Tuesday, November 17, 2009

[Druid] Cat in the Cap

UPDATED. Changed the white attack crits to be a bit clearer.

I couldn't resist.

With the new patch feral cats are going to start having to consider a couple of interesting soft and hard caps on their gear. At least I thought it was interesting; it wasn't something I was thinking about. And one...may surprise you.

First, the basic easy ones that everyone should know about already: the hit and expertise caps for ferals.

Hit Cap:
The normal hit cap for ferals is 8% hit - meaning that you need 8% hit to reduce the chance of misses to 0% against bosses that are 3 levels above you. This comes out to be 263 rating - or 230 rating if you have a draenei in your party. Any hit rating above this is wasted. For sake of completeness, here are the different caps with the most relevant ones to raiding highlighted:

Hit rating needed vs. lvl 80 mobs: 164
Hit rating needed vs. lvl 80 mobs with draenei: 132
Hit rating needed vs. lvl 81 mobs: 180
Hit rating needed vs. lvl 81 mobs with draenei: 147
Hit rating needed vs. lvl 82 mobs: 197
Hit rating needed vs. lvl 82 mobs with draenei: 164
Hit rating needed vs. lvl 83 mobs: 263
Hit rating needed vs. lvl 83 mobs with draenei: 230

Expertise Cap:
The expertise cap is dependent entirely on whether or not you took primal precision or not. Which you should almost certainly do if you do not cap for hit or expertise, as the secondary function (refund on misses) is the reason that hit/expertise are not ranked as highly for cats as they are for other classes.

For every 8.2 points of expertise rating, you gain 1 expertise skill. Each point of expertise skill reduces your chance to be dodged/parried by .25%. You can be dodged at a rate of 6.5% against level 83 mobs. So with Primal Precision, you need 16 total expertise skill. Without it, you need 26.

Expertise rating needed vs. lvl 83 mobs (with primal precision): 132 (due to rounding)
Expertise rating needed vs. lvl 83 mobs (without primal precision): 214

Armor Penetration Cap:

The armor penetration cap is simply how much armor pen it takes to reduce the contribution from armor to a maximal value. This is a fairly complex calculation used to figure out exactly how much will be reduced, but the end result is this - at level 80, you need 13.99 armor penetration rating to reduce 1% armor.

Thus, at 1399.something rating (or 1400), armor penetration is capped in contribution at 100%. That's the absolute hard cap.

There are several soft caps based on trinket procs. A soft cap means that with a proc on the trinket, you will reach the hard cap - but only with that proc on the trinket. Every simulation I've seen uses this logic: it is better from a gearing perspective to get one of these trinkets and gear to the soft cap (and go agi afterwards) than it is to go over the cap. This makes some sense, as every time the proc is up on the trinket, all that extra armor pen that you went over would be wasted.

As of writing now, there are three trinkets that provide (or will provide) armor pen procs:

Grim Toll (provides 612 armor pen for 10 seconds on a hit every 45 seconds) - found in Naxx
Mjolnir Runestone (provides 665 armor pen for 10 seconds on a hit every 45 seconds) - from hard mode 10-man Thorim
Needle-Encrusted Scorpion (provides 678 armor pen for 10 seconds on a crit every 45 seconds) - from Devourer of Souls in heroic Forge of Souls - the new 5-man.

For the sake of this discussion it's unimportant about which is going to be better than which here; I'll talk about that later this week. The numbers are what matter right now.

Armor pen cap without any armor pen proccing trinkets: 1400
Armor pen cap with Grim Toll: 788
Armor pen cap with Mjolnir Runestone: 735
Armor pen cap with Needle-Encrusted Scorpion: 722

Critical strike cap:

This was the new and exciting bit of information for me. With procs of various trinkets, cats are going to be easily blowing past their critical strike caps some of the time. For instance, Death's Choice gives a crazy 510 agility on a proc - that's 6.1% crit by itself and that's pre-Blessing of Kings multiplier.

But what are the cap levels? Well, this is where it gets complicated, as there are two different systems depending on whether it's a normal melee attack (white attack) or if it's a special attack (yellow).

For white attacks, there is a one roll system. This means that every attack rolls randomly against your hit table, and depending on where it lands will determine whether or not the attack misses, dodges, hits or crits. Paladin tanks should be very familiar with this hit table. The important thing to understand is that things on this table have a precedence, and you can push values off depending. For example, getting 8% hit pushes misses off the table completely.

But you can never remove Glancing Blows. The way it works is that there is a 24% chance of an attack being a glancing blow. This is independent of any other ability and can't be reduced, and can't be pushed off the table. In other words, no matter how high your crit chance is, you will always have this chance. In addition, there's a 4.8% chance of an attack always being a 'hit' due to some kind of crit deflation. This means that 4.8% of your crits will be converted to hits and not a crit no matter what - again, this cannot be changed to a crit no matter what. However, you need 4.8% crit to push this off the table - otherwise you'll have even more hits. Then there's misses, dodges, and parries. Those can be replaced by crits assuming you have the proper hit and expertise cap.

So the highest white attack crit % is (100-24%) = 76% crit. Any higher than that, and you're wasting crit chance on your white attacks. This means that in practice, 76% is the highest crit value (assuming maximum hit/expertise caps). And this means assuming you hit an infinite amount of times, you will get 71.2% crits from your white attacks.

For special/yellow attacks, WoW uses a two-roll system. What this means is that first the game checks to see whether or not the attack lands (using your hit/expertise skill), and then uses your crit chance to determine whether or not it was a crit. This means that the crit cap is higher, as special attacks cannot be glancing blows.

What's more interesting here is that there appears to be a special crit reduction in place on all yellow attacks. This has been shown to be 4.8% over at EJ. Which means that the base cap for yellow attack crits is 104.8%.

Yikes. Add into this that you will have Rend & Tear, and it gets fairly annoying to figure out.

But thanks to Murna, here are some fairly nice descriptions of various scenarios for crit caps (edited with more information - the 4.8% chance needs to be pushed off the table)

68.72% - White Crit Softcap with [Death's Verdict] proccing (+ 7,28% from Procc + 24% Glancing Blows = 100%)
69,58% = White Crit Softcap with [Death's Verdict] proccing (+ 6.42% from Procc + 24% Glancing Blows = 100%)
76% - White crit cap (+ 24% Glancing Blows = 100%)
79.8% - Ferocious Bite Crit Cap (+ 25% Crit from Talent - 4.8% Crit Suppression = 100%)
97.52% - Yellow Crit Softcap with [Death's Verdict] proccing (+ 7,28% from Procc - 4.8% Crit Suppression = 100%)
98.38% - Yellow Crit Softcap with [Death's Verdict] proccing (+ 6.42% from Procc - 4.8% Crit Suppression = 100%)
104.8% - Yellow Crit Cap (- 4.8% Crit Suppression = 100%)
Again, these are samples. If you have DM:G or some other crit trinket proc or a proc like mongoose, this will change. But what this does indicate to me is that you should value these possibly a bit less, especially as you get into higher-end gear. For example - right now, without procs, I'm sitting at a bit over 61% crit when I'm raid-buffed. With a proc from DM:G and mongoose, I'll be at the white cap and the extra crit is going to be wasted. Right now that can't be helped (mongoose is on the staff because I'm using it for cat and bear) but it definitely factors in my view of what to gear.

And finally, if you're not at hit/expertise cap, these numbers will be different for you.

In terms of what this likely means for Icecrown...it's fuzzy. Chances are that we wouldn't be able to hit 71.2% crit naturally, but that would definitely be a point where you should consider gearing something else (like, oddly, strength). My gut feeling is that we'll be going for the hard cap of armor pen, and if you happen to reach 1400, then you can try for the caps for crit. I also think that this clearly indicates hit/expertise capping is going to be valuable given some of the procs on trinkets, just because more crit isn't going to necessarily help you. I wouldn't worry about reaching the FB cap; FB still will not be a large component of your damage most of the time.

But again, more on that in a bit.

Thursday, November 12, 2009

[Druid] Icecrown loots for bears

Finally, we're getting to see a bit of the gear in Icecrown. I'm really shocked how long it's taken; usually we have most of the loot tables by this point in the PTR. This might mean that the PTR is going to take extra long...or that the gear is just going to have some issues.

There isn't enough to give a BiS list like normal, so I'll just be doing some analysis of it for bears (and later, for cats).

The TLDR part: T10 is great, save for that glaring lack of expertise.

First off, the links to the list of gear are at MMO champion here (for random drops, T1o pieces), here (for 5man and QD rewards) and here (for Emblem of Frost purchases).

On getting the tier gear: (thanks, Copey) - they're going to do something similar to how they did ToC tier gear. What happens is that you spend emblems of frost to buy the lowest level T10 gear. Then, you use that gear AND a token gained from 25-man to purchase the next level of T10 gear. It's not clear whether the tokens are slot-specific, class specific or what - but that's the basic system.

The first thing to note is the huge jump in T10 vs T9 and agility. For instance, the T9 hard mode chest piece has 136 agi and 153 stam; the T10 chest piece has 162 agi and 162 stam. The stam is going up as we'd expect, but a jump of 26 agi from an ilvl increase of only 6? That's pretty amazing. It may be that the tier pieces are overbudget or they're not done tweaking. It's odd that there exist so much agi on all the pieces, honestly.

The second thing to notice is that none of the T10 pieces have any expertise on them. None have any blue sockets either. None have hit.

Again, this may change. But I am very skeptical. For starters, we didn't see drastic changes on gear when it hit the PTRs prior to this. We saw changes in things like the set bonuses, but the actual gear stats tend to be fairly easily finalized. Same with sockets. We might see a random change here and there (like correcting that +5 spellpower socket bonus on the T10 gloves) but my gut feeling is that this is what it's going to be like, going forward.

The other observation is that the tier pieces have steadily been more and more itemized for cats. This is never been more the case than T10. Every single T10 piece has huge amounts of agility and armor pen - every single piece! The 5 pieces of T10 by themselves give 444 armor pen and 726 agility. They have zero hit and zero expertise on them - which is awesome for cats, who tend to not care that much about those things anyway and are swimming in hit. Every socket is red or yellow, and all the socket bonuses (save the aforementioned hands) are agi.

They're almost perfect for cats.

But as mentioned before, this makes them less than ideal for bears.

There are options for bears - but because the T10 set bonus is so strong for bears, those options come down to essentially one thing: what one non-tier piece will you take? And so far, the answer looks to be the Emblem of Frost offtier chest piece, which provides 92 expertise and is otherwise itemized the same as the tier piece. I don't honestly expect this to change either; I suspect that there won't be any legs or head pieces with that high of expertise, and hands/shoulders won't have that much expertise due to their lower itemization point total.

There are other good choices for bears, such as the crafted leather boots.

But honestly? It may come down to whether or not you can deal with socketing expertise in your gem slots. And my suspicion is that it will be down to that - if you can afford the EH hit, socketing expertise may be a very good choice to make up for this lack in other gear.

The purchaseable gear via emblems appears to be consistently full of win. The tanking cloak provides a ridiculous 849 armor. The chest is great and looks to compete with the best of things out there. The waist is also very decent. I fully expect that I'll be racing to buy the chest and the cloak first and second - likely the cloak, given that they tend to be cheaper. There's also a set of gloves, but again - you don't want these since you do want the T10 set.

On the weapon side, there are three new ones so far. There's a fairly decent weapon that's comparable to hard mode Twin's Pact. The reward from the Quel'Delar chain is lower level and isn't socketable, but does have expertise; it's definitely worth doing the chain if you don't have a hard mode weapon yet. And then there's a drop from heroic Icecrown 5-man that's...kinda meh, actually. But is at least an upgrade from Marrowstrike and better than Twisted Visage due to the blue socket.

Jewelry wise, there's nothing that grabs me. Most jewelry has either hit or parry on it. No sockets, no armor. Nothing that makes me ooh and ah like Clutch of Fortification did.

And no jewelry that provides any expertise. Only one non-leather piece that does other than that weapon - and that's a cloak from the 10-man which you'd be insane to grab compared to the buyable one.

Finally, there is a purchaseable trinket that provides 228 stamina base and has a decent on-use ability. Wow. I did mention earlier that I wouldn't put it past Bliz to do this sort of thing for tanks. There ya go.

Anyway, if you're a bear you should be hoping for rings and necks to have tons of expertise. And hope for at least one belt/bracer to have it too. Or...hope for them to reitemize the gear, which i suspect they won't be doing. Or hope for some random tanking trinket with a ton of expertise and stamina...somehow.

Tuesday, November 3, 2009

[General, Druid] Icewell radiance part 2 -the 4 fallacious myths

I've been very busy in the last couple days, so I've got some comments to answer - but of the answerable comments, enough were odd that I wanted to address them specifically.

There were 4 statements made in the blog and elsewhere that I kept seeing, and they were specifically incorrect as far as Icewell radiance goes. I'll go with them all Matticus style and talk about them separately.

Myth #1: other tanks can stack parry to make up for the lack of dodge.

This looks reasonable, right? I mean, sure - other tanks lose dodge, but they can just go with parry. All is good. The problem here is that very few tanks were going for any kind of avoidance anyway; they'd get what's on their gear and just stick with that for the most part, with the occasional defense/stam or dodge/stam gem thrown in for metas and socket bonuses.

Here's the important thing to understand about this change: it doesn't make dodge less valuable, it makes all avoidance less valuable. As I pointed out, all avoidance now reduces less damage overall than it used to due to having less of it. But that doesn't say dodge - it's all avoidance. You have less avoidance period - it doesn't matter how you got at that avoidance.

Furthermore, parry is as valuable before from a diminishing returns standpoint as it is now. In other words most tanks would rather stack dodge than parry because they'll get more avoidance out of it. So they can't just switch out dodge gems for parry gems (even if they had them); they'll likely lose avoidance doing that. And if they didn't lose avoidance doing that, they probably should have been doing it in the first place.

Another way to say it is this: if it takes 70 dodge rating to get 1% avoidance now, it'll take 70 dodge rating to get 1% avoidance then. If it takes 75 parry rating to get 1% avoidance now, it'll take 75 parry rating to get 1% avoidance then. Those values aren't going to change.

Myth #2: block is going to make up for it

Block is definitely more scalable with respect to the number of incoming attacks. Savage defense does lose out here. At the same time, every block tank just lost 20% avoidance. Chances are they weren't unhittable (most paladins and warriors aren't in their normal gear, even with HS up), which means that they'll be blocking the same amount of hits as before; dodge didn't push their block off the table. So they'd have to be stacking block to compensate now, if they wanted to - except they've got 20% block to stack.

That's a lot. Now, it's not subject to diminishing returns at least, but at the same time the block gear that exists out there...kinda sucks. Most Anub hardmode tanks know this pain; they're wearing a blue trinket, a bunch of random pieces from Naxx and Ulduar...do you think that gear set is going to cut it in Icecrown?

And they can't gem for block rating either. So where are they going to get all this block?

Myth #3: Bears are going to be dying more due to more hits in a row.

The idea goes like this: let's say that a tank will die in x hits in a row. Tank A, with 40% avoidance, has a (1-.4)^x chance of dying then. The druid, with 31% avoidance, has only a (1-.31)^x chance of dying - which is much greater. For 4 hits, for example, tank A would die 12.96% of the time; the bear would die 22.67% of the time. Therefore, aren't bears going to be at a big disadvantage?

Well...no. Here's the thing - this was true before, right? I mean, tank A had an advantage over druids before. But how much? The avoidance values before were 60%, so assuming you'd die in 4 hits, tank A had a 2.56% chance - and the druid had a 5.71% chance.

So yeah, the chances that a druid takes 4 hits has increased significantly (15% vs 10%). At the same time, the relative chances that this occurs has decreased significantly.

But the most important thing here is that healers cannot assume either tank is going to dodge that many hits in a row any more. Healers have to heal both tanks the same way. As I showed in the previous article, bears are actually going to take less relative damage to other tanks than they do now. What this also means is that unlike in the 2% case, healers can't assume that tank A will avoid an attack every 4 attacks. They'd be wrong 12% of the time, which means 1 out of 8 times they'd be dead.

So they have to heal tank A as if they weren't avoiding. Which means healing the druid is functionally the same.

Myth #4: Because GC said so

Ghostcrawler's said basically that if you think avoidance was good before, it'll still be good. If you think it's bad, that won't change. He's also said that EH is important but not the be-all, end all of tanking. Here's one commenter on this:

Interesting view, but I'll go with what GC has posted about Chill of the Throne. Stamina is arguably less important, avoidance is still valuable, cooldowns matter and are not easily compared, and effective health does NOT = best tank. I love that you write a lot on bears Kalon, and I'm sure you're a great bear tank. But, you seem rather set on the idea that bears are the best tanks. They aren't. They're not dominating leading progression guilds, general opinion of them is certainly not "#1 tank" and the developers don't think so either. So, can you make this your last "nerf bears because they're so awesome" post for awhile?

So why am I so high on druids?

It's not because of EH, at least not solely. It's not because of avoidance being nerfed or that I think avoidance sucks. It's that I think that relative to other tanks, druids improved significantly. Whatever you think of EH, after this change EH is better than what it was. Whatever you think of avoidance, after this change avoidance is worse than what it was. Both of these things favor druids.

Now, it's possible that icecrown has a ton of tank-stressing fights that so happen to stress high avoidance and high block values. That's a possibility, and I don't deny it; I've often said that the best tank is not determined by the tank's values, but by the encounters.

At the same time, GC's come out and said that tanks are going to be dying in Icecrown due to taking huge amounts of damage. And really - what tank is better at taking huge amounts of damage than a bear?

I'm not saying bears are the best tank. I don't know enough about Icecrown's fights to say that reasonably. But I am saying simply that these changes improve the things bears are good at and reduce the things bears are bad at. That's it. That's a far cry from most of the druid population that I read, who continually were worried that bear tanking was dead because they were losing all their dodge.

Though let's go back to that 'best tank' thing. It's true that they're not dominating progression, though I'd say that a druid MTing Anub'arak for most of the top guilds in the world indicates strongly that this isn't entirely true. General opinion is that paladins are the #1 tank...but they just lost 20% avoidance. And developers have stated things like 'we think block tanks are going to be the best tank in Ulduar because of large physical hits', and they've also stated that they're worried about druid levels of power but aren't going to nerf them significantly because there aren't enough to care.

I'd take that general opinion with a big grain of salt. I'd take what developers say with more. Basically, a high sodium diet should be in your future. :)