@Mijae: You forgot the flasks.

(Using 1 STA = 17.49 health)

[Flask of Stoneblood] = 37 STA
[Flask of Chromatic Wonder] = 18 STA 35 FR -> 1.84
[Lesser Flask of Resistance] = 50 FR -> 1.35

So the lesser flask is still better than using any of the TBC Fire Resist gear if you want more of it.

--

Also, from my logs I read that I resist >99% of sarths breaths and I think the others are just the paladin dying before me because it was a wipe, and then I'm unlucky and take a full breath because my FR is lower.

I'd argue that resist gear is better for tanks now than it was in TBC. On Illidan you could get very unlucky and take 3-4 full hits in a row (since it was only 0/25/50/100% on the table?). It now seems to be more of a distribution around an expected value, with extremes being more uncommon. This means we can rely more on it, even though it seems less effective per point than before.

I'm curious what the source is on this. I still haven't found any rigorous math on the new resistances. The "average" reduction of 440/(440+resistance) seems to fit, but is most likely slightly different for an 83 boss. However as Mijae mentions the most important thing here is the minimum possible resist and the chance of it.

Tbh 335 resistance for a 1% chance of full damage seems high. The possible resist percentages at each level of resistance seem fairly tight about the average.

So far I've only done one test that I'm somewhat confident in, and that was 481 samples of Magma at 350 FR. One caveat is that this would use the resistance checks for an 80 mob, so the effect of the level difference is unknown.

The breakdown was as follows:

30% Resisted: 47 (9.8%)
40% Resisted: 145 (30.1%)
50% Resisted: 214 (44.5%)
60% Resisted: 75 (15.6%)

To me, it seems highly unlikely that under the new resist system, a full damage hit is even possible with this level of FR. Based on even this moderate number of samples, there is a 99% confidence that if it is even possible to have a 20%, 10%, or 0% resist with 350 FR, the combined chance is below 1%.

In other words, I could accept that a 20% resist is possible with 350 FR, though highly unlikely - but this makes me near certain that 10% and 0% resists have been pushed completely off the table, and fairly confident that 20% has as well.

If this is the case, I'd be interested to find a rough minimum amount of resistance needed to eliminate resists of 20%, and 30%. Hitting such goals, then maxing stam, could basically allow us to be self-sufficient in OS10, and remove a lot of risks in OS25.

I'm not sure if anyone out there has or is running tests like this, but I'd certainly be interested. Though it's rather time consuming to get statistically meaningful data at many different values of FR, and again, the current effect of the level difference is not known precisely either, though to be safe I imagine adding 15-20 to w/e thresholds are found would be sufficiently accurate.

Edit: Ok got curious ab how reliable even higher resist might be and did one more -

429 FR - 538 Samples

40% Resist: 23.8%
50% Resist: 43.7%
60% Resist: 28.6%
70% Resist: 3.9%

99.5% confidence that the chance of a lower resist is under 1%, if possible at all.

For the record, average damage taken at 350 FR was 53.4%, and at 429 FR, 48.7%.

The predicted averages from [440/(440+FR)] are 55.7% and 50.6%. While the sample size is moderate, it does seem a bit larger than the Greedy Goblin link's sample size per point, and the numbers actually seem to better correspond to a 400/(400+FR) fit: 53.3% and 48.3%. This would make some sense as well, as 400 is simply Level x 5.

If this is indeed the correct relationship between FR and "average" damage, then it's quite plausible that the minimum FR required to knock a certain resist chance for an 83 mob off the table is simply the FR which provides the same "average" resistance as a level 80 minimum. This would be 415/400 times the 80 value. Of course as seen with things like +hit, level differences don't always have such straightforward effects.

So... Is (getting to and fighting for an hour) Baron Geddon soloable at this point? I bet a dragon aura in BWL would be faster to test, but requires a partial clear with a good number of people.

If not... Drakkisath in UBRS?

Alright so it turned out I was curious/bored enough today after all...



Just 4 different FR values so far. I'm pretty much certain at this point that sufficient resistance pushes certain possibilities completely off the table, just like if you have insufficient resistance you'll never randomly see a 90% resist. The sample size is large enough now for a 90% confidence that if outliers do exist, their combined probability is less than 1/1000. It appears more likely that Blizzard is using discrete cutoffs of a probability distribution (sadly, not a Gaussian one, so that part I still haven't been able to nail down).

Also, an average damage taken of 400/(400+Resistance) seems to hold.



Things to note - a resist value of 100 should give an average resistance of 20%, the data agreed well at 19.4%. Only three resist values seem possible at this FR: 10%, 20%, and 30%. Interestingly, however, the probability is not symmetric, and 10% resists are more likely than 30%.

At other resistance values, it's even harder to nail down the formula used to determine the probability of each resist, because the average resist does not fall on a 10% mark. I've been assuming it's some type of common probability distribution about the sustained average, which is subsequently chopped up into these 10% bins, but so far haven't found anything that fits.

If anyone else out there is statistically inclined I'd be happy to share more raw data to help determine a fit - just didn't want to clutter this thread too excessively...

As far as testing methods, since old world bosses are still "boss" level do they obey exactly the same hit/resist mechanics? I guess I thought so as well, but in my head it just "feels" like they barely land hits sometimes so I wasn't sure. If that's the case I could probably figure out something in Azeroth to mess with. The nice thing about Magma is I get 600 samples in 20 minutes, going to be tough to get that kind of data from any boss.

Really nice work, shamgarr.

I think that's an incorrect way of determining the damage mitigated, mijae. It certainly does not jive with what has been actually observed being resisted and the damage coming in. I suspect the formula is going to be something like:
(14062* (1-FR)) *(1.5*2*1.75) * (.88*.97*.98)

Which at 40% FR gives 37054 damage. That's more in line with my anecdotal evidence as well; the max damage I received appeared to be much closer to this than to what your number showed.

42k after the debuff is 56k before the debuff. What gear set are you using where you manage to have 56k hp using FR enchants and flask?

It's not that hard if you're a JC. If you're not a JC, I believe the absolute max stam you can have is around 54.5k.

Here's an example of one of the parses from my log:
22:10'56.375 Sartharion #14 Flame Breath hits Felhoof for 32978 Fire. (7669 Resisted) (1388 Absorbed). From that, it seems clear that the resist is pre-modified; the breath should have been far higher at that point in the fight.

Similarly, on one that killed me:
22:05'12.688 Sartharion #13 Flame Breath hits Felhoof for 40005 Fire. (5916 Resisted)
(that was after I had lost my FR aura from my paladin, hence the higher number and lower resisted damage).

Another one that killed me, barely:
20:48'02.453 Sartharion #3 Flame Breath hits Felhoof for 39031 Fire. (5793 Resisted) (140 Absorbed)

If I had had a shield on that one I would've made it; I don't have 415 FR, so some times I will take an unlucky hit.

Full results of my resists can be found here.

Here's an example of taking Vesp+Shad breaths, then shadron dying:
21:57'09.703 Sartharion #12 Flame Breath hits Felhoof for 23523 Fire. (6088 Resisted)
21:57'32.141 Sartharion #12 Flame Breath hits Felhoof for 18783 Fire. (9110 Resisted) (2336 Absorbed)
21:57'44.188 Sartharion #12 Flame Breath hits Felhoof for 31210 Fire. (6175 Resisted) (2196 Absorbed)
21:57'56.281 Sartharion #12 Flame Breath hits Felhoof for 12804 Fire. (8572 Resisted)

The last one is clearly the Shadron death.

I think it's pretty clear to me that the way I posted is basically how breaths work. And it also does mean that with minimal cooldowns or consumables a tank should be able to survive the breaths assuming they have at least 280 FR. Barkskin will deal with one, SI will deal with another, nightmare seed/FR prot pot should deal with a third, and by then barkskin will be up. A shield should even be enough to deal with this damage too.

This seems like only an option for LW druids, but it could help some.

This is correct, everything is simply multiplicative, in effect.

For example, with a 50% resist of a 10k base damage spell, with no other modifiers, you will see 5k damage, 5k resisted.

All "damage taken" buffs/debuffs, like Protector of the Pack, or Twilight Torment, apply to the actual damage taken, 5k. All this means is that with barkskin up, for example, you would see 4k damage, 5k resisted. The resisted damage is not modified by any "damage taken" modifiers, but in practice this makes no difference (you would instead see "4k resisted" if it were, but your damage taken is the same either way).

Gift of Twilight on the other hand, increases Sarth's damage dealt by 50%, so it would increase both the amount taken and the amount resisted. Again, the effect is exactly the same, multiplicative. I only point it out because it will increase the "resist" number in your combat log by 50%, so needs to be taken into account if you're trying to determine what the base damage of a breath was.

I don't think this is true. If it were true, I would expect to see far higher resisted numbers in my combat log than I am seeing. The resisted amount in the log stays reasonably close to the same regardless of overall breath damage.

Here is a simple probability distribution that seems to be reasonably close to this chart.

Compute the expected damage taken, DT = (400/(400+FR))

25% of the time blizz will "aim" at DT-10%.
50% of the time blizz will "aim" at DT.
25% of the time blizz will "aim" at DT+10%.

If the "aim point" is a multiple of 10%, that will be the actual damage taken. Otherwise it will move up or down to one of the bounding multiples of 10%. The probability that it will move up is exactly how far "up" the aim point was between the two boundaries.

Example, for 429 FR:

Expected DT is 400/(400+429) = 48.2509%

48.2509 is 82.509% of the distance between 40 and 50.

"Aim points" can be 38.2509 (25%), 48.2509 (50%), or 58.2509% (25%)

Probability that final damage taken is

DT 30% = 70% resisted = 25% * (100%-82.509%) = 4.4%
40% = 60% resisted = 25% * 82.509% + 50% * (100%-82.509%) = 29.4%
50% = 50% resisted = 50% * 82.509% + 25% * (100%-82.509%) = 45.6%
60% = 40% resisted = 25% * 82.509% = 20.6%

This distribution preserves the 48.2509% average DT, and looks like a pretty good fit for the graph.

My formula breaks down if DT > 90%. More testing would be needed to see how they handle that situation.