Druid Changes Patch 3.1 - Page 6

Allev · 02/06/09, 10:01 AM

Originally Posted by Monfalaris

People say, that SD coming along with lower armor modifiers on SotF will be an overall nerf to bear aoe tanking. Let's take a look at this (more napkin math inc):

The difference in dmg reduction is roughly 8% between bears (70%) and warriors or paladins (62%) when it comes to armor mitigation (not sure about dks at the moment). The mob taken from the ThinkTank example deals 10k with a swing timer of 1.5sec. I can't think of an example, where actually 10 mobs need to be tanked at once (maybe utgarde pinnacle or maexxna?), so let's just take this one. We leave out any avoidance, so keep in mind, that 10 hits is already the worst case when tanking 10 mobs at once.

According to the ThinkTank this can be expressed as 1 mob hitting every .15sec. for 10k. With a shield proc occuring every 1.2sec every 8th hit is mitigated by SD's shield. With current numbers, the shield will absorb 1250 dmg. So the mitigation we gain from SD is 1250dmg/1.2sec.
How much do we lose from the loss of armor mitigation?
Blizzard thinks about bringing bear armor in line with the armor other tanks have. So, regardless of the actual numbers, this would require a loss of about 8% mitigation. So lets see what happens, if the lower SotF modifiers result in a 8% mitigation loss (note that at this point I ignore the fact, that this might not even be possible, even if SotF would not give any bonus armor. We will see later on).
Every hit would be mitigated by an additional 8% which is 800 dmg per hit. In 1.2sec. we get 8 hits, so we gain 6400dmg/1.2sec. This is in fact better than SD, but it is also absolutely unrealistic, because we ignored avoidance. Considering avoidance and misses, the benefit of 8% more mitigation from armor is reduced to approximately 3500dmg/1.2sec effective dmg reduction.
So far, armor is still far ahead of SD and this is why people keep saying, that SD is bad for aoe tanking.

What a nerf of SotF is needed to achieve a loss of 8% armor mitigation?
Again, using the ThinkTank numbers 29866 armor results in 64% mitigation. 56% mitigation is reached having 21k armor. This means Blizzard had to reduce the SotF bonus to something like 5/10/15 which is highly unrealistic since this would bring us way below other tanks. It is realistic to drop us to about 25k (60%) which means something like 13/27/40. This is in fact only 4% loss in armor mitigation so the above aoe tanking example would cut the effective dmg reduction to 1/2. 4-5 hits during a 1.2sec. period hitting for 400 less dmg means 1600-2000dmg/1.2sec.

In conclusion, bringing us exactly in line with other tanks armor wise, SD will be a slight nerf when tanking 10(!) mobs at once. We reach a balanced state at roughly 3 hits taken per 1.2sec. (1200 more dmg taken because of 4% less armor mitigation and 1250dmg less dmg taken due to SD's absorb). This is the case while tanking 6 mobs at once (2.88 hits per 1.2sec). So, whenever we tank a total number of 6 mobs or less at once we actually benefit from Savage Defense, even if the bonus of SotF is reduced to bring us exactly in line with warriors and paladins. In other words, unless Blizzard reduces the SotF bonus below 40% at 3/3, SD will be a buff everytime we tank 6 or less mobs at once.

Note: Actually I had this all done earlier, but I didn't have the time to post the results. Then, kalbear came up with his ThinkTank entry and so I decided to change my calculations in a way to reflect his numbers and to prevent reinventing the wheel.

At least 3 mistakes:

1) It's not 8% mitigation, it ends up being around 4k armor and 3.7% mitigation, as per my math on the previous page.

2) I have closer to 7k AP, which would make the shield 1750, not 1250.

3) If you have 10 mobs hitting you for 10k every second, you're going to die-- they're draining your HP pool twice a second. Even accounting for avoidance. No amount of armor will save you. It does raise the point that it depends on how hard the mobs are hitting you, but realistically, you shouldn't be tanking a group of mobs that can hit you for 1/5th of your max possible health. The ratio is closer to 1/10th (if not 1/20th), which means armor is helping you half as much. Making them 5k hits slice in half your armor mitigation value, which puts it identically in line with 2), if you don't take into account the fact your armor mitigation formula is way, way wrong.

I continue to assert that a full SotF nerf won't be a negative on AOE tanking.

Allev · 02/06/09, 10:08 AM

Originally Posted by nightcrowler

I'll try out some math and comparison:

...

Now let's consider 6 target tanking:

We will do 6 attack each 1.5 seconds and 2 attacks each 1.85 seconds = 68 attacks every 15 seconds. Consider for example a swipe with 6 target we have a 90% chance to crit. Just make it 100% for simplicity. We will have our shield up every 1.5 seconds. But we will recive an attack every 0.25 seconds so we will have a 16% uptime. If we want to consider mauling for extra shiled the chance can go up a little to 20-25% maybe.

On this example you forget avoidance, which makes it an attack every .5 seconds, so 32%+ uptime at a minimum. Plus with an expected 3 hits per GCD, you have to start looking at interleaving swings, i.e. your shield might stay up in between swipes and mauls.

Rannasha · 02/06/09, 10:44 AM

I've created a simple C-program to simulate the effects of different stats on Savage Defense to get a first glimpse at how it might scale.

Assumptions on the mechanics of S.D.
Every critical hit by the Druid procs the S.D.-shield (no internal cooldown). This shield has an infinite duration (finite duration would mostly show up in cases where the Druids base avoidance is high, reducing the effectiveness of the shield). The shield only has 1 stack, critting while the shield is still up has no added effect. Once the boss swings at the Druid and the blow is not avoided, the shield is removed.

The program
The program loops over 3 different events: auto-attacks (every 2.5 seconds, possibly reduced by haste), special attacks (every global cooldown, for simplicity i've not counted debuff-renewal such as D.Roar in this 'rotation') and boss-swings (assuming a standard pattern of just auto-attacks using the 2 second base-attack speed for NPCs reduced to 2.4 by IW/TC/whatever). I'm assuming all auto-attacks to be Mauls and all attacks therefore use the 2-roll system for yellow-damage attacks.

Hits, misses and crits are determined by a simple PRNG. Every time the simulated player crits, the SD-flag is set to one and every time the boss swings, the total number of hits is increased by one, if the SD-flag is one, it's set to zero and the total number of 'blocked' hits is increased by one.

Testing methodology
I'm assuming the following 'base stats':
8% chance to miss an attack (dodge + parry + miss, note that parry-haste is not modeled).
35% chance to crit.
45% total avoidance.
2.5 second auto-attack speed
1.5 second special-attack speed
2.4 second boss-attack speed

At the start of each simulation, the 3 swing-timers (auto, special, boss) are set to a random value between 0 and the attack speed. Each simulation lasts for 24 million 'ingame' seconds, or 10 million boss-swings. The percentage of swings that were blocked by Savage Defense is returned.

To obtain results for varying stat-levels, i left 3 of the 4 parameters fixed as denoted above (i assume the GCD and the boss-attack-speed to be fixed and did not include them in the set of parameters to investigate). The remaining parameters was then varied in, what i assume to be, a range of values that you might see ingame, the range being split up in 40 equal steps:
Miss-chance between 0% and 20%. (Note that 'miss' can mean regular misses, dodges or parries!)
Crit-chance between 25% and 45%.
Avoidance between 20% and 60%.
Haste between 0% and 40%. (Equivalent with an attackspeed between ~1.79 and 2.5)

This gives 4 data-sets of "block-chance" versus parameter-value. I repeated this procedure 5 times (to obtain 6 versions of each data-set).

Results
Block-chance versus miss-chance

Block-chance versus avoidance

Block-chance versus haste-percentage (attackspeed = 2.5 / (1 + haste% / 100)

Block-chance versus critical strike chance

Discussion
There are not many surprising discoveries. Haste is very weak when it comes to scaling Savage Defense. Going from 0 to 40% haste only increases the block-chance by 2%. The avoidance-graph may look odd, but keep in mind that 'block-chance' is defined as the percentage of total hits (as opposed to unavoided hits) that are absorbed (partially) by S.D. The hit graph looks strange, at first glance, for a samplesize of 60 million swings. Looking at the 6 individual data-sets that make up this superset reveals clues to the reason:

Block-chance versus miss-chance (6 data-sets superimposed)

Since attack-speeds remain constant throughout the simulation, the relative timing of player attacks with respect to boss swings is repeated every 60 seconds (after 25 boss-swings, 24 auto-attacks and 40 special attacks). The initial timing of these swings is randomized and depending on the outcome, different 'resonances' might arise that affect the outcome . For example: a situation where more player-attacks happen right before a boss-swing than right after is favourable over one where this is reversed. And if the initial timings happen to land in such a situation, it will remain this way for the remainder of the 10M-swings data-set. Here more data-sets of smaller size would have been prudent.

A similar effect is visible in the other graphs, most notably in the haste-results, where there is a point with very high deviation from the average exactly when the auto-attack speed drops to 2.0.

In realistic combat, parry-haste on the boss combined with non-perfect timing and latency effects will destroy these resonance-effects, which means that the approach of averaging out the data should be valid.

nightcrowler · 02/06/09, 11:16 AM

Thanks for the graph they are very interesting. The avoidance one is something I don't understand, suppose you have 99.999% of avoidance, basically the rare time you were hitted you should be sure that you have the shield up. So increasing avoidance should increase blocked hit. What i think is that you are not graphing "not-avoided-blocked hit" but total hit blocked so more avoidance = less blocked hit. But you should graph blocked hit / taken hit not vs. total hit.

Edit: yes, re-reading your post, I saw that you said that. Can you post a plot with ( blocked / taken hit) vs. avoidance?

Why a shield?

When they first announced the change in SoF I have proposed an AP/crit to armor conversion. My proposed solution was simplier than giving us a shield... but... they gave us a shield for a reason. Many times they stated that they were worried about shield block scaling in later tier contents and that warr/pala will be too much better than druids/dk.

So my question is: What kind of fights make shields better than armor?

Boevis · 02/06/09, 12:10 PM

Originally Posted by nightcrowler

Why a shield?

When they first announced the change in SoF I have proposed an AP/crit to armor conversion. My proposed solution was simplier than giving us a shield... but... they gave us a shield for a reason. Many times they stated that they were worried about shield block scaling in later tier contents and that warr/pala will be too much better than druids/dk.

So my question is: What kind of fights make shields better than armor?

The first fight that pops into my mind for a shield's use is Illidan flame tanking and Kil'jaeden. The chance for the shield not being used before your next attack is next to nil, and armor is useless.

Personally I don't like it, Block still has a major advantage in multi-mob situations, the shield will be useless on the difficult parts of Sartharion as he'll be immune or we won't be attacking due to Torment, even on Sapphiron while the shield will always get used even if you crit multiple times before he hits you it won't be fully effective (for many druids) because it fades after getting hit once even if the hit was weaker than the shield.

It's also going to be very hard to balance, which I suppose is why we're seeing it now instead of added with Ulduar, attack speed and damage varies wildly between bosses. The difference in usefulness of this shield between pathetically weak bosses like Heigan or Anub where the challenge comes from other areas versus Patchwerk or Grand Widow is significant. I'm worried they'll treat it like block, but it's far weaker because of situations already mentioned.

Nathariel · 02/06/09, 12:27 PM

3.1 is when both the changes mentioned and Ulduar are meant to be released. Although I am sure that these changes will get more available time to test than the encounters in Ulduar.

Any fight where the damage is magical rather than physical will make the shield better than additional armor, assuming that we are able to keep the buff up. Also the shield becomes relatively less powerful the more incoming dps there is due to its fixed nature.

I like the concept of doing something to keep up my mitigation, although my opinion might change if there is more spells like Twilight Torment.

kalbear · 02/06/09, 12:47 PM

Personally I don't like it, Block still has a major advantage in multi-mob situations, the shield will be useless on the difficult parts of Sartharion as he'll be immune or we won't be attacking due to Torment, even on Sapphiron while the shield will always get used even if you crit multiple times before he hits you it won't be fully effective (for many druids) because it fades after getting hit once even if the hit was weaker than the shield.

This is all true. However, it doesn't make it useless. On Sapphiron, for instance, even if you overshield on the tick it's better than armor or block would have been; you wouldn't have mitigated that damage in either case. On Sartharion, armor would have slightly helped you with his harder physical hits but wouldn't have helped with the magic parts anyway.

And on other fights it's great.

It's ideal when facing a single mob who swings very slowly and hits you, after mitigation, for the full damage of the shield. That would be the perfect fight. Barring that, it is much better than armor against magic damage, against about 5 mobs that hit for 8-15k each before mitigation, and single mobs that hit fairly lightly. That's where it shines, and where no amount of nerfing armor would realistically hurt you.

Monfalaris · 02/06/09, 12:47 PM

Originally Posted by Allev

At least 3 mistakes:

1) It's not 8% mitigation, it ends up being around 4k armor and 3.7% mitigation, as per my math on the previous page.

2) I have closer to 7k AP, which would make the shield 1750, not 1250.

3) If you have 10 mobs hitting you for 10k every second, you're going to die-- they're draining your HP pool twice a second. Even accounting for avoidance. No amount of armor will save you. It does raise the point that it depends on how hard the mobs are hitting you, but realistically, you shouldn't be tanking a group of mobs that can hit you for 1/5th of your max possible health. The ratio is closer to 1/10th (if not 1/20th), which means armor is helping you half as much. Making them 5k hits slice in half your armor mitigation value, which puts it identically in line with 2), if you don't take into account the fact your armor mitigation formula is way, way wrong.

I continue to assert that a full SotF nerf won't be a negative on AOE tanking.

Well, 1) and 2) are really depending on the gear level. For comparison I used the ingame numbers I recieved by comparing me to a warrior and a paladin of equal gear level. Accordingly, the armor mitigation depends on the armor level you are looking at. I took the character sheet numbers and got something very close to a 8% difference in both cases. Additionally, many druids stated that they have an AP value roughly around 6k not 7k. You could probably guess an absorb value of 1.5k if that makes you happier or just go with your personalized 1750.

3) is right though. If 10 Mobs are hitting you for 10k each .15 sec. you are dead. That is why I never looked at the actual value hurting you, but only at the difference in armor mitigation. The case could even be, that the warrior mitigated 9k of 10k while the bear reached 9.8k, resulting in a difference of 8% mitigation, which can be quite misleading, because it is 8% in regards to the hit value. You only need to know what the unmitigated hit would have been. 8% mitigation on a 10k hit is always 800, no matter how much damage you take in total.

As for the mitigation formular I was referreing to wowwiki. Maybe this one is old and I did not even attempt to proof it myself, which might lead to errors. Furthermore, I really didn't want to post something like 63.866537728901% so I rather used 64% since we all love round numbers and in this case exact numbers really didn't matter.

Finally, this really might not be the ultimate truth, since I worked on it with maybe 30% attention as I had to do other stuff

. Then I even reworked the numbers to match them in the ThinkTank. Still I do not see 'mistakes' you mentioned.

Why a shield?

When they first announced the change in SoF I have proposed an AP/crit to armor conversion. My proposed solution was simplier than giving us a shield... but... they gave us a shield for a reason. Many times they stated that they were worried about shield block scaling in later tier contents and that warr/pala will be too much better than druids/dk.

So my question is: What kind of fights make shields better than armor?

Obviously, being attacked for about the absorb value every 2 sec. would be perfect for a shield mechanic. So I really think the upcoming encounters will include mobs with about that attack speed not hitting too hard, because this would really destroy any shield absorb mechanics, which aren't dependend on the attack value. The same goes for magical damage. Something like Sapphiron's Frost Aura would also be really good for a SD mechanic. Just thoughts though.

Pater · 02/06/09, 1:16 PM

The stacking question is very important. When I first read the announcement, I assumed 1 stack only. Is there any reason to believe otherwise?

mydhrin · 02/06/09, 1:25 PM

Originally Posted by Pater

The stacking question is very important. When I first read the announcement, I assumed 1 stack only. Is there any reason to believe otherwise?

I really doubt that it's going to be more than 1 stack. With the amount of crit rate a druid can reach and the attack speed, having 2 or 3 stacks could mean that we have a constant 1.5k shield on us at all time which is straight out OP. Besides, with the fact that any attack takes off the shield, I think it strong suggest that it is only 1 stack.

Allev · 02/06/09, 3:19 PM

Originally Posted by Monfalaris

I took the character sheet numbers and got something very close to a 8% difference in both cases.

...

3) is right though. If 10 Mobs are hitting you for 10k each .15 sec. you are dead. That is why I never looked at the actual value hurting you, but only at the difference in armor mitigation. The case could even be, that the warrior mitigated 9k of 10k while the bear reached 9.8k, resulting in a difference of 8% mitigation, which can be quite misleading, because it is 8% in regards to the hit value. You only need to know what the unmitigated hit would have been. 8% mitigation on a 10k hit is always 800, no matter how much damage you take in total.

As for part 1-- you're right, I got it wrong. Without checking the mechanics, I assumed SotF was an additive rather than multiplicative bonus on armor. While I haven't been able to pop in-game, I did just check with Rawr, and it pretty clearly looks to be multiplicative (and I assume Rawr has it right). This explains why I missed you were nerfing to be relative to warriors. This throws off all my math on the subject to this point, but especially yours since you're talking 8% character-sheet, and not 8% of the remaining damage coming in. Which is odd for a damage shield, since it mitigates after the other calculations have taken place (i.e., armor mitigates before the damage shield).

As for part 3-- I should've read more carefully to see what you did there, I assumed your mitigation values and calculated values were post-hit, since that's when damage shields are applied.

The point stands that the 10k number is arbitrary and vital to the percent that a fixed number mitigates. If a shield is able to mitigate a full hit (think Hyjal paladin blocking), then you're dealing with a higher percentage of damage than when a hit is three times more powerful than your block. Specifically, if you took your example and merged it all into one enemy, we're talking 100k unmitigated every 1.2 seconds. Malygos hits for less than half that.

Monfalaris · 02/06/09, 4:53 PM

Originally Posted by Allev

The point stands that the 10k number is arbitrary and vital to the percent that a fixed number mitigates. If a shield is able to mitigate a full hit (think Hyjal paladin blocking), then you're dealing with a higher percentage of damage than when a hit is three times more powerful than your block. Specifically, if you took your example and merged it all into one enemy, we're talking 100k unmitigated every 1.2 seconds. Malygos hits for less than half that.

Yes definately the number ist arbitrary. Actually I chose that number only in correlation to kalbear's ThinkTank entry. Using the (really high) numbers nightcrowler posted, that 100k unmitigated and unavoided dmg would be reduced. 50% Avoidance cuts this in half as he has 50%. Armor mitigation would reduce these remaining hits by roughly 69% (36.5k armor). This results in 15.5k every 1.2 sec. Yeah, you are right, this is still a little high, but this is actually intentional. The first example used a total of 10 mobs hitting, which was stated as never happening. I wanted to see the worst case for SD and how big the difference is, hoping that, even in this case, it is somewhat close, which it wasn't. Then I started to modify the situation to be much more realistic.

All in all, it helped me being able to actually judge the shield mechanic and it's effectiveness compared to the current situation and I tried to predict the changes to SotF using a different approach but getting very similar results compared to kalbear, who posted his results earlier.

Odas · 02/06/09, 5:45 PM

I recently made a model for calculating the effective armor benefit of Savage Defense and for finding out the new relative tanking values for attack power and crit rating. I wanted to see if any of you could see any flaws in my thought processes. This is all based on a single mob, although it is easily generalized by decreasing the time between player swings and boss swings.

If I calculate what the chance is that the nth swing the boss does lands and all previous ones miss I can figure out how much time would be available to generate a shield.

A. chance the nth swing hits and all previous ones miss = bossHitChance*(1-bossHitChance)^(n-1)

B. number of swings available to apply a shield = Floor((n*bossSwingTime)/(playerSwingTime))

C. chance that the shield is up when the boss hits = 1-(1-playerCritChance)^(B)

D. chance that the boss hits on the nth swing and the shield is up = A*C

E. chance that the shield is up when you are hit = sum from 1 to infinity(A*C)

Once you have the % chance the shield (%absorb) is up the next time you are hit, you can calculate the armor you would need so you could take the same damage before Savage Defense is implemented and after.

bossDamage*(1-DR) = bossDamage*(1-newDR)-%absorb*attackPower/4

Using my raid buffed values, I calculated what my new armor should be, and what the SOTF multiplier would be.

Stats:

bossHit% = .5
playerCrit% = .32
bossSwingTime= 2.4
bossRawDamage = 25000 (about 8k after mitigation)
playerSwingTime = 0.85 (assumes 2s maul swing and 1.5s specials)
durationOfSD = 6
attackPower = 6237
bearArmor = 37694
agility = 1519

Results:

newArmor = 30849
SOTF multiplier = 1.33
Effective base armor from 1 attack power = 0.11
Effective base armor from 1 crit rating = 0.24

Any thoughts?

kalbear · 02/06/09, 7:26 PM

That looks about right, save that boss raw damage is pretty low. Sarth hits for about 40-50k before armor, for instance, and Ulduar will have much higher. Patch hits for 80k and has a swing speed of 1 second. But yeah, for those low levels of damage that armor reduction looks just about correct.

smellme · 02/06/09, 8:51 PM

I have to question how they are going to play out DoT's with this to be honest, while alot more common in pvp there are some in pve encounters now and again(such as mindflay, rend effects and so on).

Would a rend that ticks for 12 damage remove this shield for example?

I'm hoping the not anyway.