T15 2-piece: Casting Word of Glory or Eternal Flame also grants you 40% (additive, not affected by DR) additional block chance for 5 sec/Holy Power. Buff duration is not additive.
T15 4-piece: You gain 1 Holy Power for every 20% of your health taken as damage (after all mitigation, but counting absorbed damage) while Divine Protection is active.
Today we’re going to look at how these bonuses will affect our play in 5.2.
The 4-piece is interesting for a few reasons. First, it turns Divine Protection into a resource generator. While we have it up, we’ll be getting holy power in chunks whenever we take damage. Unfortunately, the amount of holy power is going to vary significantly based on the content you’re raiding. Heroic modes are going to grant more than normal modes, and 25-man raiders should see a higher benefit than 10-man raiders. It may very well be that the set bonus is highly prized by 25-man heroic guilds and nearly ignored by 10-man normal guilds. For 25H raiders, I wouldn’t be surprised if it turns Divine Purpose into another Holy Avenger, which would be an amazing boost to Shield of the Righteous uptime (though curiously, SotR being up diminishes the effect!).
It also has the potential to shake up our rotation a bit since that extra holy power comes in discrete chunks, and at unpredictable times. You may normally be saving up 5 holy power before firing off SotR, but during Divine Protection you’d be at risk of wasting the holy power granted by the set bonus that way. It’s likely that during DP we’d shift to casting SotR at 3 holy power to prevent overages.
Much like early iterations of uncapped Vengeance, basing the effect on post-mitigation damage might lead you to think that it encourages gearing for less mitigation. Except that in this case, you’re doing so to get more holy power and cast more SotRs, giving you more mitigation. It’s got an interesting built-in diminishing returns of sorts: as you take more damage, you get more holy power to feed into SotR to reduce that damage, lowering the holy power gain. So I’m doubtful that it’s advantageous to “tank without pants on” for the sake of the set bonus.
However, it will generally generate more holy power for a control/haste gear strategy than for control/mastery or avoidance gear strategies. Remember that control/haste takes more damage than either of the other gear schemes, but that damage intake is smoother. That means more holy power generation, which is good, and potentially smoother holy power generation as well, which may mean less ends up wasted as overage.
Another way it seems to affect itemization is to slightly disincentivize stamina since the effect is based off of percentage of health. For example, having a little less stamina may turn a 1-holy-power gain into a 2-holy-power gain, assuming the calculation is done per-attack instead of cumulatively. I don’t think this is enough to make Stamina any less attractive to the heroic-progression raider, because earning less holy power during Divine Protection is a lot more attractive than being dead. But stamina is already much less important to players farming normal modes, especially in the 10-man format, and this set bonus may further reward them for focusing on secondary stats instead of stam-stacking.
One thing I can confidently state about the 4-piece bonus is that it will be more effective during high-damage periods than low-damage periods. Popping Divine Protection while you’re taking manageable damage will reduce your total damage taken, but won’t really make a meaningful impact on your survivability. Despite the natural diminishing returns effect, you’ll get more mileage out of this set bonus by lining Divine Protection up with high-risk periods.
Because of this, I think the dev team’s concerns about this set bonus incentivizing Unbreakable Spirit aren’t going to be problematic in-game. Certainly making Divine Protection available 33%-50% more often makes the talent even more attractive than it already was. But I also don’t think that “spam Divine Protection on cooldown” will end up being the ideal way to use the ability. It may be the default choice on normal modes, or for players who really care about TDR (why??), but on heroic modes I’ve found that I’m often choosing Clemency instead.
Aside: In fact, when Mel and I first saw Clemency announced at BlizzCon 2011, the first thing he remarked about was that it was so broken it would never make it live. I disagreed with him, of course (it’s just what we do). It turned out we were both right. The talent did make it live, and it’s just as broken (in a good way) as Mel predicted. Being able to double-BoP and double-HoSac is really good, to the point of breaking some encounters. Blade Lord Tayak is a great example – in a raid with 5 paladins, all talenting Clemency, the Wind Step mechanic becomes basically irrelevant. You have 10 Hands of Protection with which you eliminate the debuff before dangerous periods like Unseen Strike or Blade Tempest, and almost unlimited Hands of Sacrifice to cover everything else.
The 2-piece bonus poses a quandary: is spending 3 holy power on WoG to gain 40% block chance for 15 seconds better or worse than using it for 3 seconds of SotR? That’s a tough question to answer because it depends on how you value WoG itself. Are we using it when we’re already at full health just for the sake of the 2-piece buff? That “100% overheal” estimate isn’t completely fair, because the heal certainly won’t always be overheal. On the other hand, assuming 0% overheal is just as unrealistic to the opposite extreme.
Unfortunately, I think the true amount of overheal is likely to be closer to 100% than 0%. Your healers are playing a predictive game, queuing up heals 2 to 3 seconds before they actually land. Dropping an instant WoG during that process isn’t always useful. If their heal tops you off before the WoG lands, your WoG ends up being overheal. If your WoG lands first, all you’ve done is turn a chunk of their heal into overheal. It’s only when you’re low enough on health that all of the healing you take – your WoG, the healer’s long-cast-time heal, and any incidental healing from HoTs and such – that all of it ends up being effective healing.
However, if we make some assumptions about WoG overheal, we can estimate the effect of this set bonus. I’ve modified my earlier Monte-Carlo simulation code (link) to handle the set bonus, which was relatively easy. The more difficult part was implementing WoG. Remember that we don’t track tank health in this simulation, just damage intake. Thus, I had to come up with another way to simulate the effect of WoG healing. The most straightforward analogy is an absorption bubble like Shield Barrier, since WoG essentially offsets part of a boss attack. However, I’ve kept the duration short in order to make sure we can’t “save up” WoG absorption to apply to a spike too far down the line. My approximation here is to give the fake WoG bubble a 3-second duration. That way it’s an effective counter to a boss attack in close proximity to the cast, but ends up wasted if we avoid the next two attacks.
The default rotation we’ve been using in earlier simulations is just spamming Shield of the Righteous as much as possible, so that will be our gold standard. We’ll call that rotation “S” for simplicity. We could also imagine ignoring SotR entirely and spamming WoG with all of our Holy Power. That rotation, which we’ll call “T” for “tier bonus,” will use WoG any time the tier bonus buff drops off, as well as if we hit 5 holy power so that we don’t waste anything.
The simulation tracks BoG stacks appropriately, of course, so we can also evaluate rotations that mix SotR and WoG much the same way that we looked at combinations of Shield Block and Shield Barrier in the warrior simulations (link). There are two logical ways I could come up with to do that. The first is to keep the tier bonus up at all times and use SotR as a bleed mechanism rather than WoG. The way we’ll do that is to cast WoG every time the tier buff drops off and only cast SotR when we’re at 5 holy power. We’ll call that rotation “TS” to indicate that it prioritizes the tier buff over SotR.
The second is a “first come, first serve” style rotation that tries to keep something up at all times, whether it’s SotR or the tier 15 set bonus. It will prioritize SotR over the tier bonus, so we’ll call it “ST” to indicate that. The logic will be roughly “if neither buff is up and HP>=3, cast SotR; if neither buff is up and HP<3, try to cast WoG.”
You might ask why we’re not considering alternating queues, like “SotR-WoG-SotR-WoG.” The problem with those sorts of queues is that holy power generation isn’t steady. During the 5-second duration of a 1-holy-power tier buff, we could get anywhere from 2-4 holy power. As a result, those rotations are going to be very inefficient. They’re either going to overlap SotR and the tier buff and have gaps in the coverage of both, or they’re going to end up refreshing an existing tier buff early, which is wasteful since the duration isn’t additive. The TS and ST rotations employ logic that avoids these problems, and as a result should be more efficient than any fixed cast sequence.
Let’s see how this all sims out. I’m going to use all of the same metrics as in previous installments, if you want clarification on what they mean I suggest re-reading the definitions in this post. For a gear set, we’ll use the control/haste setup since that performed best in the last round. The stats on that set are:
| Set: | C/Ha | | Str | 11000 | | Parry | 2000 | | Dodge | 2000 | | Mastery | 1500 | | Hit | 2550 | | Exp | 5100 | | Haste | 4000 | | Armor | 60000 |
First, let’s take the most pessimistic view of WoG and assume that it performs 100% overhealing. That means WoG is doing nothing but applying the tier buff. This is the “worst-case” situation for the tier bonus, of course, but if it beats SotR spam in this simulation, it will be the unquestionably better choice.
WoG overheal factor: 100% | Queue: | S | T | TS | ST | | S% | 0.4540 | 0.0000 | 0.2575 | 0.2794 | | mean | 0.5794 | 0.6883 | 0.5839 | 0.5829 | | std | 0.1321 | 0.1479 | 0.1487 | 0.1423 | | ------ | --- 2 | Attack | Moving | Average | | 80% | 10.1178 | 45.2993 | 22.9590 | 21.1210 | | 90% | 6.8373 | 11.4510 | 5.7840 | 5.3520 | | ------ | --- 3 | Attack | Moving | Average | | 80% | 10.7772 | 21.0260 | 10.4550 | 9.5852 | | 90% | 0.9210 | 3.8613 | 1.5420 | 1.3373 | | ------ | --- 4 | Attack | Moving | Average | | 80% | 2.4223 | 26.1447 | 9.7880 | 8.4615 | | 90% | 0.0000 | 9.0337 | 2.7805 | 2.1452 | | ------ | --- 5 | Attack | Moving | Average | | 80% | 2.3230 | 27.5950 | 7.8373 | 6.0060 | | 90% | 0.0000 | 3.7158 | 0.8937 | 0.6600 | | ------ | --- 6 | Attack | Moving | Average | | 80% | 0.3473 | 16.4365 | 4.5367 | 3.4055 | | 90% | 0.0000 | 1.4808 | 0.2637 | 0.1745 | | ------ | --- 7 | Attack | Moving | Average | | 80% | 1.2698 | 18.7365 | 3.1425 | 2.1338 | | 90% | 0.0000 | 2.9270 | 0.2640 | 0.1210 |
So, those numbers are a pretty strong indication that the tier bonus isn’t worth giving up SotR for. The T rotation takes significantly more damage than the S rotation, and performs worse in every spike category, in many cases by huge amounts. The closest it ever comes is a factor of 2x worse, but it’s up to 50x worse in some categories, and that’s not counting the situations where SotR eliminates spike representation (because you can’t divide by zero). Maybe this rotation will improve with some of WoG’s power restored to it, but it’s got a long way to go to catch up.
The TS and ST rotations don’t perform much better. They’re both worse for 2-attack strings, though they manage to do about as well as S in in the 3-attack category. However, they quickly fall behind for 4+ attacks. There’s just really no comparison here in which they shine compared to SotR spam. Spending any holy power on WoG for the sake of the buff seems counter-productive if you’re not getting anything out of WoG.
Before going on, let’s just quickly look at what the damage profiles look like for these rotations. First, let’s look at S:
This is just like it did last time. The combination of hit/exp caps and haste eliminates spikes over 90% throughput and does a pretty good job of suppressing 80% spikes. That gives us a scattered grouping of attacks falling in the 30%-80% range, skewed slightly towards the higher end. Now let’s look at what happens when we use the T rotation instead:
As the statistics suggested, this doesn’t do anything interesting for us. It doesn’t push the mean down, unlike Shield Barrier does in the warrior simulations. It also doesn’t do much to eliminate spikes, as the representation of 80% and 90% spikes is still very large. It just doesn’t do much of anything.
The ST and TS rotations have a more interesting damage profile though:
Both ST and TS look pretty similar here. ST is a little better at spike elimination because it prioritizes SotR more highly. But the interesting part is how much different they are from the SotR spam rotation. Instead of a scattered distribution, we’re getting something that has a strong resemblance to a bell curve. That’s primarily due to the extra 40% block being up so often. The significant increase in block chance creates a larger variation in the size of 5-attack strings because we’re far more likely to get multiple blocks mixed in with our avoids and SotR events. It’s sort of a central limit theorem thing happening.
Anyway, so far there seems to be no significant advantage to prioritizing WoG for the set bonus at all. However, that was also the worst case. Let’s be more generous and assume that only half of WoG’s healing ends up being overheal. How does that change the results?
WoG overheal factor: 50% | Queue: | S | T | TS | ST | | S% | 0.4543 | 0.0000 | 0.2574 | 0.2797 | | mean | 0.5783 | 0.6060 | 0.5400 | 0.5462 | | std | 0.1325 | 0.1476 | 0.1473 | 0.1361 | | ------ | --- 2 | Attack | Moving | Average | | 80% | 10.1180 | 28.8380 | 17.5155 | 12.6595 | | 90% | 6.7907 | 5.8715 | 4.3460 | 3.3627 | | ------ | --- 3 | Attack | Moving | Average | | 80% | 10.7220 | 9.0195 | 7.1772 | 7.1858 | | 90% | 0.8855 | 1.1638 | 0.9507 | 1.0065 | | ------ | --- 4 | Attack | Moving | Average | | 80% | 2.3863 | 10.8043 | 6.0877 | 3.6110 | | 90% | 0.0000 | 2.2140 | 1.3308 | 0.4452 | | ------ | --- 5 | Attack | Moving | Average | | 80% | 2.2545 | 11.4547 | 3.7963 | 2.6248 | | 90% | 0.0000 | 0.4497 | 0.3400 | 0.0850 | | ------ | --- 6 | Attack | Moving | Average | | 80% | 0.3175 | 4.5735 | 1.8378 | 1.3150 | | 90% | 0.0000 | 0.0930 | 0.0633 | 0.0550 | | ------ | --- 7 | Attack | Moving | Average | | 80% | 1.2650 | 5.3187 | 1.1753 | 0.6032 | | 90% | 0.0000 | 0.2230 | 0.0405 | 0.0187 |
Hm… much better, at any rate. The T rotation is still garbage, unfortunately. TS isn’t a whole lot better, still trailing S in almost every category. But the ST rotation is starting to catch up in the longer-queue situations. One could argue it’s a little better for 7-attack strings, but it still trails in the majority of categories. So at 50% overheal, there’s still no distinct advantage to prioritizing the set bonus. Let’s see what happens at 0% overheal:
WoG overheal factor: 0% | Queue: | S | T | TS | ST | | S% | 0.4541 | 0.0000 | 0.2570 | 0.2794 | | mean | 0.5794 | 0.5219 | 0.4989 | 0.5093 | | std | 0.1319 | 0.1522 | 0.1505 | 0.1342 | | ------ | --- 2 | Attack | Moving | Average | | 80% | 10.1017 | 23.1373 | 15.6115 | 10.5695 | | 90% | 6.8475 | 5.8565 | 3.9105 | 2.4470 | | ------ | --- 3 | Attack | Moving | Average | | 80% | 10.7855 | 6.2480 | 6.1457 | 3.3335 | | 90% | 0.9250 | 1.1413 | 0.8900 | 0.3090 | | ------ | --- 4 | Attack | Moving | Average | | 80% | 2.4060 | 4.2743 | 4.0332 | 2.3253 | | 90% | 0.0000 | 1.0760 | 0.9870 | 0.1777 | | ------ | --- 5 | Attack | Moving | Average | | 80% | 2.3395 | 0.9550 | 2.0130 | 0.7100 | | 90% | 0.0000 | 0.0815 | 0.1878 | 0.0455 | | ------ | --- 6 | Attack | Moving | Average | | 80% | 0.3292 | 0.9020 | 0.7195 | 0.3733 | | 90% | 0.0000 | 0.0000 | 0.0227 | 0.0055 | | ------ | --- 7 | Attack | Moving | Average | | 80% | 1.2673 | 1.1312 | 0.3800 | 0.2043 | | 90% | 0.0000 | 0.0000 | 0.0023 | 0.0008 |
Aha. Now we’re starting to get somewhere. First, we see that even the T rotation has pulled ahead in TDR at this point. It still trails in most of the relevant spike metrics, but not all of them anymore. TS still doesn’t look that attractive, but ST now seems to really compete with S. It tends to permit more 90% spikes while reducing the 80% spike numbers significantly. I think you could make good arguments for both S and ST based on these numbers.
Unfortunately, I also think this situations is pretty unrealistic. It’s basically impossible to operate in the no-overhealing limit in a real raid environment. Your World of Logs reports will probably tell you that your actual healers are doing at least 30% overhealing when you’re not using WoG. Chances are your own overhealing is much smaller if you’re saving WoG for low-health situations, but it’s probably still non-zero just due to cases when your healers react to your low health the same way you do and heal-bomb you. If you’re using WoG more regularly, especially to maintain a buff, it’s likely to push both of those numbers up.
I wouldn’t be surprised if the 50% overheal data is a far more accurate estimate of what we’ll see in-game than either of the extremes. And if that’s the case, the strategy of treating the tier bonus as a maintenance buff just doesn’t seem compelling. It may be a nice temporary mitigation boost for cases where we might use WoG anyway (i.e. low health, high BoG stacks). But it’s not going to be game-changing.
The short short version is: 4-piece is good, and probably worth getting. But the 2-piece is probably at best a “meh.” It lowers the opportunity cost of having to use WoG, which is probably good because SotR is such a huge amount of mitigation to give up.
If the two were reversed, such that “WoG-block” were the 4-piece, we would probably all skip it. Since we’re likely to want the Divine Purpose 4-piece though, we’ll get WoG-block whether we like it or not.
In any event, don’t plan on changing up your rotation based on the 2-piece bonus – just stick with your usual “Spam SotR and WoG like a fiend if I’m about to die” and you’ll be playing optimally. The 2-piece bonus just makes those situations where you need to WoG to survive the next attack feel a little less punitive.