Guild Wars 2

Excelsior comrades

You may have noticed how many people come up with dps calculations for various classes. You must have also noticed how varying these figures can be. Lately it was Gaung and I who disagreed on elem staff’s max dps. We cross checked our computations and it appeared that we had a few differences in our approaches.

One of the things I would like to discuss is what buffs we should include when calculating max dps. Guang’s view of “full buffs” is quite literal, whereas I consider a speedkill situation: no ranger.

The question is: should be include ranger buffs? With traits?
also:
Should we include mob-specific potions?
Should we include bloodlust or perception stacks in addition to weapon sigils (debatable since they are going to be tied to the weapon you have equipped in the future) ?

It would be nice if we normalised our calculations a bit so our figures could be comparable and reliable

It would be easier to follow those calculation if they all were without any buffs, potions, runes and other enhancing abilities.
But as those runes and sigils and other buffs/traits are a major part of different builds and how they work, it would be best to just note max power and max crit as two base lines.
But I would leave out bloodlust and potions.

I would do an unbuffed and a buffed calc. The buffed calc would include might, vuln, fury, banners, food, potions, sigils and stacks. I wouldnt include ranger frost spirit because thats not as commonly picked. Only reason I would use banners is because every decent team takes atleast 1 warrior so it makes no sense to leave banners out.

The unbuffed calc would include sigils and stacks but not boons, banners or consumables.

I use Banners, Fury, Vuln, Might, Food, Potions, Sigils, stacks and runes in my estimations. I would like to use ranger stuff because it would be great and change a lot (with spotter certain builds could use Bloodlust where they current use Perception and get a massive DPS boost) but it’s just not practical since almost nobody has a fully geared ranger. And those that do, almost none of them enjoy playing it.

I think you should leave out the Ranger buffs. Yes, Rangers were proven to be good in some niche situations. But more often than not they’re kind of lackluster in practice, so I wouldn’t expect to see too many of them in the majority of content (speed run wise, anyway)

Keeping it as simple as possible would be best IMO.
So no stacks, no food buffs, Rangers are hardly common practice so no to their buffs.
Preferably no Vuln or Might either (beyond what the profession can create on its own.) Though you’ve got issues there with classes that can’t maintain Fury on their own when setup for optimal DPS. However I would see the precision from Banner of Discipline as a constant because you’d be nuts to not have that at every opportunity. Idk I just don’t see why you over complicate things by taking so many things into consideration when you can just compare two builds in a sandbox environment.

Yes that’s why I often make two computations:
- A realistic situation computation: with only the buffs self-provided and generic sigils.
- A speedkill situation computation: with all common speedkill buffs but no content-specific sigils or potions.

Dolan.3071:

Keeping it as simple as possible would be best IMO.
So no stacks, no food buffs, Rangers are hardly common practice so no to their buffs.
Preferably no Vuln or Might either (beyond what the profession can create on its own.) Though you’ve got issues there with classes that can’t maintain Fury on their own when setup for optimal DPS. However I would see the precision from Banner of Discipline as a constant because you’d be nuts to not have that at every opportunity. Idk I just don’t see why you over complicate things by taking so many things into consideration when you can just compare two builds in a sandbox environment.

Basically, the entire point of theorycraft is to get reliable data with which to make better decisions that will increase efficiency in game. If you use situations (like your example of only measuring the stuff it can create on it’s own) that don’t actually exist in actual team play, you will arrive at inaccurate conclusions.

In organized groups, fury and might, banners and dungeon potions and food buffs are a given. 25 stacks of vuln is debatable because it’s much harder to maintain permanently on a boss. But the other stuff is a no brainer.

I have no idea why you exclude content specific potions. Have you ever run in an organized Arah group that didnt use undead potions intentionally>

Nike Porphyrogenita.8137:

I use Banners, Fury, Vuln, Might, Food, Potions, Sigils, stacks and runes in my estimations. I would like to use ranger stuff because it would be great and change a lot (with spotter certain builds could use Bloodlust where they current use Perception and get a massive DPS boost) but it’s just not practical since almost nobody has a fully geared ranger. And those that do, almost none of them enjoy playing it.

you have just convinced me to start bringing my double bear longbow clerics ranger to more guild runs.

You’re comparing x build using a slayer potion against y build using the exact same slayer potion, you can just cancel them out and keep things simple? :/
Adding 10% damage seems unnecessary when the ultimate goal is seeing which build comes out on top. Same can be said about Banner of Strength and Scholar Runes (though most people have this anyway so I would consider it a constant between two different builds.)
I suggested excluding stacks because well… As an example if a build depends on maintaining perception stacks to give itself a reasonable crit chance (>85% or so) then that’s rather counter productive because there aren’t many dungeons where you can get stacks right off the get go, just look at CoE p1, I don’t even think there’s 25 things to kill in the whole path. Then there’s Arah where you would have to waste time killing trash at the start which isn’t ideal either.
Then of course who would want a build that is crippled if you happen to lag spike or just plain goofaboof.

I think you’re right about Might stacks though, in an organised run you should be capped all the time so relying on what a build can produce itself would be inaccurate. So either comparing 25 might (or 0) as a constant factor.
As for Vuln I would just compare the builds based on what they can produce alone, unless of course you regularly run with an Engi or w/e where the build’s Vuln contribution is irrelevant. But in that scenario it’d be like “x build is better when trying to cap Vuln and y build is better when Vuln isn’t an issue”

You only need to kill 13 things, and it takes 5 seconds to fully stack in Arah. You can also stack on the trash before path selection in CoE.

Nike Porphyrogenita.8137:

I use Banners, Fury, Vuln, Might, Food, Potions, Sigils, stacks and runes in my estimations. I would like to use ranger stuff because it would be great and change a lot (with spotter certain builds could use Bloodlust where they current use Perception and get a massive DPS boost) but it’s just not practical since almost nobody has a fully geared ranger. And those that do, almost none of them enjoy playing it.

I do. /slap

@Dolan

Just because two builds have the same buffs doesnt mean they can be ignored because they cancel out. Due to the way the buffs interact with calculations they skale differently depending on the class/build.

As far as I can figure there’s not much point in cross-referencing.

It’s not like you’re looking to crack open mathematica and plot out datasets from your combined field testing with hundreds of subjects, here. It seems to me like you guys are approaching the community with two different hypotheses derived from reasoning the same topic in two different ways. Which is totally fine.

What you guys have is Math, not Data. There isn’t any actual experimentation to compare, so why does it matter if you’re on the same page or not?

spoj.9672:

@Dolan

Just because two builds have the same buffs doesnt mean they can be ignored because they cancel out. Due to the way the buffs interact with calculations they skale differently depending on the class/build.

Mathematically, you can ignore vulnerability, non-stacking sigils, potions and frost spirit because they will increase dps by 61.8375% assuming night+slaying sigils and disregarding all condition damage.

haviz.1340:

spoj.9672:

@Dolan

Just because two builds have the same buffs doesnt mean they can be ignored because they cancel out. Due to the way the buffs interact with calculations they skale differently depending on the class/build.

Mathematically, you can ignore vulnerability, non-stacking sigils, potions and frost spirit because they will increase dps by 61.8375% assuming night+slaying sigils and disregarding all condition damage.

I support this approach.

how about when we come up with dps numbers we also provide how many stacks of vuln the build is able to produce on average?

As for bloodlust, scholars, banners, and other non-modifier stats we must include them all in speedkill calculations since they are not flat boosts.

Base damage modifiers are multiplicative. The more you have the better they are. Ignoring certain ones because “they cancel each other out” negatively biases builds whose traits give more modifiers vs ones who just increase normal stats.

1.1*1.1*1.1*1.1*1.1 = 1.61

That last 10% is worth a bit more than 10%. If the last 10% was a trait in your build and you ignored the first four, it would make that last 10% not quite as good as it is in actuality.

We are talking about ignoring modifiers that are not build-dependent: potions, sigils etc

The Mexican Cookie.3690:

You only need to kill 13 things, and it takes 5 seconds to fully stack in Arah. You can also stack on the trash before path selection in CoE.

I forget people do this. To be honest I’m not sure why I don’t have two stacking weapons myself.


I’m only suggesting that we try to keep variables to a minimum for easy comparisons. Like Haviz said I assumed that Vuln, slayer potions and all that jazz added a flat bonus to the build’s damage output. But I see what you mean about including Banner of Strength + bloodlust because it would skew results with builds that have different damage modifiers.

But really if x build was better than y build only when it had full stacks I know which one I’d rather use from day to day xD
For the record setting crowd? I can definitely see the appeal.

Zelyhn.8069:

We are talking about ignoring modifiers that are not build-dependent: potions, sigils etc

But you cant because of what Nike just said. Those extra modifiers provide a larger dps increase than they do on other builds so ignoring them creates an unrealistic difference in optimum conditions. Its simple maths :P

spoj.9672:

Its simple maths :P

Apparently no.

spoj.9672:

Zelyhn.8069:

We are talking about ignoring modifiers that are not build-dependent: potions, sigils etc

But you cant because of what Nike just said. Those extra modifiers provide a larger dps increase than they do on other builds so ignoring them creates an unrealistic difference in optimum conditions. Its simple maths :P

Well I insist:
If build A does 10k dmg without potion and build B does 12k damage also without potion, then the difference is 20%, with potion the difference would still be 20%
( 12k * 1.1 ) / ( 10k * 1.1 ) = 1.20
It does not matter whether build A has 70% total modifier while build B has only 50% and more power for example.
All common flat modifiers can be discarded regardless of builds.

However, banners and other absolute buffs must be taken into account.

Edit: just to be clear, the increase from potion is indeed larger in absolute terms for build B, but not in relative terms.

Ok lets compare two build with different modifiers and then factor in the modifier from scholar runes.

1.1*1.05 = 1.155
1.1*1.05*1.1 = 1.2705

Which means scholar provides a percentage increase of 74.5161% between the end modifiers.

Now for a build with more modifiers.

1.1*1.1*1.15 = 1.3915
1.1*1.1*1.15*1.1 = 1.53065

Which makes scholar provide a percentage increase of 35.5429%

Please tell me what im missing? Because the way I see it you cant ignore any common factors unless all builds have the exact same damage modifiers in traits.

Edit: I sort of see what your saying but I dont think its that simple. The gap is widened on certain builds when you include more or less buffs for a comparison. So build A could be providng quite a bit less dps than build B in X conditions but in Y conditions the values are actually closer? Agh my brain hurts, I should probably just do some actual full calculations to check if what im saying is true haha.

How do you calculate these percentage increases? All I can get is 10% increase for both cases!
Calculation: new / inital -1 = percentage increase

[(new – original)/original]*100

so [(27.05-15.5)/15.5]*100 = 74.51%

spoj.9672:

[(new – original)/original]*100

so [(27.05-15.5)/15.5]*100 = 74.51%

Why would you truncate them ?? :o

Try this:
( 1.2705 – 1.155 ) / ( 1.155)
and get the correct increase.

Yep ok I see what ive done wrong haha. It does seem to cancel when using the actual numbers. But that means you can only ignore potions, vuln and possibly scholar runes. Some builds wont use the same sigils (2H vs 1H) and some might benefit more from different runes.

Discussion helped me so thanks for that. xD

I still think you should include all though, just to give an accurate value to optimum conditions. Doesnt take much extra effort to factor in 3 extra modifiers to each build. :P

Np

You do need to include scholars since they provide stat buffs (power and crit damage).

By the way can we agree on an abbreviation for critical damage? “cd” is already taken, how about “bcd” (bonus critical damage) ?

As for sigils, that’s the point: for optimal dps we are all going to use a night or slaying sigil, so on 2H sigils could be ignored and on 1H we could include only one sigil.

Yes we could include all of course
Actually we probably have to since some damage dealing entities do not scale with player’s attributes: mesmer phantasm (if I am not mistaken) which would skew calcualtions.

Yeah obviously you have to include the stats on scholars. And yeah for simplicity I would suggest including all buffs. But with sigils im pretty sure most people take 1 force/night/slaying sigil and then something like accuracy/battle or energy on the offhand. It depends and I think thats why we should include them. The only time most people stack 2 dmg modifier sigils is in record attempts as far as I can tell.

The other thing about including all the buffs is it helps determine which stacking sigil is superior.

For example, a build with 88% chance to crit without Banner of Disc would almost certainly appear to benefit more from Perception stacks. However, when you include the Disc banner Bloodlust stacks are the clear winner.

Erring on the side of more (and better) data is always superior when trying to do science, as your results are only as good as the inputs.

For what it’s worth my updated spreadsheet includes all the common buffs as a base already so it’s not like it takes additional time to calculate. There is literally no harm in including those buffs and only potential gain so I’m not sure why the opposition is so intense.

Yes your spreadsheet is definitely a reliable tool

So we agree on using “bcd” as an abbreviation for bonus critical damage?

Also we agree on using realistic speekill situtation conditions to calculate max dps? that is without hunter buffs and with war buffs

How about rotation calculations?
Guang used an “unscaled base dps” approach – resulting in base dps a lvl 80, then scale for dps
Sabull and I tend to use a step by step coefficient calculation – resulting in coefficient per second figures, then apply full formula to get dps
These two methods are both valid (when used accurately), but people tend to prefer reading the results in one format rather than the other.

For example, would you rather learn about elem staff’s rotation (lava font, 3 fireballs) as: “465 base dps” or “1.38 coefficient per second modified for relative weapon damage” ?

Eh, none of the above?
The reason you’ll see a group like ElitistJerks use Rotations is because they make sense for the game.

In WoW your output is delivered automatically to the target in a more or less constant steady stream, and the actual meat of the gameplay is in resource management. They focus on rotations because they can just assume they’ll have constant unbroken contact with the enemy and the thing most people want to know is what they can do to be cost-efficient.

Rotations are an awkward choice for GW2 because it has some action game elements. When it’s at it’s best (and it’s not always) you can’t assume unbroken constant contact with the enemy, and the thing most people want to know isn’t what skills should be done in what order to keep from running out of mana, but what skills are the best for this fight/dungeon.

If you’re having a hard time figuring out how to present your information to others, maybe take into consideration how the min/max communities of action-games do it.

For example;
Older fighting game communities also couldn’t assume constant unbroken contact. They needed to be able to analyze the value of attempting a Combo even if it was broken in the middle, like you’d want to analyze the value of a Rotation even if it was broken. They did this by categorizing them according to when the most damage took place, AKA; Frontloaded and Backloaded.

Also we agree on using realistic speekill situtation conditions to calculate max dps? that is without hunter buffs and with war buffs

The real value of Theorycrafting is in how it contributes to our common knowledge and helps us to reverse engineer the game’s structure. No Theorycrafter has ever followed the scientific method to arrive at any actual honest to god conclusion. (And, for good reason, the participant sample size you’d need for even one experiment is ungodly).

So I wouldn’t agonize over your Methodology, because you’re not actually going through with an experiment. You’re creating a hypothetical scenario that will hopefully display some information that exposes the inner workings of the system that can then summarize for public consumption.

Like in all cases of reverse engineering, the more variables you play with, the better our understanding will be. It’s one thing if you’d like to focus on something very specific, but try not to limit the variables of others unnecessarily.

Honestly I don’t get where the idea that rangers are impractical is. An 8% crit boost and a 7% raw damage boost plus a respectable might and fury contribution for the entire party more than makes up for their relatively low DPS. Hell, you could let the pet die and just camp GS the entire fight and it’d still be a positive gain in most cases. I think we discussed this quite a bit a while ago, sword/axe with feline is the minmax deeps version but it doesn’t mean the easier “scrub” variants aren’t still worth bringing along.

Ranger boosts are impractical because few people have properly geared rangers, and of those that do, they rarely opt to play them.

Not to mention many people would rather not take them into dungeons like arah where the sword makes it more awkward to dodge lupi’s attacks.