Rawr sucks for healing! (SPLIT POST)

I don't like the program Rawr, because it's been clueless about tree healing in the past (when I tried it) and has suggested entirely pointless crap to me. However, as I've been looking at potential moonkin upgrades, I fell into the temptation of running my healer kit through it. All I can say is... LOL! It consistently ranks mp5 as a higher stat for me than crit and haste (wtf?), even if I set it to full raidhealing and bursty, short fights. I mean, it insists I regem to Intellect! Mana is NOT an issue to me. At all. Except maybe on very short and very bursty fights like hardmode Thorim P3, but they're over before I need to pot (and that's with Innervate still cool). Seriously, wtf?

Now I'm not sure whether it's just trees it struggles with or all healers. From my "recommendations" one could assume it knows about Disc priests (questionable), but I will uninstall it and go back to gut instinct and RatingBuster. This little experiment showed me Rawr is still a big pile of stinking doo-doo for for healers. I'm not sure whether any of you use this tool to evaluate healing gear - if you do, apply extreme prejudice!

Rawr doesn't work for me either. I've found it doesn't take Expertise into account and even when it suggests DPS upgrades, they would be downgrades as 10% of my hits won't land.

Marveen wrote:I don't like the program Rawr, because it's been clueless about tree healing in the past (when I tried it) and has suggested entirely pointless crap to me. However, as I've been looking at potential moonkin upgrades, I fell into the temptation of running my healer kit through it. All I can say is... LOL! It consistently ranks mp5 as a higher stat for me than crit and haste (wtf?), even if I set it to full raidhealing and bursty, short fights. I mean, it insists I regem to Intellect! Mana is NOT an issue to me. At all. Except maybe on very short and very bursty fights like hardmode Thorim P3, but they're over before I need to pot (and that's with Innervate still cool). Seriously, wtf?

I'm slightly puzzled here.

I just downloaded the current version of Rawr, set it to Tree model, input your talents and gear, selected a ton of available enchants and gems (allowing it the jewelcrafting gems) and ran an optimise, and it comes back with almost exactly what you have; indeed the only difference I can obviously spot is it using a 19 spellpower gem where you have a 16 haste one in your belt. It certainly isn't advocating gemming intellect at all... I've uploaded the saved file to http://jink.org/Marveen.xml if you want to take a look.

From how I look at it - I assume you're looking at relative stat values to say it values MP5 more highly than crit or haste? Note that the relative values are split between "heal burst" and "heal sustained"; MP5 is (quite rightly) rated at zero for heal burst, but rated highly for "heal sustained". If you're only looking at burst healing, then crit and haste are both quite rightly valued more highly than MP5, but in a mana-limited sustained healing scenario then MP5 quite rightly comes out on top.

Rawr is a very useful tool, but can take a bit of time to make sure that you're asking it what you think you're asking it at times! The thread here at MMO-Champion for example suggests that there are trees out there finding it working for them.

Here's an example, looking at bracers. I have Bindings of the Depths (70 spellpower, 46 crit, 34 spirit, 43 intellect), which are pretty cool. But it says Cuffs of the Shadow Ascendant (66 spellpower, 38 haste, 15 mp5, 43 intellect) are superior, both for sustained and burst healing. No matter where I put the slider of burst vs sustained healing, or fight duration, it will never rate the superior Bindings as better, even though they have more spellpower (both direct and spirit, and the added crit). And this is with the settings for raidhealing (2 tanks with Rejuv plus rolling LB, and Rejuv on everyone). I simply don't get why it's rating the Cuffs higher, the added haste is not worth that much as I'm already at the softcap (1 second GCD)!

Even more puzzling, it rates my Rejuvenation idol (reduction of mana cost on Rejuv) as insanely superior to the Lifebloom one (more spellpower on the tick). It makes no sense, as I'm telling it mana regen is not an issue and it still insists I need more regen.

Marveen wrote:I simply don't get why it's rating the Cuffs higher, the added haste is not worth that much as I'm already at the softcap (1 second GCD)!

Are you sure about that 1 second GCD? Not fully positive on druid workings, but I'm seeing a suggestion that "32.79 haste rating increases your haste by 1%. If you have Gift of the Earthmother, you need 19.98% haste to get to a 1 second global cooldown, or 655 haste rating", and on your Armory page you're listed as having 502 rating?

Marveen wrote:Even more puzzling, it rates my Rejuvenation idol (reduction of mana cost on Rejuv) as insanely superior to the Lifebloom one (more spellpower on the tick). It makes no sense, as I'm telling it mana regen is not an issue and it still insists I need more regen.

Where are you telling it that mana's not important? Assuming that under Options/Fight you've picked "Raid heal with Rejuv (2 Tanks RJ/Roll LB)", and other settings at default (four minute duration fight, with three wild growths a minute and no swiftmends), then the conclusion from the simulation is that it's mana limited - so it's not surprising that it's rating an idol that would allow you to get more casts off before being OOM?

Telluria wrote:Are you sure about that 1 second GCD? Not fully positive on druid workings, but I'm seeing a suggestion that "32.79 haste rating increases your haste by 1%. If you have Gift of the Earthmother, you need 19.98% haste to get to a 1 second global cooldown, or 655 haste rating", and on your Armory page you're listed as having 502 rating?

Well, last I checked I was at 1.04 (1.3 for non-HoTs). Add to that 3% from Swift Ret aura and I'm practically at the softcap.

Telluria wrote:Where are you telling it that mana's not important? Assuming that under Options/Fight you've picked "Raid heal with Rejuv (2 Tanks RJ/Roll LB)", and other settings at default (four minute duration fight, with three wild growths a minute and no swiftmends), then the conclusion from the simulation is that it's mana limited - so it's not surprising that it's rating an idol that would allow you to get more casts off before being OOM?

Yes, that would work .. if I was going OOM. The thing is, I'm not! Even on fights like XT-002 hard (nearly 10 minute fight), mana is not an issue. The closest I get to stretching my mana is on hardmode Thorim P3, where I spamheal one tank. But that's over in less than 2 minutes..

I think the big trouble with Rawr is it thinks like a DPSer - you're ALWAYS going to be optimising your best "rotation", so it models that. If there's even a small break in your healing (change of phases, tank dodges), we will get mana back, and their stat prios fail to simulate that. Healing is too stop-and-go in 10-mans (maybe less so in 25-mans as there's a greater need for well-defined assignments) - I don't think Rawr can model it.

Marveen wrote:I think the big trouble with Rawr is it thinks like a DPSer - you're ALWAYS going to be optimising your best "rotation", so it models that.

Well, strictly speaking it of course doesn't "think" at all, it simply runs models written by folk who've volunteered their time and effort to write them.

Healing will be harder to model than DPS, yes, as it's not an always-on nukefest for the most part - but that's not to say Rawr is never useful. It can at least provide some sensible input, as opposed to simply being left with a 'gut feeling' that item X is better than item Y.

Marveen wrote:If there's even a small break in your healing (change of phases, tank dodges), we will get mana back, and their stat prios fail to simulate that.

Rawr can only be as good as the information it's given. Breaks in healing, for example, would be reflected by moving the "% of fight spent in FSR" available under Options/Mana. Similarly you can tweak around a slider for expected uptime of Replenishment, and whether or not you'll be using a mana potion (the default assumes that you will consume a Runic Mana Potion during each fight, I don't know if that accurately represents what you actually do?).

Telluria wrote:Rawr can only be as good as the information it's given. Breaks in healing, for example, would be reflected by moving the "% of fight spent in FSR" available under Options/Mana. Similarly you can tweak around a slider for expected uptime of Replenishment, and whether or not you'll be using a mana potion (the default assumes that you will consume a Runic Mana Potion during each fight, I don't know if that accurately represents what you actually do?).

I've tried moving the slider all the way to the left/right and I still think it's spouting out rubbish.. I just don't think it produces useful numbers for trees/healers. There's something fundamentally wrong with its model of (tree) healing.

As far as gearing as a resto, it should unequivocally rate spellpower > all, with spirit as a close second, and everything else behind. It's overrating haste and to some degree intellect and mp5, and it's underrating crit and spirit. I can live with the crit underrating - but I feel it's ignoring the fact spirit gives me spellpower and in-combat regen. If that's the case, it's useless for modelling trees!

Ultimately, I was hoping Rawr would give me a nice upgrade list I could use and save me a heavy WoWhead trawl. But when it suggests I go buy the Emblem of Valor bracers (less spellpower than my current ones, no haste, just a tonne of mp5), I can't trust its results. I'm sure I could tweak it and fiddle with it until it spat out the upgrades I want it to give me, but that kinda defeats the point of the whole exercise as I may as well go into WoWhead with a stat filter and get the same results..

Marveen wrote:As far as gearing as a resto, it should unequivocally rate spellpower > all, with spirit as a close second, and everything else behind. It's overrating haste and to some degree intellect and mp5, and it's underrating crit and spirit. I can live with the crit underrating - but I feel it's ignoring the fact spirit gives me spellpower and in-combat regen. If that's the case, it's useless for modelling trees!

Or you just have a very different take on what's best?

I mean, going to our friends at Elitist Jerks, there's this pre-Ulduar list, and this early best-in-slot Ulduar list, of optimal resto gear that are referenced from the first post in the main Resto Itemisation thread.

The pre-Ulduar list suggests (with an * to mark what's regarded by that list as best in slot):

* [Unsullied Cuffs] - cloth, 54 Sta, 46 Int, 42 Spirit, Red Socket with +4 Spi bonus, 34 haste, 64 spellpower
[Esteemed Bindings] - leather, 42 Sta, 38 Int, 35 Spi, Red Socket with +4 Spi bonus, 37 haste, 56 spellpower

The Ulduar-included list suggests:
* [Grasps of Reason] - cloth, 63 sta, 53 int, 48 spi, red socket, 75 spellpower, 39 haste
[Unsullied Cuffs] (same as they were before)
[Bracers of the Broodmother] - leather, 54 Sta, 46 Int, 46 Spi, 36 haste, 74 spellpower

All four of those bracers have haste and no crit. I'm therefore inclined to suspect that if Rawr is suggesting that you should favour bracers with haste over bracers with crit, it's because it's following the same sort of logic as the theorycrafters at EH. You might personally disagree with their choices, but (particularly with my limited knowledge of druid healing) I'm not going to post there and tell them their priorities are wrong

Marveen wrote:But when it suggests I go buy the Emblem of Valor bracers (less spellpower than my current ones, no haste, just a tonne of mp5), I can't trust its results.

Erm, your current bracers have zero haste, the Cuffs of the Shadow Ascendant have 38 haste (not "no haste"). Stat-wise, they resemble the bracers the EJ theorycrafters preferred more than your current ones. You might not like what it's suggesting but I don't think it's fair to say it can't be trusted.

Telluria wrote:Or you just have a very different take on what's best?

Yes I do! At least as it applies to 10-mans with the murlocs.

Telluria wrote:I mean, going to our friends at Elitist Jerks...

I know this may amount to heresy, but I'm going to state it here too: EJ's theorycrafting does not fully apply to 10-mans non-hardcore raidguilds. Especially not healing!

Now, take a deep breath and hear my arguments..

1. When they base their numbers on their experience they are looking at cutting-edge, hardcore raids. As they face hardmodes a lot sooner than us mere mortals, they tend to be extensively undergeared when they succeed (I doubt anyone who was part of world kills in the first week had the epic Sons of Hodir shoulders, for example!). Naturally, they must min/max class abilities and consumables to a much greater extent as they don't have access to the gear we do! This is when they come up with things such as "stacking Holy priests" or "using this or that kind of tank for this or that encounter". More casual groups (like us) tend to make greater use of hybrids and are far better geared when they first start trying face hardmodes.

2. When they base their numbers on theory/numbers alone, they assume perfect class stacking and people not fucking up (as much as we do anyway! ). In the case of healing, this means they tend to stack one kind of healer and one kind of tank to make the encounter as forgiving as possible. They don't use whatever healer they can get, they use the class(es) that give them the most bang for their buck for that specific fight. So they will only use a certain class of healer when that specific healing class is THE BEST in that role. Their theorycrafting tends to apply to that class' healing niche only.

We will fuck up a lot more, and we don't risk losing people because we (healers) are undergeared - we lose them because we fuck up or we use suboptimal classes for that fight. Our raids are more random, damage is more chaotic, we have to be a lot more reactive. So theorycrafting that's meant to optimise our niche's healing leaves us in the lurch when we can't dig up 2 Holy Priests for Hodir, or a DK tank for Vezax. So, on fights where we're doing what our niche is meant to be doing, even if our gear is not perfectly itemised it's still much better than what hardcore raids had access to, and any minor gains/losses are not as crucial to us as they were to the cutting edge raiders (the marginal value of an upgrade is lesser to us). Alternatively, on fights where we're having to do something atypical for our class, standard theorycrafting will tend to say "don't be losers, stack holy priests or mages and warlocks or use a dk tank".

For example: exactly 1.00 second GCD is a bonus for a tree on fights where we're used for what we excel at - buffering raidwide damage (Iron Council). But when the murlocs do the fight in normal mode I am not on raid healing - I swap with D (OT healing), D swaps with Zuzu (MT healing) and Zuzu swaps with me (raid healing), because we need all cleansers on the Steelbreaker tank when he's not being OTed. And their theorycrafting doesn't take that into account! :O

Telluria wrote:All four of those bracers have haste and no crit. I'm therefore inclined to suspect that if Rawr is suggesting that you should favour bracers with haste over bracers with crit, it's because it's following the same sort of logic as the theorycrafters at EH. You might personally disagree with their choices, but (particularly with my limited knowledge of druid healing) I'm not going to post there and tell them their priorities are wrong

Their priorities aren't wrong for the encounters they face in the raids they are in. I don't say haste is bad, I love the stat, but after the softcap is reached crit is much better. If I'm looking at a 15+ spellpower upgrade and a minor haste upgrade, it's a nobrainer for me - and I don't understand why Rawr rates haste higher than spellpower!

Telluria wrote:Erm, your current bracers have zero haste, the Cuffs of the Shadow Ascendant have 38 haste (not "no haste"). Stat-wise, they resemble the bracers the EJ theorycrafters preferred more than your current ones.

>.< My bad, I was referring to the Bands of the Great Tree (which topped my upgrade list although I didn't mention them in the first comparison - and they have no haste or crit, they're pure regen bracers).

PS: I'm enjoying the discussion!

Marveen wrote:

Telluria wrote:Or you just have a very different take on what's best?

Yes I do! At least as it applies to 10-mans with the murlocs.

Telluria wrote:I mean, going to our friends at Elitist Jerks...

I know this may amount to heresy, but I'm going to state it here too: EJ's theorycrafting does not fully apply to 10-mans non-hardcore raidguilds. Especially not healing! [etc]

I'm going to go ahead and say that this is the root of the some of the problems you are having with Rawr... seeing as it's the EJ guys who are most likely doing the majority of Rawr changes!
(And it is entirely possible that you are right - most EJ people don't care much for 10-man raiding)

And yes, Rawr is occasionally out-right wrong - see Dark Moon Card: Greatness - which is incorrectly implemented in Rawr (It's 300 "highest stat" (though including sta - which the trinket doesn't buff) - triggered by healing...)

That, and in my experience, it's very easy to miss the "options" tab - which makes a huge difference at times.

Ultimately, I think that for the ultimate usage of Rawr, you need to be able to motivate every upgrade and if something goes against your beliefs, you are either right - and you've just proven your superiority of your class - or you are wrong, and you've got a clear way of improving how to work your class.

Anyways, that was off topic.

Mnemo wrote:I'm going to go ahead and say that this is the root of the some of the problems you are having with Rawr...

Aye, if you've decided that you don't like the 'accepted wisdom' of what's best to get, it's hardly likely that a program designed to optimise characters (which will follow the 'accepted wisdom') then tells you things you don't like!

Ultimately it's a useful tool for those of us who are more or less happy with 'accepted wisdom', and it's very handy for juggling around multiple pieces of gear to tidily reach hit caps etc, it's never going to be a total replacement for brain.exe though

Mnemo wrote:And yes, Rawr is occasionally out-right wrong - see Dark Moon Card: Greatness - which is incorrectly implemented in Rawr (It's 300 "highest stat" (though including sta - which the trinket doesn't buff) - triggered by healing...)

While it doesn't buff sta, it sure as hell IS triggered by healing! On the rare, nay freak occasions when we wipe, if I res Timber and pop a Mend Pet on to cheer him up, I'll almost always proc Greatness from it...

Telluria wrote:

Mnemo wrote:And yes, Rawr is occasionally out-right wrong - see Dark Moon Card: Greatness - which is incorrectly implemented in Rawr (It's 300 "highest stat" (though including sta - which the trinket doesn't buff) - triggered by healing...)

While it doesn't buff sta, it sure as hell IS triggered by healing! On the rare, nay freak occasions when we wipe, if I res Timber and pop a Mend Pet on to cheer him up, I'll almost always proc Greatness from it...

What I meant is that the way they have modeled Dark Moon Card: Greatness in Rawr is the following settings:

300 Highest stat (in Rawr, this does not exclude Stamina)
Trigger: Healing Spell hit.

This is quite incorrect.

Dahlaine, for instance, has higher stamina than agility. So DMC:Greatness will (according to Rawr) increase Dahl's stamina by 300, not the agi (Which DMC:Greatness will actually increase - as it does not increase stamina!).
(Note: I checked Taqwa 1496str/1439sta; Akari 1295 agi/1028 sta; Telluria 1653agi/1119sta - so you those won't notice this first problem... Bal was in prot gear, so was not checkable. (That's everyone I know who has the card))


Secondly, the Trigger does indeed trigger from practically anything - including healing, attacks and procs.
However, for a correct modeling, it is the incorrect trigger.
For a melee class, it would be a much better idea to change it to "PhysicalHit".

So basically, Rawr is wrong for this trinket

(Note: this problem exists in Rawr 2.2.7 (latest version I'm currently running) - and can be fixed by editing the trinket manually)

edit: and this is very offtopic - we should make a seperate thread if we continue this discussion! (a mod could split the topic even?

Mnemo wrote:300 Highest stat (in Rawr, this does not exclude Stamina)

You're quite right in saying it won't buff stamina (which is what I meant with "While it doesn't buff sta"), I actually can't quickly tell myself whether or not Rawr includes stamina as a possibility since the detail just says "Highest Stat" in a rather short field.

Mnemo wrote:Trigger: Healing Spell hit.

This is quite incorrect.

Secondly, the Trigger does indeed trigger from practically anything - including healing, attacks and procs.
However, for a correct modeling, it is the incorrect trigger.
For a melee class, it would be a much better idea to change it to "PhysicalHit".

So basically, Rawr is wrong for this trinket

Hm, I've actually just downloaded 2.2.7 to check this ... and yes, it's just a simple error for Enhancement Shamans! If you load up Dahlaine, and then change the class via the drop-down to Druid, you'll then find if you choose the Bear, Cat, or Moonkin models the DMC:G shows with a "15 sec 33% on damage dealt / 45 sec" trigger, while if you choose model "tree", then the trigger is "15 sec 33% on Healing hit / 45 sec". Basically whoever was putting together the enhance shaman model picked the wrong option for the card, is all

(Same error for enhance shamans in 2.2.8 (which works with the EU Armory), and 2.2.9 (which doesn't - since the EU Armory hasn't yet been upgraded to the newest version, 2.2.9 works with the US Armory), incidentally).

And yes, we've gone very off course and splitting this into a Rawr topic might well make sense

Mnemo wrote:edit: and this is very offtopic - we should make a seperate thread if we continue this discussion! (a mod could split the topic even?

Done! *waddles back to the Oracle hut and lurks from the shady interior*

*returns from the hut*

So, if Rawr just did as it was told it'd be alright. I just want to tell it "these are the stats I value - gimme ultimate power and ignore regen, I'll manage that myself tyvm", and for it to spit out a list for me. At the moment that's not doable unless I go in and change all the default settings (and even then some things may be wrongly coded, like the DMC thing - we know about that, what about other things?). It fails as a filter, in that it's making a judgement (and imo a wrong one - but we've had that discussion ) about what's best for my class!

If you want to completely ignore regen, just tell Rawr that it's going to be a very short fight; it'll then show you a "Cast Time Limited" result.

Thing is, it's still going to show the Cuffs of the Shadow Ascendant over your Bindings of the Depths, evidently because haste will allow you to pump out more HPS than crit will.

Interestingly it does clearly value spellpower over haste, since it recommends a pure spellpower gem where you have a (single) pure haste gem matching a haste socket bonus!

But the problem basically is - Rawr is doing as it's told; it's calculating optimal gear setups for a given task (which is 10-man as well as 25-man relevant, since when you go through Buffs you'll only have a slack handful selected as opposed to most/all which you'd have in a 25-man). This is why it's an external application which can consume significant CPU time, as opposed to a simple in-game addon like RatingBuster or Pawn.

If you just want to use a 'gut feeling' of what 'feels right' to you, then fair enough - but it's not reasonable to say "Rawr is wrong" - it's "Rawr is giving me answers that I don't like". It's not failing at all, nor is it making a judgement - it is giving calculated results; you're the one making a judgement by deciding that everyone else's theory and calculations are wrong for you. But then the onus is on you to show some calculations or proof for why they're wrong - or to simply say that you want to do things your own way, whether or not it's strictly optimal!

Which is cool, you're totally entitled to do that, and I've used somewhat suboptimal setups in the past myself because the optimal one was too damn ugly (see for example Cryptfiend's Bite versus the Wraith Spear, and thank goodness for helm/cloak graphics off at times)