Guild Wars 2

Here is a spreadsheet I was working on that I use to calculate mesmer DPS in various environments. At this moment I wouldn’t consider it final because I still need someone to check it over to make sure everything is actually calculated properly.

Spreadsheet (please download to edit, atm its read-only)

• https://docs.google.com/spreadsheets/d/1Yt0WNbgZkbn1W0Y89v7IatIujauTTQ3XbU1iTqRd7qc

Spreadsheet has multiple tabs but all user input is limited to the “Variables” tab only. Enter your data/options in yellow cells only. Calculated results will show up in the green cells. There is no need to edit anything in the other tabs unless you want to force override something.

How cool! This is excellent, and super clear and easy to use. I’ll start validating insofar as I can. Thanks for this excellent resource!

This is so cool. I’ve been playing with it for a little while now. Very cool how you can test/play with almost every single trait and configuration!

One question though— I haven’t seen where we input armor choices. Are you assuming full exotic zerk for the equipment?

No assumptions are made, there is no real need to make any. Stats with 0/0/0/0/0 should cover that. You can use gw2skills editor to equip whatever gear you want (weapons, armor, trinkets, and runes) and copy/paste the base stats you get with that combo into spreadsheet.

The reason on spreadsheet you can choose different weapon quality is because different weapon tiers have different strengths (and stats too so adjust 0/0/0/0/0 accordingly) and that in turn also affects your damage output.

frifox.5283:

No assumptions are made, there is no real need to make any. Stats with 0/0/0/0/0 should cover that. You can use gw2skills editor to equip whatever gear you want (weapons, armor, trinkets, and runes) and copy/paste the base stats you get with that combo into spreadsheet.

The reason on spreadsheet you can choose different weapon quality is because different weapon tiers have different strengths (and stats too so adjust 0/0/0/0/0 accordingly) and that in turn also affects your damage output.

I see. Thanks!

This is very useful, thank you kindly frifox once again for your great contribution to the mesmer community!

In the expected DPS column, what trait dps setup are you using to meassure current dps vs previous?

Whatever you have spreadsheet use in the calculations produces the results in green box, which then in turn is compared with whatever results you entered in the “current vs this” section.

This is quite decent

If you are going to work on actual DPS calculators you may want to collaborate with me and DEKeyz: we have done it for elementalists and we could expand it to other classes like the mesmer with your help.

I’ll take a note of that. For now I’ll be limiting myself to just the mesmer since it’s the class I know about the most and who’s dps output has probably the most variables compared to other classes also set up for full dps.

I do have an ele question though. I want to add a Lightning Hammer for cases where mesmer would be using that instead of the sword AA but I can’t find any skill coefficients for the lh aa chain. Do you have any that are valid post patch? And also, if mesmer equips the weapon, what weapon strength would I use. Sword MH? Sword OH/Focus? Or some other predefined weapon strength?

I cannot say for sure but it should use the weapon strength of those you are wielding at the moment when you pick up the hammer.

Just something odd coming from a non-mesmer: I don’t see blurred frenzy mentioned in your spreadsheet, is that normal?

I’ve always had a bit of trouble computing mesmer dps for some reasons: it is incorrect to assume phantasms will be up and running from the beginning of the fight until the end. In my calculator I made a rotation input, which allows to “design” the fight as you would do it in game (with some limitations). This would better reflect the time required to spawn the phantasms. But then it is hard to include damage sources other than that of the player. That’s why it would be interesting if you could take a look at it and let me know what you think.

My calculator was designed to support all classes, but I never had enough data for other classes than ele. So I would need help from people who play those classes to finish it

frifox’s calculator doesn’t support rotations like yours, Zelhyn— it just tells you a DPS for a build + combination of phantasms.

This is good enough for mesmers… we basically only have one good weapon (mainhand sword), and cast time for all the phantasms is identical. The only thing that matters when comparing mesmer builds is just the DPS of the traits + gear + phantasm combinations, since summon times, etc. are constant across all variations. This is of course very different in ele, where there are tons of different weaponsets you need to compare and therefore you need the full rotations to have any idea about comparisons. Mesmers get away with something much more like Effective Power because there is a little less variety in what you can do.

Anyway, a full mesmer rotation is kind of annoying because it is very situational. Phantasms die and need re-summoning, but only if the fight is going to go on for a while longer, etc. etc. This is why knowing the personal damage (which you can always rely on) vs phantasm damage (which can be sporadic/unreliable) is so nice.

Finally, afaik Blurred Frenzy is a minor DPS loss (and you don’t get the boon strip + vuln from the auto-attack). But if it saves you a dodge, it’s a DPS gain, of course.

EDIT: One final thought— you bring up exactly why it’s hard to compare mesmer DPS to other classes. The question of “how long does it take to summon your phantasms” and “how reliably will they stay up” is a big part of deciding how mesmer DPS compares to others. But this is hugely situational, so figuring out the exact timing with a lot of assumptions may not be very helpful.

Yeah, with mesmers the direct damage dealers are very few – sword mainhand AA and whatever phantasm you have out. Everything else is too dynamic to properly predict. The skill/utility rotation most of the time are reactive to whatever enemy is doing. It gets even more complicated once you start considering reflects. Also, it’s quite hard to predict how many phantasms you will have alive and how often you will have to re-summon them because that depends not only on you but also your party members and the enemy AI. Builds also change depending on the dungeon, path, or even boss. Most of things above also heavily depend on the player skill, which is quite hard to quantify/measure. To top if off, there is quite a lot of RNG involved with things just bugging out and flat out just not working. This is why I figured it would make more sense to keep it simple as is and let you decide which build will work the best for him.

Btw, where can I get the calc you mentioned? I’ll have to check it out.

Here is the calc:
https://drive.google.com/folderview?id=0B88y8fHnpBfZRnJRdVNqVC1MYXc&usp=sharing

It’s still in its pre-patch version. I haven’t had the courage to update it since my main account got perma-banned from the forums and I got more into PvP
Applying the patch changes would be easy though.
Initially I designed it so that it would support all classes. However I had very limited knowledge on other classes so I was hoping to collaborate with people interested in developing it for their own class. Let me know if you are interested, maybe we can work something out

Ok, after almost a whole week of testing and re-testing everything I can finally say I’ve gotten it as accurate as you can get. Simulated many builds / setups with spreadsheet and then verified the actual DPS with the help of my dps meter and all results were within 1% of the predicted. Guess that’s good enough

Notes:

• Phantasms ignore your weapons (white or legendary, doesn’t matter as long as your core stats are same) and use Exotic weapon strengths instead
• Warden uses Axe stats, not Focus
• Tooltips / Indestructible Golem are both assuming 2,560 armor and not 2,600.
• Mind Spike (3rd hit of sword AA) does 10% less dmg than what tooltip claims and about 4% less than what wiki has (coefficients). This is most likely because Mind Spike actually uses WARHORN weapon strength instead of Sword for combat… gg anet.

Also a related conclusion I came to while doing the tests – using Blurred Frenzy does not lower your dps. In fact, if you use it without breaking the sword AA chain you will gain +12% dps for the following 4.3 seconds and + 14% dps for 3.4 seconds if you actually do end up breaking the chain (aka BF after only 2 AA’s). Assuming you have your sword traited that should average to about ~6% personal dps gain if you use BF on cooldown. The dps gain goes even higher if enemy has boons (thus proc’ing the weaker mind spike)

ps: all coefficients were tested in hotm so unless there is an unknown skill nerf/buff for pve counterpart then everything should still be valid in pve as well.

You mean like a certain studio once infracted me for posting about my email being banned by them and in their answer they told me to email them if I wanted to discuss such things? Game, set, and match.

How did you find out about the weapon strength used for mind spike?

Edit: do you know if phanstasms scale with weapon quality? it does for conjured weapons so maybe it works for all utilities? I found that utilities used the inherent strength of an exotic sword in pvp, but in pve it depends on the quality of your equipped weapon.
Last time I checked, if we assumed that tooltips use 2600 armor then the undestructible golem would be around 2660. I guess we need to cross check our results on that.

In PvE I equipped Legendary offhand sword (94 power) with no mainhand and recorded 1,000 non-crits against a target dummy. Then I did the same but with a white lvl 80 offhand sword (54 power) and a Cherry Tart (40 power) to match the legendary’s power stat. Minimum, maximum, and average hit were identical. Heck, I even unequipped the whole weapon during the skill cast and number stayed the same as well. Kind makes sense since otherwise Disenchanter/Defender would have no weapon to inherit the stats from since they are summoned by a utility, not a weapon skill.

Figuring out weapon strength took some math. The general idea was to derive a coefficient from the tooltip data. Then, record 500 to 1,000 non-crits and then use that coefficient to figure out the weapon strength. The equations:

• d = p * w * c * m / a
• c = (a * d) / (p * w * m)
• w = (a * d) / (p * m * c)
  d = damage; p = power; w = weapon strength; c = coefficient; m = multipliers; a = armor;

Taking Swordsman as an example. At 2,213 power (soldier amu, str runes, 4pts into domination) using the tooltip’s damage number we figure out the coefficient (1.15 multiplier was from Empowered Illusions, because otherwise the tooltip is bugged… but that’s a completely different topic):

• 1.895 = (2,560 * 1,794) / (2,213 * average(905 – 1,000) * 1.15)

Then, do ~1,000 hits and take the lowest and the highest non-crits to figure out the weapon strengths

• 905 = (2,560 * 1,705) / (2,213 * 1.15 * 1.895)
• 1,000 = (2,560 * 1,883) / (2,213 * 1.15 * 1.895)

Getting exactly 905 – 1,000 (100% match to an exo sword) after the real test means the tooltip really is accurately representing the proper damage calculations..

Doing the same check for Duelist / Warden, I get these weapon strengths:

• Duelist: 876 – 1,029 (exo pistol is 876 – 1,029, perfect match)
• Warden: 857 – 1,048 (exotic axe is 857 – 1,048, perfect match)

If I assume 2,600 armor for both tooltip and golem, I get:

• Swordsman: 905 – 999 (shoudl be 905 – 1,000… off by 1, acceptable)
• Duelist: 873 – 1,025 (should be 876 – 1,029… off by 3 & 4, not acceptable)
• Warden: 854 – 1,044 (should be 857 – 1,048… off by 3 & 4, not acceptable)

This is why I stuck with 2,560. The coefficients are nice and round and the calculated weapon strengths match the tooltip pretty much to 100%.

Now, for the Sword AA’s part.

Mind Slash/Gash and Mind Spike coefficient from tooltip:

• 0.573 = (2,560 * 472) / (2,213 * average(905-1000) * 1.0)
• 1.145 = (2,560 * 943) / (2,213 * average(905-1000) * 1.0)

And going back to calculate the weapon strengths, Mind Slash/Gash:

• 903 = (2,560 * 452) / (2,213 * 1.01 * .573) // 1% vuln from the skill
• 999 = (2,560 * 500) / (2,213 * 1.01 * .573) // 1% vuln from the skill

and Mind Spike

• 814 = (2,560 * 806) / (2,213 * 1.0 * 1.145)
• 900 = (2,560 * 971) / (2,213 * 1.0 * 1.145)

Mind Slash/Gash is 903 – 999 which off by 2 & 1 vs the intended 905 – 1000 exo sword stat, but I considered that acceptable knowing that vuln on hit may have come into play and messed with the data a bit. Mind Spike on the other hand was completely off (no vuln here to mess up the results). Looking up the exotic weapons I’ve noticed that the 814 – 900 I got is actually weapon strength for an exo warhorn, a 100% match.

If we assume anet really is using 905 – 1,000 for Mind Spike in combat then the tooltip is bugged then the coefficient should actually be:

• 1.031 = (2,560 * average(806 – 891)) / (2,213 * average(905 – 1,000) * 1.0)

I do believe coefficient gotten from the tooltips are “intended” damage numbers because:

• Mind Slash/Gash => Mind Spike (w/ boons) is 50% increase
• Mind Slash/Gash => Mind Spike (no boons) is 100% increase
• Mind Spike (w/ boons) => Mind Spike (no boons) is 33.3% increase

Kinda hard to assume that those nice round numbers are NOT intended and the actual Mind Slash/Gash => Mind Spike (no boons) somewhat random 78% increase is intended.

Hard to say what is intended or not with Anet

Alright, thanks for the display of maths. So the answer is that you assume tooltips are accurate. I have a bit of trouble accepting that

But before we go deeper into this, the only reason why you take 2560 armor for tooltips is because of the roundness numbers? This could be correct, but this figure does not necessarily make numbers round for other professions!

I wouldn’t say he’s assuming they’re accurate: they’re the simplest thing which explains his quite detailed and complete observations. Leave it to ANet to be inconsistent and get ele tooltips wrong, and Mesmer more correct.

Well, there is a 0.06% difference between the relative spreads of the warhorn weapon strength and that of the sword. So for all I know, it could be considered equal. Taking this into account, we could assume that the coefficients are simply wrong. This would hardly be a surprise. Many tooltips are wrong, not just for the ele.

Now maybe if we actually recorded and found differences in the weapon strength distribution of those two weapons then we could come up with more accurate conclusions. I have no idea how frifox has managed to gather such large samples, doing so manually would take ages, so I can only let him refine his findings based on some friendly remarks

I guess that at this point, if we take into account all the uncertainty that arises from having Anet as a developing studio, the proper course of action would be to go very basic: test for weapon coefficients in pvp on a target which armor you can control.

I am growing very tired of how inaccurate every thing is in this game.

Also, maybe if mind spike really used the warhorn weapon damage instead of the sword then you could notice this with progession in http://wiki.guildwars2.com/wiki/Weapon_Master ?

The dps meter I use can log all hits enemy takes to a text file. After that it’s a simple matter of copying the data into a spreadsheet and doing the calculations.

I’d say the observations are as accurate as you can get because after doing same tests over and over, some even days apart from each other, the results are identical every time. I’ve noticed that sometimes even 200 data samples was accurate enough but I still went ahead and recorded 1,000+ data samples, multiple times, before assuming them being accurate.

Darox.8069:

proper course of action would be to go very basic: test for weapon coefficients in pvp on a target which armor you can control..

I was looking for someone on JQ/TC (I’m BG) that would be willing to let me use them as a punching bag but I still haven’t found anyone with the enough patience to commit.

frifox.5283:

Darox.8069:

proper course of action would be to go very basic: test for weapon coefficients in pvp on a target which armor you can control..

I was looking for someone on JQ/TC (I’m BG) that would be willing to let me use them as a punching bag but I still haven’t found anyone with the enough patience to commit.

I would but I’m EU

Can you send me an email ingame to let me know how I can get that dps meter please? ign=Zelyhn.8069

Ive got gw ESP downloaded and working, but it doesnt seem to have the same functionality as some of the tools ive seen frifox use. Is that the dps meter youre referring to here? Are you using the same version as is publicly available, frifox?

Also, back to LH dmg, coefficients, and ascended rarity… anything new since Zel’s last post in that ele thread? Or nothing concrete yet that i can use for updating real-dps for various LH builds?

He said somewhere he is making a completely new version.

frifox.5283:

Darox.8069:

proper course of action would be to go very basic: test for weapon coefficients in pvp on a target which armor you can control..

I was looking for someone on JQ/TC (I’m BG) that would be willing to let me use them as a punching bag but I still haven’t found anyone with the enough patience to commit.

I can give you a hand Fri,

TC here.

what kind of time table would you like?

I also have a guild Arena in PvP that we can use.

message me in game if you would like.

Darox.8069:

Can you send me an email ingame to let me know how I can get that dps meter please? ign=Zelyhn.8069

The old one does work but it’s not very flexible (no dps data vs objects) and you can find it on my github, my username there is same as here – frifox, you can probably find it through google. I’m coding a re-worked version that will be flexible and more accurate (down to microseconds, if needed). Atm I’m in the middle of a big project at work but should be done in the middle of this week and will continue working on the new version. I can send you an “alpha” version if you really want it, otherwise just wait a week or so and I’ll share it.

Ropechef.6192:

I can give you a hand Fri,

TC here.

what kind of time table would you like?

I also have a guild Arena in PvP that we can use.

message me in game if you would like.

Perfect! I’ll contact you ingame once I’m ready. Just want to sort out a few small UI quirks with the dps meter before continuing.

frifox.5283:

Darox.8069:

Can you send me an email ingame to let me know how I can get that dps meter please? ign=Zelyhn.8069

The old one does work but it’s not very flexible (no dps data vs objects) and you can find it on my github, my username there is same as here – frifox, you can probably find it through google. I’m coding a re-worked version that will be flexible and more accurate (down to microseconds, if needed). Atm I’m in the middle of a big project at work but should be done in the middle of this week and will continue working on the new version. I can send you an “alpha” version if you really want it, otherwise just wait a week or so and I’ll share it.

Ropechef.6192:

I can give you a hand Fri,

TC here.

what kind of time table would you like?

I also have a guild Arena in PvP that we can use.

message me in game if you would like.

Perfect! I’ll contact you ingame once I’m ready. Just want to sort out a few small UI quirks with the dps meter before continuing.

sounds like a plan,

and for all of you that may or may nit need it. Zel i am looking at you ,

i have a guild arena in PvP that i have access too so if you want at least numbers on a “player” target we can work something out.

hit me up

(the perks of being a guild leader. )

Nice! This is good to know

I will wait until frifox has finished his new version and then I’ll get to testing.

Big thanks to Ropechef for donating 4 hours of his life to become my punching bag… for science

With enemy Armor known, the only unknowns left are,

• PvE: Phantasm weapon strength
• PvP: Phantasm weapon strength
• PvP: Mesmer weapon strength

I will go ahead and make the following assumptions,

• PvE: Phantasms use exotic weapons
• PvP: Phantasms use exotic weapons
• PvP: Mesmers uses exotic weapons

Making these assumptions should not affect the calculations as long as the respective coefficients match those assumptions. Think of weapon strength and coefficients as 2 parts of a single multiplier, as long as both result in the same multiplier it doesn’t really matter which weapon strength and coefficient you use. They’re proportional.

Now with everything else being known, assuming maxed might/fury/vuln, both banners, same core stats, and exotic weapon qualities for both gamemodes:

• In PvE mesmers will hit 3.13% MORE than in PvP
• In PvE phantasms will hit 2.7% LESS than in PvP

I’m sure ANet has some kind of logic to justify these buffs/nerfs for each game-mode. I won’t bother trying to justify that logic. All I know is that’s how it works.

Finally, after doing another round of tests vs the indestructible golem all calculations point to his Armor being right around 2,600. If I do calculate the same vs the tooltips then the armor used for Sword AA1/AA2/AA3/BF/IR tooltips is 2,595 and 2,600 for S/D/W/B phantasm tooltips. And yea, Mind Spike tooltip is still out of whack. If we decide that tooltip is actually correct then it’s the in-combat damage (ie weapon strength) that is broken.

I have updated the spreadsheet with proper coefficients for both game-modes and added a switch where you can toggle between PvP and PvE. Ran a few tests in PvP mode with dps meter and results were within < 1% of the predicted values. Looks good.

Now, if I keep testing and find out that there is a split between WvW/PvE or even PvP/HotM… I don’t know if I’ll have enough self restrain to keep being civil. I’m already at my limit.

frifox.5283:

I don’t know if I’ll have enough self restrain to keep being civil. I’m already at my limit.

I can imagine

So you confirm that the armor used in tooltips is 2600?

Same class but different areas (mesmer’s vs phantasm’s tooltips) have slightly different armor values. Maybe I’m just splitting hairs because the difference between the two is only about 0.5%. Either way, I do believe 2,600 is a safe bet. In order to to get more accuracy we would actually need to test all classes individually. Also, with Ropechef we limited ourselves to around 100 data samples. Most of the time coefficients stabilized after ~60 samples but I do think in order to get most accurate Armor values and coefficients it’s necessary to do 500 to 1,000 samples.

Feel free to check out the “Coefficients (PvP/WvW)” tab to see how I went about calculating the coefficients for both game modes.

Oh, and one more thing. Tooltips were not always ~2,600 armor. In some cases I ran into tooltips that calculate to 2,490 Armor and sometimes even 2,580 Armor, both in PvP and PvE. I’ll try to figure why exactly that was happening but for now I would suggest avoiding relying exclusively on the tooltips for coefficients. Always need to test and re-test them vs an enemy with known armor.

Wow thanks for publishing all your data and calculations! This is very valuable to me!

Do you have any idea how much Defender bonus Ropechef had in WvW when you did the test? It is supposed to apply to monster hits only, but that could include phantasms, and this could explain the inverse split compared to normal skills.

Good catch. I’m not sure but I believe it should have been right around 5% or 6%, which aligns with the observed shift in coefficients. This actually can be tested quite easily. Do a test for one mesmer skill and one phantasm right before reset and then do the same right after when all bonuses reset to 0.

I’ll see if someone will be online at that time to do the test.

as of this posting.

Defender Bonus for me is at 6 percent.

tool tip does in fact say against monsters. But yes the only way to really test it would be to do it right after reset.

Just don’t be surprised if Defender bonus doesn’t exactly work like you think it should.

On the topic of sample sizes, I tried to do a little check on your data. Here is my thought process. It could be wrong, but it’s worth investigating.

I compare the relative spread ((max-min)/average) of the damage you observed to the relative spread of the expected weapon strength. Then I check if the variation between the two would be enough to justify at least a 1 point difference in observed weapon strength range. In layman terms: “did we miss a min or max value for weapon strength rolls by at least one point in the experiment?”

Input the following formula in G37:
=((G20-G18)/G19)/((G31-G29)/G30)*(G31-G29)-(G31-G29)
(which can obviously be simplified but its just so you know the though process)
The output value is the difference between the observed damage spread translated into weapon strength spread and the expected weapon strength spread.
If its absolute value is higher than 1 then it means that the observed spread is different enough to say that we missed a min or max value by at least 1 point of weapon strength. Anything below an absolute value of 1 should be fine because of rounding down.

What I am not sure about is that the threshold value is indeed 1. It could be more, or less, due to rounding down in the damage calculations. Let me know what you think.
Edit: also I am not sure what is the threshold to justify that coefficients are wrong.
Edit2: I guess the proper threshold would be to use the value computed by my formula as a weapon strength in the damage formula using the computed coefficients to see if the difference is large enough to say that we missed a min or max observed damage value by at least one point. (input =G37*G26*G8/G23*G35 in G38 and =G37*G26*G10/G23*G35 in G39, if any has an absolute value superior to 1 (or 0.5?) then we have a problem)

Anyway, some results are outstanding: -5.6 for swordsman in wvw, +33.8 for warden in wvw, and -4.5 for berzerker in wvw.

This suggests that something really went wrong with the warden.

To verify my data I usually do it the more straight forward way.

• abs(MinCoefficient – MaxCoefficient) / average(MinCoefficient, MaxCoefficient)

Data set resulting in coefficient deviation by more than 1% is not good. Ideally I would prefer to have less than 0.1% deviation. Most of the data I gathered however deviates less than 0.5% (your -5.6 / -4.5). The WvW Warden column was off by 3.57% (your 33.8).

Either way, I found the problem. Once again you caught a mistake in my data. The problem lied in a wrong weapon strength used for the WvW Warden column. Instead of using Axe weap strenghts for spreadsheet I somehow used Pistol weap strengths instead. Probably a copy-paste fail when I was copying formatting (colors, number format, etc) from the Duelist column.

Good news is that this doesn’t affect the calculated final coefficients. While Pistol’s and Axe’s min and max weapon strength values are different, the average weapon strength is exactly 952.5 for both, which is the actual value used for calculations.

@#$%!#….

Looks like the “Defender” wvw bonus DOES apply versus phantasms. Did a round with +7% bonus active for the enemy and the calculated coefficients were -6.70% off from the tooltip. Did another test right after reset and, what do you know, the calculated coefficient were within 0.10% of the ones calculated from the tooltip.

On the bright side that means phantasm coefficients are SAME in both pvp and pve as long as we assume they use exotic weapons in both game modes.

On the not so bright side, mesmer coefficients are still ~3% lower in PvP than in PvE (assuming exotic weapons in pvp). To have coefficients be the same in both game modes, PvP sword weapon strength has to be right around 876 – 968 (excluding Mind Spike, which is even lower), which is somewhere in between Rare and Exotic weapon strength. Funny to notice is that an exotic level 78 sword is 876 – 969. Coincidence? I don’t even know anymore… lol

Finally, probably the best news of all, the tooltips are actually correctly reflecting expected damage in both PvP and PvE (assuming tooltips use a 2,600 armor target). One thing I’ve noticed is that my tested coefficients were consistently 0.001 or 0.002 lower than the ones from tooltips. While this is almost nothing to worry about, I still wonder why that keeps happening.

I haven’t been following your math very closely so forgive me if I say something something.

IIRC, tooltips round the values while actual damage floors. Results in one damage difference in some cases.

Based on my tests, some WvW defender values increase armor by (Defender – 1)/100*Armor. So 7% bonus should result in 1.06 more armor which means 5.7% damage reduction. I have only tested this for values 1%-5% (and 1% worked correctly so 1% Defender = 2% Defender). Based on your test, did 7% work as it should?

Hits gathered for Swordsman at 7% (7% on tooltip, 8% in wvw tab).

Assuming bonus does nothing,
- coefficient is off by -6.7% from the tooltip.

Assuming bonus increases armor (100% doubles armor),
- armor*(1+7%) = coefficient is off by -0.17% from tooltip

Assuming bonus reduces dmg (100% negates all dmg),
- dmg/(1-7%) = coefficient is off by +0.32% from tooltip

I feel like I didn’t gather enough samples (data spread was 9.92% but should have been 10%) for a definitive answer as to how the bonus value is used in calculations, but it’s definitely there. I’ll go even as far as assuming it also works on ranger pets, engi turrets, necro minions, and all the other summons too.

Once I assume that “Defender” simply multiplies enemy armor, then my recorded WvW coefficients for phantasms will match the PvP recorded coefficients to within <0.1%. Pretty kitten consistent if you ask me.

I know it increases armor from testing against Ballistas but good that you also got the same result.

I was just curious whether you had 7% or 8% on WvW tab.

Current findings for weapon strength scaling so far:

Data was gained from characters with exactly correct level (in level 30 dungeon a level 30 character and so on).

Until level 60, doesn’t matter whether you use Ascended, Exotic or Rare weapon as they scale down to same weapon strength. Until level 80, doesn’t matter whether you use Ascended or Exotic as they scale down to the same weapon strength.

Data: (from Ascended to normal, percentiles based on level 80 values)

Level 30:
516 Armor: 20%
Stat scale: 22%

293 293 294 278 261 212
27% 28% 32% 32% 33% 33%

Level 35:
614 Armor: 24%
Stat scale: 28%

362 362 362 343 323 262
33% 35% 39% 40% 40% 41%

Level 45:
885 Armor: 34%
Stat scale: 40%

454 454 454 429 403 328
41% 43% 49% 50% 50% 51%

Level 50:
1047 Armor: 40%
Stat scale: 47%

500 500 500 473 445 362
45% 48% 54% 55% 56% 57%

Level 55: (not sure how reliable as the area downscales to 56, not 55)
1217 Armor: 47%
Stat scale: 55%

541 541 541 511 481 391
49% 52% 58% 59% 60% 61%

Level 60:
1450 Armor: 56%
Stat scale: 62%

684 684 616 582 547 445
62% 65% 66% 67% 68% 70%

Level 70:
1923 Armor: 74%
Stat scale: 80%

821 821 739 698 657 534
75% 78% 80% 81% 82% 84%

Level 79: (some magic guessing)
2487 Armor: 96%
Stat scale: 98%

966 966 870 821 772 629
88% 92% 94% 95% 97% 99%

Nice job Weth! Can you give us an equation now? :p

I wonder what kind of formula can explain THIS:
- http://i.imgur.com/lXAzWXm.png

and THIS (going a lvl higher but tooltip dmg actually goes down):
- https://www.youtube.com/watch?v=d1s4uEvlhCg

#ANetSwag2014