Guild Wars 2

So they retired the hint achievement, I think those of us who were tracking an achievement caused an issue when our achievement tracker tries to load.

Same problem here:

• Crash <—*
  Assertion: tierDenominator > 0
  File: ..\..\..\Game\Ui\Widgets\ObjectiveList\OblAchievements.cpp(1758)
  App: Gw2.exe
  Pid: 1820
  Cmdline:
  BaseAddr: 01210000
  ProgramId: 101
  Build: 38569
  When: 2014-09-09T21:25:06Z 2014-09-09T17:25:06-04:00
  Uptime: 0 days 0:04:09
  Flags: 0

No, PvE stuff like Fractals and whatever else there is nowadays.

Thanks for putting this together, I do have one question though, regarding:

Tarsius.3170:

Q25 What are effective guardian builds?
There are different builds guardians can run depending on PvE, WWW, sPvP, or tPvP. Here are some links to the various guardians that have tested and contributed builds to this forum. – Blasino.3128

• Jax’s (Silven) sPvP/tPvP builds
• Christos’ Healway
• Kharel’s Boonway
• Amin’s WWW builds
• Obal’s fractal builds
• Silentstorm’s Paladin build
• Harbinger’s Paragon build

Are these builds still the standard builds? I haven’t played since Jan/Feb and I know there have been changes to the class so I wanted to know if the builds linked to were still relevant in the current meta.

Thanks!

Sounds like good changes, well other than making Spirit Weapons even worse than they already were!

I havent played since January or February, at that time AH was the generally accepted hotness with glass cannon builds with lifesteal food being the choice of the “elite”. What has changed for Guardian? did they change how any of or stuff works? buff/nerfs?

What are the “standard” builds these days?

Thanks in advance for any help you can give me =)

I have two ati 7970m video cards and they don’t seem to work in crossfire in gw2.

I got the newest drivers and the newest CAP but the crossfire icon never turns on and the gpu utilization for my 2nd card is always at 0%.

Does it just not work in GW2 or is there something I can do?

I think that for Hammer on a stationary boss the “best dps” in theory land is to just repeat 1 but Mighty Blow adds more than just raw DPS.

Mighty Blow in a Light Field (always available due to swing 3 of your Auto Attack) grants retaliation to some number of people which will trigger Altruistic Healing (assuming you are one of the 98% of guardians who use that trait). It also adds a small amount of damage when the target is hitting someone with Retaliation (looks like about 350 damage everytime the target attacks someone affected)

Where Mighty Blow really shines is if there is a Fire Field down, granting 3 stacks of Might to nearby allies. This not only heals you via Altruistic Healing (I haven’t had a chance to test if it heals you for the # of targets or the number of boons applied) but it also increases Power by 105 for 20 seconds (plus boon duration bonii). Assuming the group isn’t capped on might stacks and the average group member has between 2,000 and 2,500 Power buffed you increase each affected person’s dps by 4-5% for 20 seconds.

Old damage = power/armour*weapon*coefficient
New damage = (power+105)/armour*weapon*coefficient

since the players weapons, the coefficients of their skills and the bosses armour remain constant the ratio of new to old damage is:

(power+105)/(power)

If you hit 3 people (including yourself) with this at least once every 20 seconds you are effectively adding 12-15% of a person’s damage. If you have +boon duration, can coordinate with other team members regarding the placement/timing of fire fields and if the whole group is within range of the AE Might given out by the Blast Finisher you could be granting 3 stacks of 34 second might every 9 seconds (an average of ~11 stacks of might or +385 Power) to 5 people. Again assuming Power of 2,000 to 2,500 you are increasing their damage by between 15.4% and 19.25% each – multiplied by 5 players is another 77% or 96.25% of the average of the 5 players. I don’t think you’ll realistically get to that point but it’s good to know the potential, and I know some of the people in the thread had expressed that they wanted the figures in a “best case scenario”.

Tests of Significance

This has no place being used to compare damage, Foofad has it 100% correct when he says:

Testing for significance in this case is totally pointless. You test for significance when you have a data set from which you are gathering statistics, and you are attempting to ascertain whether or not the data reflect a statistically significant change. We aren’t working with a data set here. Increasing power by 100 has an absolute effect on your damage. That damage is a % increase for a given critical chance and damage, and that’s absolutely true. There is nothing to test. You will do x% more damage with 100 more power. It’s a given. There’s no chance in a sterile environment like this of it being any different.

Where you would use a test for significance, is if you smacked a monster 100 times and then had someone else smack a monster 100 times, compared your data sets, and then tested to see if the fact that your numbers are seemingly x% higher than his is in fact statistically significant given the sample size or whether that x% is just noise from your sample. That’s an appropriate test for significance.

Exactly as he said, if you were collecting data and had recorded the damage done by each weapon you could use a test for significance then, if you only took 4 swings and one weapon had 10% more done that may not be a significant enough difference – perhaps it just got good RNG. After 100 swings with each weapon that same 10% difference might be more significant, after 1000 swings even more so. In essence as you record more data points (swings) your data becomes more accurate because the RNG factor should even out over larger sets of data. The more data points for each set the more likely any difference between the sets is a significant actual difference in sets and the less likely it’s due to random “noise”.

Here is an example, I have 2 six sided dice with nonstandard numbering. Which dice is “better” (i.e. going to roll higher on average)? We have 2 approaches to this.

a) Add up the values on all 6 sides and divide by 6, this gives us the EV of the Dice, if it is higher than the other dice it is better – no test for significance needed.

b) Roll each dice a number of times, record the values, average them and compare – in this case we need to perform a test of significance to see if we got enough data points in the set to justify the difference as being related to the value of the die or if it’s just because you rolled better with one die than you did with another.

Do you see the difference? Do you see how a test of significance is appropriate in situation b but not in a?

Obtena: I can criticize him because he’s mis-using a statistical tool on things are are not statistics! It’s a fine tool but it’s being completely misapplied.

Balerathon: I don’t think you a troll and I hope you don’t think I am! Yes we are approaching this differently but I’m trying to get you to see that a) Your use of a test of significance is not appropriate. b) Some of the other numbers you are getting kind of have no relevance on anything – if I’m missing how the “Total Average” coefficient you give for each weapon has ANY bearing on how much damage you do please explain it to me.

So, some of the numbers are correct, others aren’t – I think they’ve been corrected in the original post now. I gave both numbers for a stationary boss – which is the theoretical max you are interested in, as well as an arbitrary situation where a boss is moving. Just because you are not interested in the additional data I provided for those who are doesn’t invalidate the “Stationary Boss” figures.

Re: Activation/Cooldown times.

Yes, the reported activation times are not correct, I have recording the animations and timed them frame by frame to find out the actual animation times.

If you want your analysis to have any bearing on the relative damage done by two weapons you absolutely need to take into consideration the activation and cooldown times. Imagine these 3 theoretical weapon sets.

Feather:

1. 0.8 coefficient, 0.5s attack speed, no cooldown
2. 1.0 coefficient, 0.5s attack speed, 5 second cooldown
   Total Average = 0.9

Car:

1. 3.0 coefficient, 5s attack speed, no cooldown
2. 8.0 coefficient, 10s attack speed, 20 second cooldown
   Total Average = 5.5

Wet Noodle:

1. 1.0 coefficient, 1s attack speed, no cooldown
2. 10.0 coefficient, 1s attack speed, 120s cooldown.
   Total Average = 5.5

Clearly the Wet Noodle and the Car are better than the Feather right?

Wrong.

Feather will in 5.5 seconds perform 10 #1 attacks and 1 #2, giving a total coefficient of 8.2 over 5 seconds, or an average of 1.6364 coefficient per second.

Car will in 30 seconds perform 4 #1 attacks and 1 #2 attack, giving a total coefficient of 20.0 over 30 seconds, or an average of 0.6667 coefficient per second.

Wet Noodle will in 121 seconds perform 120 #1 attacks and 1 #2 attack, giving a total coefficient of 130 over 121 seconds, or an average of 1.0744 coefficient per second.

DPS

Ok regarding your quick point about DPS and asking if I calculated numbers, lets look at the formula

[Damage] = [Power]/[Armour] * [Average Weapon Damage] * [Coefficient]

Point number 3 of your assumptions is “everything else is equal”.

So I assume my Power will be equal regardless of the weapon choice I use (innacurate since GS gives a might stack from it’s Auto Attack Chain, but we can disregard that), I assume the target will have the same armour regardless of my weapon choice and that both weapons I’m choosing between are the same tier and therefore have the same Average Damage (If we’re comparing an Exotic Hammer to a Fine Greatsword it’s not really a reasonable comparison).

So I’ll say Power, Armour and Weapon damage are constant for any given situation and I’ll define a constant “k” to be = [Average Weapon Damage] * [Power]/[Armour]

Our formula is now

[Damage] = k*[coefficient]
DPS = k*[coefficient]/[time]

I can compare the dps of two skills or rotations as follows:

dps ratio = (k*[coefficient a]/[time a]) / (k*[coefficient b]/[time b])

k appears in both the numerator and denominator of this so we are left with:

ratio = ([coefficient a]/[time a]) / ([coefficient b]/[time b])

So “relative dps” can be determined by comparing the coefficient/s of the two abilities.

Moving Target:

Again, if you aren’t interested in the moving target numbers just scroll on by that section. Some people may be interested in whether a moving target affects both weapons equally, if you’re not one of those people just look at my stationary boss numbers for a potential damage comparison.

Theorycrafting is about using math to determine relevant information on how your choices affect the game. Adding up the number of letters in all the abilities in a weapon set and dividing by the number of abilities may be a legitimate way of determining the average number of letters in the names of a weapons abilities but it has no relevance on which weapon does more damage just as averaging the damage done by abilities without considering how long that ability takes to use or how often you can use it has no relevance on which weapon does more damage. It’s the theorycrafting version of looking at your screen and saying “I see big numbers, big numbers = good”.

Ok, I guess if I’m going to criticise your math I should do the courtesy of showing you how I feel it should be done right?

Hammer:

1. 0.8, 0.9, 1.0 (symbol for 1.5) – 3.7s activation, no cooldown.
2. 1.75 – 1.75s activation, 5s cooldown.
3. 0.8 – 2s activation, 15s cooldown
4. 1.0 – 2s activation, 25s cooldown

Assuming all other factors are constant (Power, Target Armor, Weapon) Dps can be compared by dividing the coefficient by the time taken to perform the skill. I’ll present numbers for both 1 tick and the full 3 ticks of the symbol.

1. 0.8649 (1 tick) 1.1351 (3 ticks)
2. 1.0
3. 0.4
4. 0.5

As you can see #3 and #4 are abysmal for DPS and should only be used if you need the utility. #2 does worse than just auto attacking if you are getting all 3 ticks on the target but it will blast combo fields so it’s usually worth doing for that reason!

For Greatsword we’ll assume single target dps with both a relatively stationary boss so you can get all ticks of your #4 and get inside the boss to hit all 14 hits of WW and a moving boss that gets 2 ticks from #4 and only the 7 main hits from your WW.

Greatsword:

1. 0.8, 0.8, 1.2 – 2.4s activation, no cooldown.
2. 3.5 – 2.75s activation, 10s cooldown.
3. 1.1 – 1s activation, 15s cooldown
4. 2.5 – 1.75s activation, 20s cooldown
5. 2.0 (it will hit twice vs single target) – 2.25s, 25s cooldown (also puts a 10s dot on the target that ticks for 160+0.2 of your condition damage each second for 10 seconds, this is unfortunately very difficult for us to model)

DPS is

1. 1.1667
2. 1.0309 (7 melee hits) 1.2727 (7 melee + 7 projectile hits)
3. 1.1
4. 0.5714 (2 ticks) 1.4286 (full 5 ticks)
5. 0.8889 (plus the 10 second dot)

For the purpose of evaluating damage (rather than utility) I’ll assume Hammer uses a (1,1,1),(1,1,1),2 rotation and that GS uses Auto-attack and #2/#4 on cooldown vs a stationary boss.

Moving Boss:

Hammer
8.15/9.15 = 0.8907dps

Greatsword
1.1667

Greatsword pulls out ~31% more dps than Hammer, granted the Hammer is also blasting combo fields and if they are fire fields could be adding a bunch of might to the party or healing if they’re water fields. Hammer is also putting down a protection symbol which supports the party.

Vs Stationary Boss

Hammer
10.15/9.15 = 1.1093dps

Greatsword

1. provides 1.167dps
2. provides (1.2727-1.167)*2.75/12.75 = +0.0228
3. provides (1.4286-1.167)*1.75/21.75 = +0.0210
   Total DPS = 1.2108

Greatsword only enjoys a 9% lead in dps here, Hammer making up a ton of damage by getting the boss to sit in it’s symbol consistently. Hammer is still blasting combo fields every 9 seconds buffing and supporting it’s party and is still dropping protection around the boss so it and other melee are benefitting from -33% damage taken.

Remember math and theory can only take you so far, with so many other factors in GW2 combat not considered here you cant use this to definitively decide on a weapon. Hammer’s symbol damages 5 targets so it’s good vs large numbers of targets, GS#5 was disregarded due to difficulty modelling the DoT damage, but also it has great utility pulling targets together to allow not only the Guardian but the other party members to AoE more efficiently.

I think different encounters may just call for different weapons, and when in doubt as to whether to bring a Greatsword Guardian or a Hammer Guardian with you into the Fractals of the Mists, I say you can’t go wrong with bringing both!

Even when theorycrafting I think it’s important that you clarify the assumptions – in this case you are assuming the target is relatively stationary, in a lot of fights this will not be the case and the numbers will be very different.

Even if you are going to assume the target is relatively stationary, you only get 4.7 if traited with “Writ of Persistence” otherwise the number is 4.2. While it’s a valid point to say “Hammer should take Writ of Persistence so the numbers should include this” you should also take into consideration the traits Greatsword should take – otherwise your numbers are an unfair comparison.

Looks like I may have misunderstood what your “Total Average” was and that’s why I asked you what they were! You are saying you took the coefficient for all abilities that do damage and averaged them? You seem to agree that the average coefficient per swing is not very useful, how is an average coefficient of all your abilties useful? Have you weighted the average to take into consideration activation or cooldown time? How can you use the average coefficient of all abilities, without taking into account the activation or cooldown times of skills, to determine which weapon does more damage?

On to your “statistical” analysis of the coefficients, you say that:

“The point is to say whether or not the difference in damage between 2 weapons is likely to deviate from one another significantly by chance alone.”

I’m not sure what you mean here, how can the “difference in damage deviate from one another”. The damage of GS and Hammer may deviate from one another (that amount would be the “difference in damage”). Also, why would I base my choice on if my damage will be significantly more based on “chance alone”? Wouldn’t I want to pick my weapon so that my damage is more based on weapon choice?

I’m not saying a test of significance isn’t a useful tool but I think you are applyingit incorrectly. In your example (and I disagree with the numbers used but we’ll use your example) you say that:

“Comparing the means of Hammer and Greatsword (1 tailed @ .05 ) yields a p-value of p = 0.4174. This is not a significant difference. What that means is there is no real statistically significant difference in choosing Hammer over Greatsword based on these numbers”

You are comparing 1.375 to 1.275, that’s an increase of 7.8%

As you say

“You could argue that even 0.0009% increase in DPS is totally worth swapping weapons. Is that wrong? Absolutely not. But the info should be out there”

The info should be out there, you should report the dps increase as 7.8% and let the reader decide if that’s a significant difference, not that “there is no statistical difference in damage done”.

Are you sure it’s Pure of Voice removing the condition and not your 6/6 Soldier’s Runes?

Also, is it possible it’s moving all the conditions to you, then removing 1 condition from you?

Generally if everyone has 1 condition it will be the same one so if they all moved onto you into 1 stack, and then removed by either 6/6 soldier or PoV wouldn’t that appear the same?

I guess one way to know would be to see if the affected allies got a boon from it?

Some issues with your data analysis.

First of all, the total coefficient of Hammer is 4.2 assuming the target stands in your symbol for the full 3 ticks, it’s only 4.7 if you take Writ of Persistence trait.

Secondly, you quote the total as 4.7 divided between 3 attacks, to get an average per swing. That gives 1.5667, not 1.375.

Thirdly, I don’t think the average coefficient per swing is terribly useful, it would be much more useful to take the coefficient of the chain and divide by the time taken to complete the chain so 4.2 (untraited, assuming a relatively stationary boss) divided by 3.7 (time to complete the chain) so 1.1351/second, Greatsword auto gives us 2.8 coefficient divided by 2.4 seconds so 1.1667

Lastly, I’m not familiar with the “t-test for significance” but it seems to be being applied incorrectly here. This is not a set of statistics that you can run a test for significance on to determine if one data point falls sufficiently far outside the expected range to question the validity of the set. This is just a number, 1.1667 is 2.75% larger than 1.1351 – so over time you will do 2.75% more damage – whether a 2.75% damage increase is significant is determined purely by the player, not by a formula.

Not trying to be a jerk, but saying that you ran a formula and it told you that the dps difference between two weapons is insignificant is wholly inaccurate.

Balerathon.2184:

Kaptain O.1348:

What do you mean by this:

Balerathon.2184:

Hammer

Auto Attack Cumulative = 4.7
Total Average = 1.375

Well, the first 3 attacks are really just the chain of auto attacks, so you hit once, then it’s a different attack, you hit the second time and it’s yet another attack. Three hits to complete a cycle. The cumulative for all 3 attacks is 4.7.

If you calculate the average coefficient for all attacks per weapon, and divide by the number of attacks, you get an average coefficient per weapon.

Then a simple t-test of significance and Bob’s your uncle!

That’s it really

What do you mean by this:

Balerathon.2184:

Hammer

Auto Attack Cumulative = 4.7
Total Average = 1.375

I know that, that’s not what I’m saying. I’m saying the Attunement Swap Sigils share the internal CD with the % to happen on Crit Sigils. When either trigger it creates an internal cooldown period for BOTH sigils.

What this means is if You have Sigil of Battle and Sigil of Fire, switch attunements you get 3 stacks of might and both Sigils go on cd for 9 seconds. If after the 9 seconds are up you crit and Sigil of Fire Procs before you swap attunements they will both go on CD again.

Since DaPhoenix is running Battle/Corruption primary and only switches to Fire after Corruption caps at 25 I would assume he wants the Might stacks from Sigil of Battle more than he wants the AoE Fire damage from Sigil of Fire.

I don’t think Sigil of Earth has an internal cd (but I haven’t tested it personally) so I think it would continue to proc while Sigil of Battle was on CD and it wouldnt place a CD on Sigil of Battle when it does proc.

If that doesn’t work, or if Sigil of Earth is deemed too weak I think that another Sigil that doesn’t share internal cd with Battle is best, like the +5% damage, or +5% critical chance sigils.

Daphoenix:

I’m pretty sure that Sigils that trigger on weapon/attunement swap share the internal cooldown with weapons that trigger on crits. If you use Sigil of Flame or Lifesteal it will not proc while your Sigil of Battle is on cd from swapping, and if it does proc your sigil of battle will not be able to proc until the cd is up.

Another idea would be sigil of earth, since it doesn’t have an internal cd I think it would still proc while sigil of battle is on cd, and I don’t think it would interfere with sigil of battle proccing.

I have been using the +5% crit sigil though.