Anaxie's Totally Revolutionary SoO Retribution T16 BiS & Theorycraft

[Huntingbear_grimbatol]

Why not go with what is best on average?

Go with the spec that is ~1% higher on average
Go with the spec that is closer to the average every pull.

Crit you might get higher average number over a night of progress.
Mastery you'll be more consistent on your numbers.
Not saying one is better than the other, it's player choice

- - - Updated - - -

You don't argue with the logic of Zardox.

I don't argue or disagree with him at all, but let's say you go crit over mastery. You'll do crits let's say around 40% of the time if you go balls out on it, now you can be lucky one pull and get crits when it's optimal (adds, cds, whatever) or you can be unlucky. Same as our 4set, you could get lucky and chain proc 10 times when the Thok bats come out... or you won't. It's RNG and while on a complete night it's gonna average out to a number comparable to someone who went for mastery the person that went mastery is gonna have a pretty rough number that he can go on. I've tried crit build, didn't like it because it "swung" too much from the good to the bad. Does that make crit build bad or worse than mastery? No. Does it make mastery better than crit? No.

I'm missing a couple HC+WF pieces but I'm pretty much BiS if you ignore warforged, raidbuffed 17k haste and 20.8k mastery and 3.5k crit iirc. in full BiS (wf inc) I might go above 21k mastery and maybe 4k crit whilst keeping my 17k haste, I'm also draenei so I got a bit more room with the hit offpieces. Don't think I've seen anyone with higher mastery while keeping a reasonable haste number.

[Neldarie]

Every class/spec has plenty of rng to them (sure some less than others). But you maybe should look into the bossfight/design of SoO instead and how those "im now god" moments become shorter as gear better. Or say you're playing with subpar dpsers in one guild, then switch to another being clearly superior in every aspect and you will see what I'm talking about instantly. The reason why crit is rather strong this tier is very simple: Mastery per point gain is suboptimal at best (just look at flat % increases back on Primordius and you see what I mean by that), while crit benefits ALL your dmg (mastery-based included), together with metas and this tier cloakprocs... But ye it's SoO and as someone pointed out earlier (siegecrafter belts 20secs) + fights getting retarded short in most cases. Then if you go and look at stuff like Paragons/Garrosh (10min+ stuff), you will again see what I mean there.

[Deleted]

I've tried crit build, didn't like it because it "swung" too much from the good to the bad. Does that make crit build bad or worse than mastery? No. Does it make mastery better than crit? No.

I'm glad you wrote this. While it is personal preference which stat to go for, one will, on average, always be superior to the other. I try to always go for what is best on average, and discard the RNG between attempts, though it is frustrating to see your DPS fluctuate by upwards of 50k.

[Huntingbear_grimbatol]

I'm glad you wrote this. While it is personal preference which stat to go for, one will, on average, always be superior to the other. I try to always go for what is best on average, and discard the RNG between attempts, though it is frustrating to see your DPS fluctuate by upwards of 50k.

Exactly, and while it's not only the "luck" of crits that's gonna impact your dps it's one more thing to put your trust in.
And yes you're correct, on as low as 1000 or even 500 pulls you'll see the average of crit go above the average of mastery. That said you'll get bumps and dips along the way which you'll see less of with the mastery way, I'd rather be mediocre always than awesome every other pull but that's my preference.

[Meiffert]

As for crit vs mastery? I prefer limiting the rng in my damage to the set bonus and instead just buff the damage it does (through mastery) every time it happens instead of crit which only buffs your damage once in a while, or every hit for 20sec, or not a single attack for 20sec... whatever it feels like.

Can you quantify how much more consistent mastery is compared to crit?
You just stated that there is a huge difference as a fact and I'm not convinced.

I'm not saying you are wrong, but just look at some examples.
You have 30 % crit, sometimes you crit a lot and sometimes you don't, there is a big difference from pull to pull. Now you get a lot more crit and manage to cap it at 100 %. Suddenly there is no difference at all, you crit exactly the same on every pull (you crit with every attack). In this example getting more crit made you do way more consistent damage.
When I look at mastery, it seems to increase the damage difference between good and bad procs a lot. If you get a lot of DP and 4pc procs, you use a lot of strong attacks (TV and DS) while without these procs you are forced to fill the GCDs with Judgments and Exorcisms. Now while you may sometimes crit on TV/DS, you can just as easily crit on J/Exo. On the other hand mastery will always make your strong procs even stronger and it will never help when your damage is already weaker.

Overall you could easily come to the conclusion that crit is the more consistent stat compared to mastery. Now I'm sure we could just as easily come up with scenarios/examples where the opposite is true, that's why I would like to see some hard numbers on how much these stats are (in)consistent.


That brings me to my second point and that is the value of consistency. You seem to overvalue it a lot in many posts in this thread. Usually you argument with the scenario where you need to make a close DPS check on every pull and not wipe your raid on half of your attempts by having below average damage.
This situation seems somewhat rare though and you can make almost the exact same argument against having consistent damage. Just imagine there is a key DPS check which is the hardest part of the fight and with a highly consistent setup you just don't meet it. Now you wipe every single time missing just 1 % of DPS. If you had a more inconsistent damage, you could meet this DPS check maby 30 % of the time and since it is the hardest part of the fight, you have a good chance of getting your server first on every third pull.

I would still personally prefer more consistent damage if both setups were equally strong, but I wouldn't typically go for a weaker stat just because of it.

[Imperviable]

It's my opinion that you should just spec for the higher value between crit and mastery. You perform so many actions during a boss fight that crit isn't really random. If we were talking about a 20 second window, I could see someone preferring to go a less damage, more consistent stat priority, but that situation is incredibly infrequent. And then when you're talking about a 6-10 minute boss fight, crit will average itself out. Your dps swings far more due to timing and frequency of 4pc procs, yet I don't see anyone dropping their 4pc tier gear in favor of better warforged pieces.

[Meiffert]

Sigh. I'm aware of all the arguments you wrote, I even said in my post that we can easily come up with specific scenraios/arguments for mastery > crit consistency. I showed some specific scenarios where the opposite is true specifically so that you can see that comming up with these proves nothing.

Mastery damage will always increase the value of Crusader Strike, Hammer of Wrath/the Righteous, Templar's Verdict and Divine Storm. There's no arguing against that.

That's somewhat my point. Mastery will increase the damage of your strong attacks. Therefore if you are getting lucky with DP/4set procs, it will increase your damage even more, if you are unlucky being stuck with J/Exo/empty GCD, it will do nothing. That only increases the difference between a good and a bad RNG.

Saying that Crit is a more consistent stat over Mastery in a situation where you have 100% crit is flat out dumb, because said situation will never exist.

Fair enough. Maybe my example was too extreme and unrealistic, let me give you another.
You have 20 % crit. You can be unlucky on one pull getting only 5 % crit or lucky on another and critting with 35 % of your attacks.
Now you have 30 % crit instead. You can be unlucky again with 15 % crit some time and lucky with 45 % other time.
Where is the big increase in inconsistency? In either scenario you can get equally lucky or unlucky, you can crit say 10 % less than is your average or 10 % more.

I never understand all these discussions.

We discuss these things because we are in the theorycrafting thread. Someone came up with the argument that we could prefer a weaker stat for the sake of consistency which might be an interesting idea and I like it. So now we are exploring possibilities how this idea could be used with our specific secondary stats.

Average doesn't mean much if 20% of your wipes comes from no crits. (Extreme situations)

Consistent doesn't mean much if 100 % of your wipes comes from missing the enrage timer by 20 damage. (More extreme counter-example. )

I do agree that having more consistent damage is better more often than not, I even agree that it may be valuable enough to prefer a weaker but more consistent stat if the difference is as small as it is between crit and mastery for ret paladins.


Let me make another attempt to explain my position: I'm not saying that mastery is more consistent or that crit rating is more consistent. I'm simply saying that I don't know. That's why I asked if you have any evidence/proof for your claim.
Doesn't simcraft show the consistency of damage in its results? If so, we could use it to calculate how consistent the damage was with mastery setup and then do it again for a crit setup.


Also from the examples you are choosing to show where crit is inconsistent it seems like you are confusing two things: crit and mastery mechanics versus crit and mastery rating.
As a mechanic itself, crit is highly inconsistent, which is what you try and show in your examples. But we can't just eliminate this inconsistency because we will always have some crit (there is 5 % crit raid buff and 10 % from inquisition at least).
Therefore we always have this inconsistency and now you are taking it a step further saying: the more crit we have the more inconsistent it becomes. And that's where you skipped some steps. How did you come to that conclusion? As I already showed in an extreme example, if you got to 100 % crit chance, we would actually eliminate the inconsistency.
How consistently does crit rating behave before this point is what we need to figure out.

[Deleted]

The standard deviation of crit on any lengthy fight (5+ min) is quite small. As Imperviable posted further up, if it is so crucial to your raid that your DPS varies as little as possible, why not drop 4p in favor of more consistent warforged items?

[Meiffert]

With mastery, this factor is not there because it's a set value.

That is not correct.
If the difference between good and bad pull is 20 % dps and you add another 5k mastery rating, the damage of both bad and good pulls will be increased, but the difference between the two will be even higher.

During progression, would you want to spend 20 attempts to see if you can do it with a ret paladin in a mastery build, or would you want to spend 80 attempts praying for the ret to crit at the right time?

Would you rather spent 80 attempts wiping with the boss getting very low before giving up or would you rather kill it on your 5th pull where you were a bit luckier with some crits/procs?
Equally ridiculous argument.

Read above. If the ret is the issue for wiping to enrage, replace him with a better dps.

How is this relevant to the question of how valuable consistency is for your raid dps?
Now we have a warlock in our raid and with his more consistent but lower damage setup he is missing the enrage timer by 0.01 % as well. If he used a more volatile reforging, he could kill the boss in time 8 times out of 10.
Now who should we replace him with?

By having less crit and more mastery, you have less inconsistency because there is a less chance for you to crit.
It's all fun to theorycraft and all, but the fact is that crit is random and mastery is a static increase.

That is NOT a fact. You (and Huntingbear_grimbatol) came with this claim and I told you that I don't believe it and I think it is wrong. I asked you 3 times now to prove this exact claim to me somehow and even sugested that maybe simcraft could be used to provide us some answers (standard deviation is I believe the correct statistical term, thanks Zardox), but please feel free to use any other way to prove it or at least provide some arguments for your claim.

I already showed why I believe mastery rating is NOT a static increase and how it actually increases the dps difference between good and bad RNG.
I also showed that in some instances having more crit rating actually makes your damage more consistent, so it is by no means always true that "by having less crit, you have less inconsistency". It may be true for realistic values of crit, it may not.

[paoani]

The point of consistency that they are trying to make isn't that crit will perform worse on average. Any basic statistics knowledge will show you exactly what they are referring to.

Let's assume we treat your dps with crit as a binomial distribution. The distribution of dps you would see at various levels of crit over 100 equal hits is as follows, with 0 representing the worst cast (0 crits) and 100 representing the best case (100 crits):





This obviously is not entirely representative of a real raid scenario, but for comparison purposes illustrates what I am trying to show nicely. The actual distributions for a real combat scenario will be significantly thinner. The average increases as crit chance increases, as you would expect. The minimum and Maximum dps do not change, but the probability of getting either does. The worse case for crit is the same at 0.00001%-99.99999% crit, as is the best case.

By increasing mastery, we raise the average damage by a smaller amount (although only slightly in this case), but we do it by raising the worst case and best case at the same time. No matter how you attempt to argue, the mathematical worst case for mastery is always higher than the worst case for crit simply because the worst and best case scenarios never change by increasing crit. By increasing crit you do not increase the floor of your damage, you increase the probability that you will perform better within that distribution. This means that you can better increase consistency by increasing your mastery such that the worst case is good enough to succeed.

Using Siegecrafter belts as an example:
Lets say that you need to do 1.3m damage and that you do 1.0m damage with the worst case (0 crits, assuming procs are irrelevant for the sake of argument), with the base build in this case you would need ~30% of attacks on the belt to crit (Roughly anyway and close enough for the purposes of illustration).

To succeed on the belt with a crit build, you still need to crit ~30% of attacks on the belt. By increasing your crit, you increase the probability that you will crit more than 30%, and therefore increase your probability of success asymptotically up to 100%, but never reaching it.

To succeed with a mastery build, you will lower the percentage of attacks that must crit in order to succeed. By increasing mastery enough to raise your worst case damage by 30%, you have guaranteed success in 100% of cases.

This phenomena is what Etsumii and others are referring to by "Consistency." In the event that you are very close to success, you can increase your likelihood of success by a much greater margin by increasing the minimum damage you do than by increasing the likelihood of a higher damage pull. Therefore mastery would provide greater chance of success than crit in the event that they provide similar increases in average damage.

[Deleted]

Excellent explanation paoani, you should post here more often. The example you gave works well in a ''breakpoint'' scenario (such as the belt), but is not applicable if you don't care about big variation in your DPS, and only care about maximizing your average damage(?).

[Meiffert]

Thanks, paoani, your idea of worst case scenario is interesting and you are right that you can't increase your worst case performance by getting more crit rating unless you manage to get to 100 % crit.

Yet I don't like this definition of consistency for 2 reasons.

1. It is too unlikely to get this worst case scenario even with low crit such as 20 %. (Since we get 5 % raid buff and another 10 % from inquisition, 20 % is very low number I think - I'm sitting way above it with haste > mastery > crit setup myself.)
The chances of not getting a single crit out of only 100 hits is about one to five billion. I'm sure your raidleader will forgive you for wiping your raid that often.

2. If we tried to plan for worst case scenarios (no crits, no abilities procs, no trinket procs etc.), we would come to the conclusion that we are unable to kill any bosses without overgearing the content by significant margin.


Let's consider a situation where we have 2 different gear setups (or classes etc.) both doing 105k dps on average and now:
- first of them is doing between 100k and 110k 98 % of the time, but has the potential to drop as low as 60k in the absolute worst case scenario (no crit or proc in the entire fight)
- second of them is doing 90k-100k 20 % of the time, 100k-110k 55 % of the time and 110-120k 20 % of the time with the worst case scenario being 70k


I would consider the first of these two to be more consistent even though his potential worst case scenario dps is lower.
I believe there are some good ways to measure consistency in statistics and we could find a way to quantify this.


Edit: an interesting measure of consistent dps could be something like 5 percentile (if you do this damage/dps at least 95 % of the time, you are considered to meet this dps check "consistently").

[anaxie]

Merry Christmas HO HO HO

Replaying Baldurs Gate 1/2 / IWD during the 6 month downtime between expansions. Anyone interested in a 6 man multiplayer?

[Butosai]

Anaxie how do i increase my DPS? Here are my logs.
http://www.davesgames.net/papercraft...ry-logs-01.png

[Huntingbear_grimbatol]

Well said Paoani

The example you gave works well in a ''breakpoint'' scenario (such as the belt), but is not applicable if you don't care about big variation in your DPS, and only care about maximizing your average damage(?).

If you want to minmax dps and get ranks then go 70% crit and pray that your raid manages to kill it on one of your good attempts, crit has a higher potential dps than mastery does, most of us have cleared content now so all of us pretty much argue on how to rekill bosses or whatever (right?). But if you're still progressing (which is the only thing I care about to be perfectly honest) then one of the best (imo) things you can do as a raider is being consistent. Consistent in performance, numbers, attendance, calling things every time if you're set to do so, stun at the same time, interrupt in your place in the lineup and 1000 more things. Back to the siegecrafter belt, you could send up say 3 firemages right and hope they all have huge crit luck and combustion ticks for 300k or whatever, chances are they won't do that every single time. That means you'd send the 3 mages and one or two additional players OR you could get 4 stable dpsers to do it and potentially gain 1 more dpser on the boss.

If you end up doing 300m damage on Thok one pull and 310m the next pull doesn't matter as much because in your raid you'll have people with bad and good rng every single time you pull a boss, but for smaller things (like belts and siege engineers) where the difference between good and bad rng will wipe you? Ask 100 raid leaders what they'd preffer, overkill it 90% of the time and wipe 10% of the time or kill it 100% of the time.
If all you care about is "scumbagging" dps and getting that rank 1 on WoL to say "YES, I HAD RNG ON MY SIDE!" then that's your choice and I don't think that's wrong.
Because that's what WoL ranking is isn't it? Doing the same thing over and over and just waiting for that 1 kill that makes everything line up perfectly.

[anaxie]

Anaxie how do i increase my DPS? Here are my logs.
http://www.davesgames.net/papercraft...ry-logs-01.png

thank you for that

[Maegor]

It absolutely blows my mind that we had a post that had probably the best explanation of what consistency means in 300+ pages while showing and explaining the math behind it, and then we have another poster still arguing against it with voodoo math - I haven't seen a single bit of hard numbers, only hypotheticals that aren't based in math at all.

The mastery vs crit can be broken down further (and dakeshi has done a pretty good job) - you know what abilities mastery effects (and it includes our most used hp generator and our two other best ST attacks in HoW and TV). You don't get the same guarantee from crit. It doesn't guarantee (if your raid buffed crit is at 25%) that 25% of your TV's will crit. In my last IJ fight I used TV 52 times in a 4:55 fight. That is way too small of an amount of attacks to say would be consistent fight to fight. I also used HoW 57 times, again not a large enough amount to consider that the average will play out over a fight. That is the benefit of mastery, we know how much it will increase my dps based on the number of those attacks.

We cannot get to high enough levels of crit to make the damage from it consistent. The highest crit I've seen on somebody is 28% un buffed, meaning he would be at 43% in a raid setting. One reason we can't get high enough, is because we have to dedicate a certain amount of our stats to haste in order to use our finishers. To get to 28% base crit, that pally is at 30% haste and 42% mastery.

Also, saying that the 4pc is rng, so if you don't like crit you shouldn't like the proc is absurd. The main benefit of the proc is providing an attack that helps fill gcds. That is why we can stop at 40% haste (and why haste is theoretically superior to 50%). We don't see as many empty gcd's, and people mistake this as more haste equaling less downtime, when really, you have relatively the same amount of downtime at 30% haste as you do at 45% haste if you don't have the procs filling the empty gcds.

Edit: I'd add that the attitude of only having 30-40% of attempts be successful personally is a terrible mentality to permeate through a raid group. Do the math on a 10 person raid where the 5-6 dps are only going to do good personal dps 30% of the time. If it takes 4 of your 6 dps having a good pull to down a boss, then you will only be successful as a group 7% of the time.

[Huntingbear_grimbatol]

It's not THAT drastic Maegor but yeh I agree.
Also our 4set (this and previous tier) will always benefit from mastery increases and we can dictate when and where we want to use our current 4set proc, we can't do that with crit. Take Garrosh HC, you get a 4set proc and MC is up in 5 seconds. You hold on to it and BOOM you do a huge cleave on 5 important targets, you can't guarantee that with crit. Our 4set isn't "great" but in some situations it is insanely strong and in most cases it's gonna help you fill gcds (which is why we don't "need" more than 40% haste like you said), now that progress is over I do kinda want to find my old 4set bonus, lei shen trinket and see how high I can crit with a pure holy TV and 40k mastery... *edit* for some reason when the lei shen trinket procs I go up to 65k mastery... guessing it's because of Thok's trinket but still, just wow with 216% mastery XD

[Maegor]

The 7% in my edit is wrong, but eggnog math can't be checked on the couch. Still, it isn't a good attitude for the raid to have.

[Neldarie]

Kinda funny how people fail to understand that higher crit vs higher mastery is bigger consistancy in the long run for very obvious reasons. And that it is actually mastery stacking that has the highest dmg swing (from low to high). The ultimate low will always be lower for crit because mastery = free dmg from thin air (which is also exactly why Mastery stacking has the highest overall performance once in a bluemoon). But when talking about consistancy and the long run of things the probability of having higher crit %age will make your overall performance alot more stable (funny that my best siege belt performance was when I was reforged highest for crit).

There hasn't been any math to talk or show any of mastery or crit in this thread. What paoani talked about was something else. It was related to the discussion. Where true math would lie in this eternal debate of ours would be the benefits and gain/loss of either of those 2 stats PER rating. Because utlimately you always trade 1 crit rating for 1 mastery rating. And how both of those scale with higher number of each + their trinket/ap/wpdmg/setbonus interractions as gear gets better. That would show the only true/real value of either of the stats.

The curve in paoani graph for maximum mastery stacking would be fat and short in comparison to the one of crit which would be taller and more narrow to show probability of things at work.

Bottomline for people who don't want to understand or read (those are mathematical definitions of them):
Higher consistancy (smaller swing from low to high) = more crit.
Higher potential (bigger e-peen? + making total fail rng scenarios not as painfull) = more mastery.