Hi,

Last night, I did some quick math regarding resto trinkets (and also tested them).
I have got these trinkets now:
Spirits of the Sun (496)
Qin-xi's Polarizing Seal (502)
Horridon's Last Grasp (522)
Inscribed Bag of Hydra-Spawn (528)

And I was wondering, which one to use. Now, I'm running with Qin-xi's and Horridon's, but maybe hydra-spawn would be better? It's +26 item levels.... I did some rough calculations and figured out, that qin-xi adds, on average, +400k to healing rain per encounter (7minutes, 20% spell coefficient, 16k base tick) and this hydra-spawn bag only 270k absorb (1.2 RPPM, 12.5% haste - on average 1 proc every 45secs). Although it has over 300 more spirit.

And from my experience, I have found out this:
- Horridon's Last Grasp is likely to proc at the beginning of fight, thus wasting the proc. You can solve this by re-equiping trinket right before pull, because it puts it on 30sec ICD. It can also stack, I have very often seen 2 stacks and sometimes even 3. Not more than that though.
- Inscribed Bag of Hydra-Spawn - haven't tested this on raid yet, but my fellow monk healer did. I think the proc should be similiar for both classes (logs from Megaera: Details for Dudlîk - 14-03 19:21 - Paradox - World of Logs )

I will probably be using Horridon's Grasp, because if you look at the logs and click at me (Details for Zeef - 14-03 19:21 - Paradox - World of Logs ) it gave me 76k mana. Over 9 minute fight, it's roughly 700mp5. Along with the static 1467 intellect, I think it's really good trinket. Qin-xi has similiar mp5 (682, not counting MTT) but higher intellect (1624). So my guess is to stick with qin-xi and horridon's.

What are your opinions?

EDIT: Tonight, we finally managed to kill protectors and tsulong on HC, so I had a chance to test the hydra spawn trinket. I tested it on protectors, and it was stunning 1% of my healing (Details for Zeef - 16-03 20:23 - Paradox - World of Logs , 326k).

Personally, I would go:
Qin-xi's Polarizing Seal (502)
Horridon's Last Grasp (522)

Looking at the Monk's proc rate / healing done on the trinket, it was only a 0.7% increase in output.
At 30k SP (with another 10k SP equiv from base spell values)... So effectively 40k SP.
To boost the size of your spells by 0.7% you'd need 0.007*40,000= 280 SP.

So while the TF-Hydra Bag has ~300 more spirit, the proc is only worth about 280 SP, while the Seal proc is worth 1624 int on average (I assume you are right, I haven't looked up he number lately).

You can make an argument that the Hydra Bag proc is always effective, but it might be "effective" at a time when a more powerful healing rain would have topped the player off otherwise. Neither trinket will be reliable in how dire the situation will be or who it'll actually proc on. So go with the stronger proc.

Revised HST/HTT haste math.

I made a pair of incorrect assumptions when working out the breakpoints:

1. The tickspeed for the breakpoint wasn't rounded down to the nearest whole number
2. The "adjusted" tickspeed (ie: the one where I look for the 0.5 ms value where the game would round to our required tickspeed) should be the inverse of normal hot calculations

This means that we're looking at slightly modified breakpoints for both totems.
HTT: 2791,11471,19203
HST: 3764,8916,14044

Further to this, I suspect at low latencies you can get a slight negative impact on the totems duration depending on how long the "create" vs "destroy" messages take to occur. As latencies increase, this duration variation stops having the potential for negative values, and increases the chance for additional ticks as we've observed previously. Threya's testing with low latency had her off the HST breakpoints by 3 rating for +2 and 20 rating for +3, which are fairly minimal amounts, especially when you note that her ping was varying from 51 to 95 where the breakpints are <=5 away from 15000ms.

i've started playing wow again and was looking at some of the new trinkets.
i noticed that some of the math done for the trinkets were incorrect, the ones which use RPPM.

One example is Horridons last gasp (ilvl 522).

The mana return is listed as 828 MP5 @ 20% haste.

The actual expected mana return is 910 MP5 @ 20%
and 828 MP5 @ 10%

Some proofs and examples are in a document attached.
(You can find the examples in bordered boxes)

Although I havent done the examples of other trinkets, the paper should shed some light on RPPM. (I have added notes on how the others can be calculated)

https://docs.google.com/file/d/0B6Ee...it?usp=sharing

As a side note:
I did take a look at the worksheet provided to how therya calculated mp5 for trinkets.
all that needs to be changed it the "Chance to proc" column.
as the values in there is actually the rate parameter and not the probability.

I calculated the trinkets a while ago and I haven't updated the table after the 5% increase they got which is where the difference actually lies I think.

i'm not sure what you mean by that. but from what i understand, the "unlucky streak" prevention was for item droppings in lfr and quite possibly has nothing to do with this.

if you are referring to that horrible post about p = RPPM*(1+Haste)*seconds_from_last_proc/60
it's a laymans explanation of what the poisson process does already with the rate.

Real Proc Per Minute (RPPM) Trinket Changes - Forums - World of Warcraft

"Multiply the proc chance by MAX(1, 1+((TimeSinceLastSuccessfulProc/AverageProcInterval)-1.5)*3). For example, if a proc has an average proc interval of 45 sec, and it’s been 72 sec since your last successful proc, you’ll get a 1.3x multiplier to your proc chance. If you’ve been out of combat for a few min, and it’s been 5 min since your last successful proc, you’ll get a whopping 16.5x multiplier to your proc chance."

I understand the Poisson process, but this is a new factor on determining proc rates of trinkets.

Really good change IMO. Will make us more viable on stack fights, although we will still be near useless on fights like dark animus Would be worth to re-calculate CH and HR viability (for example, if it's now viable to use HR on 3 people or CH on 2 people).

I don't know if the 20% buff applies to base value or scaling, so I will apply it afterwards (it should be similiar results)
According to wowhead, healing rain is 1983-2358 + 19.7% of SP and it costs ~22k mana. Let's average it to 2200 + 19.7% and 22k mana.
With 40k spell power, healing rain ticks for 10k. With unleash elements, it's 13k. With the new buff, it should be 15.6k. Let's say it ticks 6 times, that's 6*15,600*n , where n is number of players standing in it (lower than 7).
I will try to put it in a table
CH is average 6,325 + 57.3% of SP.
GHW 15,181 + 137.7% of SP

NOTE: I haven't calculated crit, resurgence, riptide (chain heal) and earth shield at all, so the results (especially for GHW and CH) will vary.

Heal	Mana Cost	Healing per tick	HPM - 1 player	2 people	3 People	4 People
Healing Rain - NEW	21,981	15,600	4.26	8.5	12.8	17
Healing Rain - OLD	21,981	13,000	3.54	7.1	10.6	14.1
Chain Heal - NEW	13,500	35,094	2.6	4.42	5.7	6.58
Chain Heal - OLD	13,500	29,245	2.16	3.68	4.74	5.48
Greater Healing Wave	16,140	70,261	4.35	X	X	X

For example, if you have 25% crit chance (I do, with 517 item level), the results for GHW will be like this:
70.261*2.68*0.25 = 47k heal gain. So GHW will heal for ~117k on average. Also, it will return 2212 mana per cast = it costs ~14k mana. This will net 8.35 HPM, but we also need to calculate crit for other heals, although for HR it's not that powerful because of not proccing AA and resurgence.

Don't take this table much seriously, it is really not accurate, but it seems quite possible that HR will be efficient even on 3 people (instead of 4) and CH will be usable even when there are 2 targets


Again, please correct me if I am doing this wrong, but I hope that this post will be at least a little bit useful for you

The results in the table below are calculated with :

Int=20241
Crit=25.53%
Haste=25.37%



While we can leave Riptide out of rotation due to its special nature we can conclude the following assuming that every spell is under the influence of Tidal Waves :

1. HW remains the best HPM spell but the worst HPS spell save CH when it hits less than 2 targets.
2. HS remains the best HPS spell but falls behind every other spell on HPM save CH when it hits less than 3 targets.
3. GHW has very strong HPS and strong HPM.While not the best in any case it is the best spell by far for when you want a strong heal that will not cost as much as HS.
4. CH has almost double GHW HPM if it hits 4 targets making is almost as efficient as HW and slightly more HPS than GHW. If 4 targets are quaranteed it's by far a better choice than GHW in both HPM and HPS and even as mana efficient as HW.
5. HR remains our biggest and AOE healing spell. With the recent 20%buff and the old mana nerf it is our best HPM and HPS spell for 3-6 targets if it hits them for full duration without overhealing them. It is better HPM also with as less as 2 targets than every other heal save CH when it hits 4 targets and HW. As for its HPS it's our strongest healing source for 2 targets and above.

Hello to the one who reads this.

I have been looking lately at the ticks from Healing Stream Totem that it was acting so strange lately since the current patch. Now I know about the red warning where HST and HTT depend on world ms when you have >50 MS where it ticks more often not then it should be, but I play on a constant 23 ms. Even if this was the case, here is what I've seen from my own view.

From this log only, I was on 1983 haste, which is far below what was first considered to be the first breakpoint where you gain 9 ticks. So my Healing Stream should tick 8 times. But when I look at the Expression Editor, I've received a constant 9 ticks. Expression Editor WoL 28th March

On MMO-champion, I've asked for a few more if they notice the same thing. Binkenstein responded to it once that this was already been talked out in this topic here, but I do not see anything other then the haste breakpoints were wrong what he first claimed to be and that he gave the new numbers. But the Healing Stream is unchanged, it's still at 3764.

The point I'm trying to make here though is that currently everyone is saying to go for 3764 haste. Taking the two breakpoints of HST and HTT out, there is a lot of room between 871 and 5676 haste, where you can go into a certain amount of haste that feels much more comfortable. Or, if you prefer, possibility to aim for 5676 haste after the 871 haste breakpoint. As right now, it's all dependend that you either have too much haste as you have a good pc and have a good connection or you have too little, while the other haste breakpoints are set in stone.

I don't know if this was meant to be in the Simple Questions/Answers topic, I've looked to be sure, but I still wasn't. So if this doesn't belong here, feel free to move it there.

I hope a more abstract theory-crafting question/idea (albeit one focused on Resto Shaman) won't be unwelcome.

Due to the dynamic nature of healing, stat weights are obviously harder to formulate for healers. Therya has done a number of spreadsheets, but I find them rather difficult to apply directly to reforging and gemming. Resto Shaman have the added complication that the strength of our heals, and the value of one of our stats, depends on the target's current health percentage.

My question/idea is this: Both problems that make stat weights hard to calculate could, in principle, be addressed by retrospective combat log analysis. It should be possible, either with raw combat logs, or an addon designed for the purpose, to determine, for every heal cast during an actual fight, what the target's health % was. The combat log of a fight also records the specific collection of casts, including the buffs active at any time.

For a given cast, on a target with a specific health%, with a specific set of buffs, it should be possible to determine the effect of each relevant stat on the throughput. It should also be possible to track mana effects from Resurgence. If you can do it for a single cast, then you can do it for ALL the casts of a particular fight, using the actual selection of spells used and the health% of each target actually healed.

Like many DPS-related analyses, this approach would assume the effect of stat changes is incremental (in effect, everything changes linearly). Stat changes that are of order 1% or so can be assumed to have relatively minor effects (which we can try to model if we wish) on overall spell selection. Of course, this is not perfectly true, but it seems to me something approximate would still be much better than anything currently available. It would be less reliable, but maybe not completely impossible, to analyze the effect of larger changes like hitting a higher haste breakpoint.

Something similar has actually been done already to analyze the Glyph of Water Shield.

Before closing, maybe a simple example will clarify what I'm proposing.

Suppose in a particular (very short) fight, you cast Riptide once on Player A, and Healing Wave once on Player B, and then the fight ends. By post-combat log analysis, you extract the health percentages of Player A each time Riptide ticked and Player B at the moment when you cast Healing Wave.

Supposing that these data are representative of the next time you will do the same fight, it seems to me it would be possible to determine (pretty accurately) how much 1 point each stat would increase your healing throughput and/or mana regen for the fight.

The analysis for a real boss encounter will obviously involve a lot more data, but I don't see anything that would prevent doing it, in principle.

I'm not 100% certain, but a quick web search gave me the impression that the target's health cannot be directly determined from the combat log. But a custom-written, lightweight addon could certainly collect all relevant information for later analysis. It might also be possible (but probably more difficult) to determine the target's health percentage to within the necessary accuracy by the size of the actual heal in relation to the nominal value (i.e. using the Mastery bonus itself).