread over a 1 second beam duration) or barely more than 3/4ths of the 20 points. A better match are 4xML for 2 more heat generated (2 less if we had TT values) and can actually do 20 damage (again spread over a 1 second beam duration).

LL does 9 damage in MWO so it is actually 18 damage. And who requires 19 DHS to make 2 LL effective? I have a K2 build that runs 2 PPC and 2LL with 17 DHS and it will kill a dual AC20 Jager every time unless I make a bad mistake and let it get too close. Weapons do not need to fire indefinitely in this game. You just need to be able to fire in 5-10 second windows without overheating.

[ShadowedR]

Why Run, on 14 November 2013 - 07:05 AM, said:

Are you missing a critical variable; Effective range? You kind of factor it, but by taking the values at the most effective range, you ignore what this variable really means. This is the real flaw with the AC/20 arguments, and a huge downside on the SLs that does not necessarily manifest in the pure statistics of damage and heat.

Personally, the AC/2 is the scariest weapon when mounted 3+. They tear off armor faster than nearly anything else.

I actually did 'group' weapons into range categories, that but decided not to post it just yet.
I want to extrapolate a bit more from it.

And regarding the 3x AC/2 : I have a Shadow Hawk with that build. Lots of fun, but the ammo requirements are mad.

[ShadowedR]

CapperDeluxe, on 14 November 2013 - 07:09 AM, said:

Unfortunately I think this is false, because the one problem with all lasers is duration, and in many cases its just not possible to keep the duration of the laser on the same part, let alone entirely on the mech in some cases. Looking only at heat vs. damage skews that representation without considering this factor.

Conversely, To hit a moving target with an ER PPC over 950m isn't as easy as it used to be, so you end up "spreading" your damage due to aiming issues.

Every weapon has a degree of aiming involved - Due to the travel time of AC weapons, it is hard to hit a moving target at exactly the right spot the whole time. Unless he is an *****, standing still. Or you are up close to him. In which case, it becomes very easy to keep your lasers on target.

ArmageddonKnight, on 14 November 2013 - 07:17 AM, said:

props for trying to make some sense of the games weapons, there will always be other ways to change the figures bassed on ones view of how the player affects the figures, but i give u props for trying non the less.

Thanks

[GalaxyBluestar]

remember these stats are telling of only a jigsaw piece in the whole puzzle. for on the field effectiveness you have to consider the parameters actually involved on the field such as:

pilot skills

mechs speed, armour and weapon capacities

enviroment, hot or cold, inclines and obstructions

costs

situational; organised play, wolf pack play, solo play

situational 2: deployment of range in weapons, distribution of damage in weapons ie: splash, hitscan or pinpoint.

state of the game; latency, hit detection, statistical buffs and nerfs in the system, heat management.

etc etc

when factoring these in it's usually the most manourvarable and well armoured/less likely to be cored mechs stacking the most firepower damage output over time which can negotiate the enviroment, situational and state of the game aspects which are deployed on mass. check high tier play and a year's worth of meta and you'll find at the time the most deployed units will match all those criteria not just the weapon stats. only when you understand all of that can you tell what weapons are OP or UP.

in other words to understand how to even begin with balancing the game...



still op you're getting close

Edited by GalaxyBluestar, 14 November 2013 - 10:56 PM.

[ShadowedR]

GalaxyBluestar, on 14 November 2013 - 10:48 PM, said:

I'm planning on doing at least another two posts in the same fashion.
Try to combine as many factors as statistically possible.

What I think most people miss is, I've evaluated only a certain portion of weapons attributes and any conclusions made are towards the specific factors that I've included, not the weapon as a whole.

Analysis and statistics is a process, rather than a one shot answer.

[MustrumRidcully]

If you're interested in this kind of mathematical analysis, be it the OP or the poster, see also my signature.

One thing to consider:
Damage and Heat after 60 seconds is not that practically important. Think about how often you actually get to fire for 60 seconds straight at an enemy. I think you find that rarely happening.
When I made my first analysis taking into account the heat mechanics (including free engine heat sinks and all that) I also started with very long durations. Then I watched a few MW:O combat videos and played a few more games, looking at how long a typical fire exchange lasts, and the minimum duration I had started with (20 seconds) already looked like an outlier.
So I started using shorter time frames.

One of the challenges in balance in MW:O is that you cannot just look at heat per second to determine the impact of heat. In the table top, this might have worked, because you quickly suffered heat penalties and thus wanted to compensate most of the heat generated. But in MW:O, you have a long way to shutdown before heat means something bad, so you only have to compensate just enough that you don't overheat before the enemy blows up or the engagement ends otherwise. But once heat threshold has to be considered, you also can no longer look at a single weapon, because all weapons add their heat against the same heat threshold, they don't each have one. So you have to look at builds. My final result was to calculate how much damage a "build-like" set of weapons would produce within a certain time frame and how much heat sinks and ammo it would need to do that, and set damage output and weight requirement in a relation.

And that's basically what I am doing in the thread in my signature.
By which I mean this one: http://mwomercs.com/...rts-2013-11-10/
And this was the long explanation form: http://mwomercs.com/...5th-2013-patch/

And these are the results for the current weapon stats, described via graphs.

Spoiler

Here are the charts without the missiles, for better readability.
High Damage Output Chart



Low Damage Out Chart

I wonder if PGI has made analysis like the OP or me... And I'd love to see them.

Edited by MustrumRidcully, 15 November 2013 - 01:12 AM.

[ShadowedR]

MustrumRidcully, on 15 November 2013 - 01:09 AM, said:

If you're interested in this kind of mathematical analysis, be it the OP or the poster, see also my signature.

Spoiler

Here are the charts without the missiles, for better readability.
High Damage Output Chart



Low Damage Out Chart

I wonder if PGI has made analysis like the OP or me... And I'd love to see them.

Thanks. Interesting read.

Regarding the "engagement time"
Engagement time varies between Mechs, Weapons and Pilot (also, how dumb your opponent is) IMO

That is why I decided to plot all the way to 60s.
Although, I could have stopped @ the 30s mark.

Anyway, you keep these stats up to date with every patch?
I was thinking of restructuring all of my excel sheets for easy data entry and keeping it up to date.
But only after I'm done taking into account as many aspects of the game as possible.

As for PGI pulling stats, I am sure they are monitoring the game to some degree, although I would question the depth at which they calculate their stats.
Not too brag, but I did go a few steps deeper than the apparent stats most people would drum up on excel. Specifically when you plug in player stats to adjust the final results closer to the real world.
I doubt they do that.

[MustrumRidcully]

Yes, I am usually updating them after some patch. It's not a big deal to do so - there is a link in one of the posts to a spreadsheet that I use. You can easily enter updated new stats. I intentionally set the thing up to also handle changes to heat threshold and heat dissipation. Occassionally I experiment with different stats to see how the charts could look "nicer". If I could have my own M:WO Test Server, that would actually be useful and fun.

There are two avenues to me that would be particularly interesting to explore to expand the methodology:
1) I only look at boated builds. I don't think my Excel abilities are sufficient to actually create distinct, seperate builds, but I think it might be possible to write a computer program that could also iterate different weapon combinations (so you can see if there are some ways to get "ahead" by mixing weapon types, for example, or see if this usually hurts.

2) I would love to find a way to model the damage over time vs single projectile damage vs multi projectile damage and how this affects accuracy. Unfortunately, I don't think you can mathematically model that in the cold, you need actually server statistics (and even the ones PGI offers under your profile might not suffice.)

[ShadowedR]

MustrumRidcully, on 15 November 2013 - 04:47 AM, said:

Yes, I am usually updating them after some patch. It's not a big deal to do so - there is a link in one of the posts to a spreadsheet that I use. You can easily enter updated new stats. I intentionally set the thing up to also handle changes to heat threshold and heat dissipation. Occassionally I experiment with different stats to see how the charts could look "nicer". If I could have my own M:WO Test Server, that would actually be useful and fun.

There are two avenues to me that would be particularly interesting to explore to expand the methodology:
1) I only look at boated builds. I don't think my Excel abilities are sufficient to actually create distinct, seperate builds, but I think it might be possible to write a computer program that could also iterate different weapon combinations (so you can see if there are some ways to get "ahead" by mixing weapon types, for example, or see if this usually hurts.

2) I would love to find a way to model the damage over time vs single projectile damage vs multi projectile damage and how this affects accuracy. Unfortunately, I don't think you can mathematically model that in the cold, you need actually server statistics (and even the ones PGI offers under your profile might not suffice.)

What I would appreciate from PGI's side is the global figures, same as in your profile, but for everyone. Separate these numbers per patch. The amount of analysis we could do then...

As for modelling and determining the stats for each Variant + Possible Load Out, it is possible. If you have the time to sit and do it

Just a lot to consider in this process: Ammo amounts, Favored Weapons, Include JJ's, BAP's ECM, etc.
A program would "optimize" a build by loading just 1 ton of ammo and ignoring BAP/JJ/ECM completely.
Also, stripping the maximum amount of armor would allow the program more room to add more guns -> so you'll have to create a ton of control mechanism.

But it is doable.

[Dr Slap]

These Analysis are always fun to read - thanks for saving my lunch break - though unfortunatly completly academical. What you need is what PGI does. They Track the game statistics. We see the Part in our Personal Stats. They can then use a simple KPI mechanic to adjust balancing.
E.g. Lasers 40% of all weapons fired, Ballistics 30% and Missiles 30%
Or same for damage done.
Also they can have a kpi like no weapon should have a below 3% Share of being equipped. They might even have a curve for weapon equiping should Look like.
If they find a deviation to their KPI they can just Look into this weapon/ weapon Group and adjust accordingly.
same is TRUE for mechs used.
Brian once said that they need 1-2 months to understand the effect a weapons Change had on Balance. This is because after a buff everybody will try a new weapon ( Hype) but this will even out after a few weeks and a Plateau will Form (we can see this with artillary now - still Hype Phase)
If you want good, game relevant data get 200+ people to give you a weekly Update on their stats ( though they need to be from several Clans to lessen bias). Then Analyse this data and you can challenge PGI!
game balancing is not about heat or Range these are the levers a Designer uses. It is about making a game challenging without being too difficult and and closing loopholes.
and most important it is about making a game FUN!