Continuing my series of posts in which I lay out basic theorycraft topics (see more here: http://elitistjerks.com/blogs/1152-hamlet/tc101/ ), here's one on mana regen and throughput.

This is a bit different from the previous entries; I'm not laying out much in the way of game mechanics, just describing a technique I often see mishandled in healer discussions. There's no general formula here since applications will be class-specific, but the principle applies generally. As before, I'll work a simple example with Druids.

 

Contents

1Introduction 2The Method 3Example: Druids 4Analysis 5Conclusion

[top]Introduction

The discussion as it often goes, in forum/blog post:

1) Introduction: this is a debate between Talent A (regen) and Talent B (throughput).
2) The bulk of the post is mathematical analysis of varying amount of detail in order to attach regen/throughput values to the talents.
3) Conclusion in which the poster makes some judgment about how much regen they feel like they need.

This is sort of a workable start, but it tends to ignore some very important factors and spend undue time on details that are often unnecessary to reach a useful result. In particular, the link between 2 and 3 is usually tenuous at best.

The piece that is usually missing is context. Specifically, the context of how the regen/throughput tradeoff in question compares to other regen/throughput tradeoffs that are being made or even that could be made. Without any information on what regen/throughput exchanges you currently make and which ones you refuse, you have no basis with which to process any information you get on the talents/stats you're looking at. Also, once you have a feel for what's a reasonable exchange, you can often make good decisions with only a quick estimate of the value of a particular talent.

[top]The Method

The basic algorithm for setting up the comparison is as follows:
0) Start with a hypothetical character setup that favors pure throughput.
1) For each regen/throughput exchange you want to consider, estimate the regen you can gain and the throughput you would lose.
2) In each case, divide the regen gained by the throughput lost.
3) Rank your possible regen sources in decreasing order using the ratio from 2.
4) Use the regen options from the top of the list, working downwards, until you have sufficient regen for the purpose of whatever encounters you're doing. Don't use a lower-ranked choice if a higher-ranked one is still available.

The point here is that is it very silly (with rare exception) to take one regen-for-throughput tradeoff while passing up a more efficient one. If you do that you're simply leaving healing potential on the table. You want to find your necessary regen by taking only the least throughput-costing options available, and no others.

As I said, this is all very general--any particular application is class-specific and depends on how you detailed you want to be. But below will be some examples showing I can use this to make some decisions about setting up a Resto Druid.

[top]Example: Druids

Starting from a hypothetical max-throughput Druid, let's consider some ways I can convert throughput to regen:
1) Regem Int to Spirit.
2) Reforge mastery to Spirit.
3) Take Furor over Genesis.
4) Use Ember meta instead of Revitalizing.
5) Use a mana trinket (Jar of Ancient Remedies) over a mana/throughput trinket (Fall of Mortality). I chose options of equal ilvl here.
6) Enchant Darkglow over Lightweave.
7) Enchant Heartsong over Power Torrent.
8) Cast less Rejuvenation and more Nourish.

Now I have to estimate the regen/throughput effects of each one. For this you can use whatever theorycraft tool exists for your class. Here I'm simply going to pull values from TreeCalcs ( http://elitistjerks.com/f73/t110354-...2_a/#TreeCalcs ). This post isn't so much about how to get the numbers but about what to do with them once you find them (I might do a later post on how to estimate things when you don't have a full calculation handy). You may also want to see my earlier post about how to compute your mana regen.

1) Regem Int to Spirit. A point of Int gives me 3.2 HPS (in the default TreeCalcs spell usage) and 0.92 MP5, including the mana pool increase averaged over a 5 minute encounter. A point of Spirit gives 0.7 MP5. Here, the Spirit gives me less mana than the Int. So my regen-for-throughput "benefit" is actually negative--I'm never going to choose Spirit over an equal amount of Int.
2) Reforge mastery to Spirit. A point of mastery gives me 2.2 HPS overall. A point of Spirit gives 0.7 MP5. Trading mastery for Spirit would give me about 0.7/2.2 = 0.32 MP5 per HPS lost.
3) Take Furor over Genesis. A point of Furor gives me about 220 MP5 (again, including the mana pool increase averaged over 5 minutes). A point of Genesis gives me about 470 HPS (even without something like TreeCalcs, it would be easy to get a good estimate just by checking how much of my healing is from HoT's). Moving points of Genesis to Furor gives me about 0.47 MP5/HPS (quite a lot better than the prior option).
4) Ember meta over Revitalizing would gain 44 MP5 for 324 HPS, a ratio of 0.14. Once again, a straightforward estimate of Revitalizing could be made just by looking at your overall crit rate.
5) Heroic Jar over Heroic Fall would pull in 88 more MP5 for 226 HPS lost, a ratio of 0.4.
6) Darkglow vs. Lightweave at first looks to be the same as Int vs. Spirit. But, temporary Int effects like Lightweave aren't quite as good as normal Int since they don't improve your starting mana. Without that, the MP5 value of Int goes down to 0.58. Also, after a recent stealth nerf, Lightweave as a longer ICD than Darkglow (60s vs. 45s), so I'll bump up the value of Spirit by 1/3 to compensate, making it 0.93. So swapping Lightweave to Darkglow gives 0.35 MP5 for each 3.2 HPS lost, a ratio of 0.11.
7) Heartsong vs. Power Torrent. Similar to the prior Int. vs. Spirit. But PT is up for 12s out of every ~50, and Heartsong for 15s out of every ~28. Since Heartsong has about twice the uptime of PT, I'll just give the Spirit a x2 boost, for 1.4 MP5. Heartsong would give 0.48 MP5 for each 3.2 HPS lost, a ratio of 0.15.
8) Spend more time casting Nourish. This isn't a really a setup choice and it's not exact by any means, but when doing all this it's helpful to have a ballpark estimate of what extra mana means for us in practice. Looking at spells, I see that Rejuv spam burns around 17000 MP5 to Nourish's 4600. But it does 34300 HPS to Nourish's 7000. So any time shifted from Rejuv to Nourish gains 12400 MP5 for 27300 HPS lost, a ratio of 0.45. This is a useful figure to have--if we imagine that our WG, Swiftmend, and Lifebloom casts are largely fixed and independent of mana concerns (we tend to use them on cooldown), what's affected by mana tends to by our choice of "filler" casts.

Resorting our options in order of MP5-given to HPS-lost ratio:
--Furor (0.47)
--Less Rejuv (0.45)
--Jar (0.4)
--Reforge Spirit (0.32)
--Heartsong (0.15)
--Ember meta (0.14)
--Darkglow (0.11)
--Gem Spirit (negative)

[top]Analysis

What do I learn from this? Firstly, I really should have taken another look the Revitalizing meta after the healing crit change in 4.2. I'll probably be changing that in my guide ( http://elitistjerks.com/f73/t110354-...taclysm_4_2_a/ ), given the apparent weakness of the Ember meta. But more generally:
We can see that if anyone is feeling limited by mana, the first thing you want to do shore up regen is to have Furor in your spec instead of Genesis. Beyond that, reforging Spirit or equipping a good mana trinket (ignoring for now Shard of Woe, which well-known to be overpowered on the regen front) are also reasonably solid if you need more mana. Finally we have a bunch of much less favorable options (Heartsong, Darkglow, Ember). You can theoretically go all the way down to these if you're really desperate for mana, but the "Less Rejuv" estimate might indicate that this is a poor idea, as I'll explain in the next paragraph.

The "Less Rejuv" is, once again, a rough estimate of how much mana I can save by trying to be more economical about Rejuv. This is important because, in the end, the whole point of getting extra mana as a Druid is more or less to be able to use Rejuv more often. So if we make regen tradeoffs in other areas that are significantly inferior to what we can obtain by adjusting our casting a little, we can expect healing to suffer as a result. Now this isn't hard and fast--sometimes you might simply need to be able to use Rejuv a lot even at the cost of overall strength on all your spells. But attaching some numbers to these things helps us see that we're very unlikely to ever use e.g. Darkglow, because for each added Rejuv it gets us, we're giving up around 4 times that amount of healing on other spells.

I'm going to leave more detailed discussion to Druid-specific forums, but hopefully this is a good demonstration of how I approached the whole problem.

[top]Conclusion

This may look like a lot of work--I'm basically saying you can't look at a regen option for a healing character without looking at all of them at once. But that's only half true. Once you have some context built up of what's a good tradeoff and what isn't, then each time you're facing a new choice you already have a proper background against which to evaluate it. And even that first time, you don't have to do it all yourself; this analysis should really be a subject for each healing class forum to address. Even though the numbers will slightly change on a per-character basis, the basic scale of what's a good tradeoff and what isn't can be easily be set out on a general basis.

Which brings me to another major point here: for most evaluations of possible regen options, you need loose estimates at best. If I can value a regen/throughput tradeoff to even one significant digit, I have a good enough number for everyday use. A napkin calculation can easily tell me if a regen tradeoff is giving me MP5/HPS to the tune of 0.5 or 0.1. This is why say that many analyses I see of (for example) Furor vs. Genesis tend to miss the point. Instead of pages of math to try to evaluate the precise benefit of each talent, spending half the effort simply building the context of how much mana regen is worth to your character will give a much more meaningful result.

As I said at the beginning, this post is about a general principle that anyone analyzing a healer regen question needs to find the best way to apply to their own situation. So it doesn't come down to some final equation like my prior posts, but hopefully it provides some good direction on what to do with the numbers that you do obtain when trying to analyze your class.