[top]The cost of control and Stat weights

Wrathblood · 09/06/12, 12:01 PM

Oops, accidental double post.

Sagemor · 09/06/12, 2:37 PM

I actually find myself casting WoG often enough. But I use my paladin for the most part when A) raid NEEDS a tank just to happen, or B) running heroic HoT 5 mans. I play dps so much that I do not often care to raid tank. So I guess the point of this post is that it varies from situation to situation. I often have macros for tanking that utilizes putting my healer as my focus, and so that when I right click certain spells (heals, cleanse, LoH, Hand of Sacrifice, AoE taunt before it was removed, etc) it would cast on the healer. I also picked up Divine Purpose for the 75 talents, which I must say in a couple of fights procced A LOT. Seriously. I had 4 ShoR back to back off of just 3 HoPo for the first one. Then followed it up with a WoG on myself cause I could.

Edit: Read it and fixed a few grammar issues.

Tyvi · 09/07/12, 1:06 PM

Originally Posted by Wrathblood

Edit - I'm afraid I have to disagree with your take on HoPo generation, though. I don't think there's an "enough" level for HoPo generation. While its perfectly fine, advisable even, to "bank" extra HoPo in case of emergency, you really want your ShoR uptime to be as high as possible. And while there are times when WoG is the right move, ShoR should be the default use for HoPo and should consume the vast majority of our HoPo.

If WoG or another reactive move was our main HoPo use, then I might agree with your take on avoidance, but its not. ShoR is primary and its proactive. You can't know ahead of time whether you will Dodge an attack and thus not need to use ShoR against it. That's how DKs work, not Paladins I'm afraid. Rather, you want to just keep ShoR up for as many attacks as possible and hope for the best.

This is why the Avoidance stats and the Mitigation/Control stats don't really play well with each other, by the way. Its nice to avoid an attack and its nice to have ShoR reduce damage by more, but you can really only do one or the other at a time which is why those stats make each other worse, not better.

Yeah, I agree there is a good chance I might be overestimating the reactive nature of the Prot Paladin active mitigation because that is what I am used to from playing my DK. But I am not convinced that it is ideal to just try to increase ShotR uptime without regard of your own health (let's ignore telegraphed attacks for a moment, ShotR usage for that isn't going to change regardless of which method you use). Using a 40%+ damage reduction ability while you are at full health just doesn't seem effective because you are safe as it gets at 100% health. Wouldn't it make more sense to use it when you are, say, half health? As long as you don't die with the ShotR buff up, you don't waste damage mitigation and doing it this way allows you to make your survival more likely.

Maybe comparing ShotR with Shield Wall will help illustrate my point further: Do you think that using SW at full health is better than using it when you are low and truly need it (assuming you don't wait long enough so that you actually die with the SW buff up because you used it at too little health for the damage reduction to actually help you survive the next swing)?
It also makes avoidance easier to swallow if it happens while ShotR is up. If you avoid swings at full health with ShotR up it feels like a waste. If you avoid swings at low health even with ShotR up you still achieved the goal of buying more time for your healers to bring you back to full health, didn't it?
(Time to live doesn't change for either scenario but you shift your ShotR usage to where you are most likely to die.)

Charybdis · 09/07/12, 5:34 PM

I'm not so sure Shield Wall versus Shield of the Righteous is an effective comparison since SW is an ohshit cooldown while SotR is closer to a rotational ability. Having played both DK and pally, I'm not even sure if there's a good equivalent rotational ability between the classes.

As far as Shield Wall (and similar abilities like Icebound Fortitude and Guardian of Ancient Kings) go, the choice on whether to use them at high or low health will depend on the situation:

The pros of using CDs at high health are more peace of mind for the healers, which might be enough to keep you from getting too low in the first place. This also has the advantage of mitigating damage before it even happens, and as such is proactive. This might also let the healers conserve mana for when the shit does manage to hit the fan.

The cons of doing so might be that it's not available if you do start getting pounded and end up with low health during the cooldown, thus needing a different and possibly worse CD to get you by. It might also lull the healers into thinking you're easier to heal, so when an unlucky hit streak happens their reaction times aren't as good.

The pros of using such CDs when low on health are that if the damage is consistent then it means the healers will almost certainly be able to keep up if they're on their toes. It might allow them to use cheaper/slower heals and not have to go batshit crazy in an effort to get you up.

The cons of doing so are that you didn't really lower the damage that got you to low health in the first place, which is probably the most important part. The CD might not even be needed because the burst was survivable without it and you can be brought up to a good health level without needing to change anything. Put another way, it might turn into a reactive ability that didn't really do anything because there wasn't much for it to do.

So which way is best? It'll depend on the situation, not just the encounter itself but how well your healers respond to damage. Based on my experience healing as well as how my healers have communicated while I tanked, I've had better a better experience using CDs while I was high on health. I've generally had good or excellent healers who knew the encounters decently though, so it will probably change based on how well they perform and how well they know what's going on.

If I were to play in MoP, I'd probably use SotR more while I had higher health so that my healer(s) weren't so stressed seeing my health rollercoaster.

Theck · 09/08/12, 7:01 PM

A few comments:

Tyvi: I consider SS a better anti-burst tool for several reasons. One you seem to have already discovered - that the SS absorb bubble is actually larger than the FoL heal that Selfless Healer will grant you. Another is that SS cannot be lost to overhealing, while your FoL can be (as can WoG). SS guarantees ~100k absorb during any burst window, which isn't EH, but plays a similar role. Furthermore, SS costs you a GCD to put up, but doesn't necessarily cost you a GCD during the burst window. SH does, which may mean delaying HP generation that denies you the SotR or WoG that would save your life.

Wrathblood: Regarding exp and haste, recall that stat weights are very sensitive to gear. For a nearly naked paladin, my analytical calculations showed haste and expertise more or less neck and neck as far as TDR value, as is shown on the first two graphs. Stack some hit and expertise, though, and haste pulls ahead of exp. Add some mastery and exp falls even further behind. But add some haste, and suddenly exp gets better and catches up, surpassing haste once again.

This is part of the reason I took the time to write the spreadsheet I linked earlier. It's going to be nigh-impossible to come up with one set of TDR stat weights that's "true" for all gear configurations. Even two different paladins using the same "control" scheme may have different stat weights due to having picked up significantly different pieces.

I think the general point, however, is that if you're following a "control" strategy, you're not doing it for TDR anyway. You've already decided to toss out the high-TDR stats (dodge, parry) in favor of low-TDR stats that give you more control. As a result, I don't think it's necessary to use TDR to justify your stat priorities in the control scheme - it's clear that you want hit and expertise to smooth HP generation and give you control, and it's clear that you want haste to increase your generation rate and improve SotR uptime. I think that exp>haste in that scenario simply because a missed HP generator at the wrong time is far more destructive than losing a percent or two worth of SotR uptime. Expertise gives you more control than haste does, in short.

Other news: I've finished running some Monte-Carlo sims that support the analytical results. I'll be posting them this week. I'm hoping to be able to address the "smoothness" question with them as well, because now I can generate statistics regarding the moving average of DTPS to characterize spike damage events (and, for example, show them on a histogram so we can see how things change by stacking different stats).

Also, it looks like Eternal Flame has received a stealth buff. The HoT is now being affected by Bastion of Glory.

Wrathblood · 09/08/12, 10:49 PM

Theck,

Good points as usual. First off, SS. Am I crazy or is this ability borderline (or possibly not so borderline) OP when looking at how it scales with Vengeance? I'm a little disappointed at the lack of tank comparison theorycrafting going on as it makes me feel on shaky ground making statements like that which shouldn't be hard to verify one way or the other.

I can't recall if I mention it, but I actually came up with my weightings by plugging my own gear into your spreadsheet as a starting point, and then varied different stats by various amounts and noting how the stats impacted each other both in degree and magnitude. I do note that weightings can vary based on differing gear, direct people to the spreadsheet, encourage them to plug their own stats and see how they shake out but I probably don't make the point as firmly as I should have. I know that people have a strong tendency to ignore disclaimers, latch onto any numbers presented as universally applicable gospel and run with it. Which is bad. On the other hand, I feel like I'm letting people down if I don't at least give them some idea of how things look, even if there's a risk of misinterpretation of the results.

Regarding a Control strategy, I totally agree that you wouldn't be following that for the damage reduction, but I did feel it was important that people had at least a ballpark idea of how much damage reduction you're giving up by following that strategy. The difference between it and Avoidance (on average) is actually quite significant. Doesn't mean Control isn't generally better, but as with everything, there's a price to be paid for not missing those CS's.

I take issue with your comment about missing a HoPo generator being critical, though. Certainly, it *could* be critical (as with the Impale example I gave) but for progression content, I argue that you frequently don't know in advance when those critical moments are doing to be (Slashlove and Celyndrashad and an interesting debate on this in the WoW tanking forum).

You know when the better telegraphed spikes will come (Impales, Morchok stomps, etc) with a decent amount of accuracy. While its possible to mis-time things, once you have a basic knowledge of the fight, accumulating sufficient HoPo to prepare for these situations isn't very challenging even for someone with minimal Hit or Expertise. For the spikes that aren't telegraphed as well (or are caused by non-damage mechanics, such as mechanics that cc healers or force them to move), Control can help by ensuring your next generator hits, giving you the HoPo for a ShoR to follow-up the WoG.

However, you can get most of the way there by simply making of habit of building up to 5 HoPo before ShoRing. If you're at 5 HoPo, 7 times out of 9 the Avoidance strategy will give you the same result as Control. Additionally, IMO, being able to fire off two finishers quickly is small potatoes compared with getting lucky and having the crisis occur during ShoR uptime. Because the damage is already done. Obviously, big mitigation is a lot less useful when you're already low on health from an EH-boosting standpoint. Using two finishers is great for bailing you out, but its better to avoid the situation in the first place.

I believe that maintaining the highest possible ShoR up-time is the best way of doing that. I suspect that anything which increases your ShoR uptime is going to be extremely valuable for survivability. However, my last couple paragraphs are shot through with "I think" and "I suspect" because while I can math out averages and guess at distributional effects, I'm very curious what your Monte Carlo work says because I hope it'll flesh out the results (especially the variability in damage intake) in illuminating fashion. I think I know how things will turn out, but I recognize its possible that I'm wrong and will have to change my tune in short order.

Edit - I've very curious to hear if you're planning on pushing for hard capping EXP or not. From a theoretical standpoint, maximizing HoPo generation and being able to perfectly deliver a critical HoPo requires going all the way to hard cap. On the other hand, broadly speaking, your results variability will certainly drop the higher you get and I'm curious to see if there's a point at which your variability is low enough that its simply not worth getting all the way there. I wonder if there's a sweet spot at like 12% Exp where you get 97% of the benefit with only 80% of the itemization cost and it turns out to be worth stopping there.

Charybdis · 09/09/12, 7:23 PM

I wouldn't worry about giving people exact numbers because they wouldn't be accurate anyway. The setup you currently have going over things like Hit>Haste>Mastery>Expertise should be fine as well as pointing them to the spreadsheet and saying "The values you get are for your unique setup and might not be anything like someone else's setup. That's just how the game works."

You could also emphasize the fact that there are no universal stat weights with big bold red letters so that a person would almost have to be blind to miss it. That or a troll decides to go "lolwut? me no get it, need numbers!" and then get a well-earned infraction.

Exemplar · 09/10/12, 8:56 AM

A little late to the party, but here's a non-Prot perspective on a few things (I've played a healer a lot of my time in WoW, and listened to them closely the remainder).

SS is powerful compared to the other abilities due to it being passive, not despite it. Think about it this way - yes, SH or EF may arguably be better for total healing/absorbtion/whatever. However, realize that by the time you cast WoG/EF/FoL on yourself, your healers have already picked their next spell and begun casting. If you spiked low enough that WoG/EF/FoL is truly useful, your healer probably has already started their fast-manahog spell. By the time they may realize you don't need it, too late, mana is gone. No one wants to hear "Oh, I didn't heal you because I thought you were going to self-heal" after a wipe, so the healer feels they must heal. Thus your self-heals are either overheal or causing your healers to overheal/over-manaspend. SS being passive helps reduce the spikes that cause a healer to panic, and doesn't make the healer second guess when they do swap to a fast heal.

Similar with tanking cooldowns. If you use a 40% damage reducer at 100% health, you're less likely to spike low and cause healer worry. Healers can potentially regen, use a slow heal, or perform some other action knowing you can survive the very next attack. If you wait to low health to use the damage reducer and it keeps you alive, your healers are still panicking and working to get you healthy enough to survive the next blow ASAP.

The total effectiveness of a damage reducer remains the same whether used at 100% or 10% (as long as you survive), but using it at the high end of the spectrum is a quality of life improvement for your healers. Also you don't have to worry about waiting a hair too long and popping the CD at X% and dying by overkill of 10 damage. As long as it's not wasted (popped during period where incoming damage is very low, or used during normal phase when a known damage spike phase is upcoming),

If any spec can understand that 'it's not always about what's best for me, but what's best for the raid', it would be the tanking specs. After all, you dedicate your lives to not topping either the DPS or the Healing meter.

Wrathblood · 09/10/12, 7:26 PM

Decided that I wasn't nearly clear enough that the stat weights were based on my own gear and that people should really work out their own. Plus, I couldn't get a combo of readability and usefulness on stat weights that I was happy with. So instead, I took out the section and I'm putting it here in this post. If someone really wants to see some sample stat weights, they're quite findable without causing issues in the OP.

[top]The cost of control and Stat weights

Lets give some perspective on what's at stake in deciding between Avoidance and Mitigation stats. Sure, giving up Parry for Hit will give you more damage smoothing, but how much extra damage will you take? In a nutshell, every 100 itemization points you put into Hit, Haste or Expertise instead of Strength will cost you about 0.4% damage taken. Hit is a little better, while Haste and Expertise are a little worse.

Reaching 7.5% Hit and Exp will take about 1525 points. If you were able to take all those points and put them into STR, you'd take a little more than 6% less damage on average.

Damage Reduction Stat Weights
1.00 STR
0.88 Parry
0.88 Dodge
0.75 Mastery
0.47 Hit
0.36 Haste
0.30 Expertise

Roughly speaking, each 0.7 of weighting is worth 0.5% damage per 100 itemization points. Also remember that stat weights will varying based on specific gearing and these are averages. Theck has kindly created a spreadsheet for you to calculate your own and its accessible here https://docs.google.com/spreadsheet/...ESkVXb0E#gid=0

Edit - Also, going to start redoing this thread for MoP launch. Will be heavily based on this thread, but edited to be 100% focused on level 90 play, and will be its own new thread. Hoping to have it up middle of next week.

bubblecannon · 09/14/12, 10:52 PM

I've been trying to test out the SS absorb and it seems weirldy tuned (if I can trust the values from SPELL_AURA_APPLIED and SPELL_AURA_REFRESH):

At 8363 AP absorb amount is 4922
At 4061 AP absorb amount is 1472

Assuming base + percentAP works out to 30 + 0.585*AP.

I also did some testing to see when the absorb used the vengeance value - it appears only to use it on a manual reapplication rather than a refresh. (i.e. when you recast the spell manually)
(Edit: for clarity - the SS 6 second shield size is calculated on SS manual cast - automatic refresh uses the original value and ignores current AP)

This seems to greatly encourage using a vengeance tracker and reapplying on either of two conditions for some quite considerable changes in absorption:
* always reapply when current vengeance AP > when cast (use some threshold to decide)
* when vengeance is about to drop

Given how easily vengeance can spike from burst this seems a very easy way to game SS usage.

Iroared · 09/18/12, 8:15 AM

DP: 40% 51%
HA: 39% 49%
SW: 35% 44%

Do the ShoR uptime calculations take into account the fact that the buff doesn't overwrite itself - instead, it adds 3 seconds to the current duration? I've done some quick math and it seems HA is undervalued in the small uptime table.

If 41% is the untalented uptime with 100% hit/exp, simulating HA as 100% uptime for 21 seconds every 120 seconds (since it would last 3 seconds beyond the 18 second duration) yields 51% average uptime. Which is pretty close to the 49% in the table.
If, on the other hand, we use 4.4 CS, 3.3 J and 0.88 AS(GC) during HA (calculated as 18/4.5, 18/6, and 18/4.5*0.2, plus 10% haste raid buff), we get a total of 8.58 casts of ShoR in an average HA, resulting in 25.74 seconds uptime. This crude math yields a 55% average uptime.

Or am I totally off and this has already been accounted for? Skimming all 4 pages, I didn't find any mention of that.

Wrathblood · 09/18/12, 2:51 PM

I did indeed assume the +3 seconds rather than overwriting. I don't have my spreadsheet here, but I believe I worked out that you'd be able to average... hmm, either 5.5 or 6, I forget which, ShoRs per 18 seconds of HA and went with that.

Theck · 09/18/12, 6:54 PM

Originally Posted by Iroared

If, on the other hand, we use 4.4 CS, 3.3 J and 0.88 AS(GC) during HA (calculated as 18/4.5, 18/6, and 18/4.5*0.2, plus 10% haste raid buff), we get a total of 8.58 casts of ShoR in an average HA, resulting in 25.74 seconds uptime. This crude math yields a 55% average uptime..

Note that you don't actually get 18/6 Judgments. In a CS>J rotation, Judgment's effective cooldown is 6.75 seconds.

Wrathblood · 09/18/12, 7:25 PM

Still haven't gotten to check what I did in Excel, but I'm at least sitting now. With perfect play, over the course of the 12 GCDs of HA, with a little luck, you should be able to fit in something like: CS - J - AS - CS - Cons - J - CS - HW - AS (GC proc) - CS - J - Cons.

Assuming you're ShoRing instantly, you'd be able to fit in 8 ShoRs. Between having sub-perfect reflexes, potentially not getting a GC proc, and possibly not having your rotation lined up ahead of time, you could reasonably get as few as 6 ShoRs during that time. On the other hand, with perfect luck, you could get as many as 10.

If you're hit/exp capped, you should be getting about 41% uptime during that period anyway. So, if you're getting off 7 ShoRs, that's 21 seconds of uptime, during which you'd have expected to have perhaps 8.6 seconds of uptime time already, for a gain of 12.4 seconds, just over 10%.

I think I know what I did. I set the HA period to 12 GCDs and didn't account for HA procs affecting uptime out of that window. I'll fiddle with it a bit when I get home, but I suspect 51% is more accurate (I'd be willing to assume you can get 7 ShoRs off consistently during an HA, but 8 seems slightly optimistic). In fact, thinking back on it, I believe I originally had DP and HA tied at 51%, but during an edit I bumped HA down to 49% for reasons I'll have to look up.

Edit - Finally got to check, and the bounding was indeed the issue. Not sure where I got 5.5-6 in the earlier post (going off-the-cuff usually ends badly), I actually used 7.5 which would actually put HA uptime at 52%, above DP.