On Restoration Shaman’s Relationship With Critical Strike

Introduction

Today’s post is all about stats for Resto Shaman in Mists of Pandaria. I have some very important notes and caveats to state before I get into the graphs, but first a short introduction.

It’s been a long time since Restoration Shaman have had meaningful stat weights. In Cataclysm, the general consensus was that it was meaningless to state that any of our secondary stats was simply better than any other. The reason for this was that different stats are good under different conditions. Vixsin did a good analysis of the idea by comparing Mastery and Crit in early Cataclysm (linked: updated version for Mists of Pandaria).

We have had a good idea for a long time now of how to incorporate Haste, Mastery, Crit and Intellect together to find out what circumstances each is best at. However, we have mostly focused on throughput from these stats. It has been tacitly assumed that it is impossible to look at how mana regen and throughput interplay – the analyses are always separate. Indeed, the interplay is known by many theorycrafters as “pseudo-throughput” and left at that – an unmeasurable and ethereal quantity that nobody wants to touch.

The effect of this has been to make decisions difficult when considering which stat one should be trying to maximise for any given situation. The net result of this is has been the following;

Deciding on stat distributions has for a long time been relegated to two methods: intuition and trial and improvement.

Neither intuition nor trial and improvement are ideal methods for assigning stat distributions. Firstly because intuition is likely to fail, and secondly because trial and improvement is laborious and also liable to failure.

I hope that we can remove a lot of this uncertainty through improved methods for analysis of such things. Some weeks ago, I published a post on a potential baby step forward in this direction; Healing Effective Power (HEP). Today I shall be utilising that method to look at the effects of stat scaling of Crit, Spirit and Intellect on “pseudo-throughput” in Mists of Pandaria.

Ass-Covering

Some important caveats must be made. Firstly – HEP is not a direct measure of healing. The interpretation of HEP is ambiguous, so one cannot directly say that stacking X Crit will get you Y healing directly and Z from “pseudo-throughput”. Instead, HEP gives an estimation on the effects of regen on your total throughput. There are valid reasons not to take HEP at face value.

Secondly – In this post you will see graphs which show a normalised scaling coefficient between 0 and 1. These are not stat weights. Do not use them as such.

Thirdly – I expect that there are better ways of looking at “pseudo-throughput”, and I actively encourage you to seek them out (mathematically or on the interwebs) and disseminate the results. I want to see better ways of quantifying “pseudo-throughput”, for the benefit of everyone.

Finally – This post will not tell you exactly what stats to stack. You must still decide which stats you want. In that sense, I have already failed in my aim. However, this post can help you decide. Hopefully that is enough to warrant your attention and feedback.

Starting Assumptions

I decided to look at scaling at the level of T14 Normal raids in Mists of Pandaria. With that in mind, I used Zagam’s MoP Dungeon and Loot Guide to assess what gear I would be starting MoP raids with. The following table summarises the gear I found:

Piece Int SP Crit Spirit
Head 899 626 555
Neck 501 285 363
Shoulder 668 465
Cloak 501 254
Chest 819 486 603
Waist 668 465
Bracer 501 344
Boots 668 412 0
Gloves 668 471
Legs 899 609
Ring1 566 368
Ring2 501 285 363
Trinket1 847
Trinket2 847
Weapon 435 5812 261
Shield 501 334
SUM 10489 5812 3071 4478

Which by my calculations totals up to somewhere around;

Intellect 10489
Crit 16.7%
Spell Power 18240
Spirit 4478

I based the rest of my sample HEP calculations on these base values.

First Look

First I calculated the HPS, HPM, and HEP for various small sets of rotations called phrases – so called because, like phrases in English, they can be put together to form larger rotations or responses to the situation (see my original post, linked earlier, for a full explanation). The phrases were;

Phrase 1 – Riptide, Healing Wave, Healing Wave, Healing Wave

Phrase 2 – Riptide, GHW, GHW, GHW

Phrase 3 – Riptide, HS, HS, HS

HEP and HPS for three different “mini-rotations” or phrases.

By playing with the stats, it was easy to see that increasing any of the stats in this study also increased HEP and HPS. The important question is not how much the HEP is though, but how fast it increases with each stat. With this in mind, I went through the process of finding out.

Closer Analysis

The next stage is to look at how HEP increases with various stats. With that in mind, I recorded the HEP values at intervals of each stat (varying them independently). The following graphs show the results:

The amount of HEP as Crit increases, assuming the other stats stay stationary at the values shown earlier.

Here we can see that Crit increases HEP quite significantly as the % chance increases. Using a linear fit, you can see that the gradient varies for different phrases. Phrase 1 is the most efficient phrase to begin with, and seems to gain the most from Crit here.

The amount of HEP as Spellpower increases. Other stats are stationary.

With Spellpower, the difference seems less pronounced! That’s interesting because we know that Spellpower increases HPS by rather a lot!

The amount of HEP as Spirit increases. Other stats are stationary.

In this form, the stats are rather hard to compare. We can see that blue (Phrase 1, Healing Wave dominated rotations)  seem to gain most from stat increases. This seems to make sense, as low healing, mana efficient spells stand to gain more from stat increases. However, we want to know which stat is best, not which phrase! With that in mind, time to look at how each stat compares. This happens in two stages; first, extract the gradient from each line of each graph shown. Next, make a conversion from the graph’s gradient to a quantity of HEP per stat rating (e.g. HEP per Crit Rating Point). Here’s the results:

How different stats compare, by their effect on HEP. The values have been normalised to the value of the highest for each Phrase, so Spirit has been forced to 1 in each Phrase.

Some interesting deductions:

  1. Spirit is by far and away the best stat for HEP, which means that you should be looking to have as much of it as you can [which reinforces what I stated here on the subject].
  2. Critical Strike Rating is powerful. Indeed, Crit is on a par with Intellect as a rotational stat. Damn. I didn’t see that one coming! I’m pleased to see that, though. It means more interesting gearing.
  3. The higher HPS rotations gain less in terms of “pseudo-throughput” than more sustainable rotations.
  4. Which stat is “best” depends on your spell choices. It always will. You have to take account of that when you decide which stats you want.

TL;DNR

There are a couple of things you should take away from this:

  • Crit cannot replace Spirit, even though it provides regen. Spirit is by far and away the best of the two for rotational regen.
  • Crit is a much better stat than we give it credit for. Don’t neglect it. It really, really makes up for its RNG. Consider reforging for it, based on this and Vixsin’s assessment of Crit vs. Mastery.
  • Intellect is still a strong stat: You still want to gem for it with red/purple gems.
  • Still no stat-weights to make things brainless. If anything, it’s now more of a choice than ever. Hopefully I’ve made you a bit more informed.
About these ads

About stoove

Steve lives a double life: opinionated Physics student at Surrey University by day ("Steve"), computer games and World of Warcraft nerd by night ("Stoove"). Possibly lives in Guildford for most of the year. It's rumoured that he's a gigantic geek who reads New Scientist for fun. Some people think he's sane but they've been repeatedly proved wrong. His influences probably include such mythical beings as Ben Goldacre, PZ Meyers, Richard Feynman, Jim Al-Khalili, Animal from the Muppets, Ian M Banks, Aron Diaz (AKA Dresden Codak), and the leprechaun he knows only as "Petey". You can follow Steve on twitter but it probably won't help you when he conquers the world.
This entry was posted in Maths, World of Warcraft and tagged , , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s