gcbirzan

Crit doesn't give you haste, but a fight isn't simply "the longest time you can spend spamming Holy Light". A fight is a period of time over which you have to maximise your HPS. And for that, you use a ratio of Holy Light/Flash of Light, to the extent that your mana allows for it. If you have more mana, you can cast more Holy Lights. Crit is different from mp5 in that, as you said, the mp5 it gives back is dependant on how often you cast Holy Light and the more crit you have, the more often you can afford to cast Holy Light.

 

Abeness

I want to point out again that using Flash of Light lowers the effectiveness of crit in terms of mana regen (so if we modeled a player mixing Holy Light and Flash of Light casts, crit would seem worse). The rest of what you said is in agreement what I have said, so I think you've got a point . Although remember that when you say, "[...] and the more crit you have, the more often you can afford to cast Holy Light," that this is also true for Mp5.

 

gcbirzan

We are, indeed, arguing your interpretation of the mathematics you presented. You are using it to say that crit doesn't scale with itself in terms of mana regen, and that is only true if you ignore the fact that the mana regen you gained from crit allows you more mana regen. I have no idea how I can explain this better than I just did, I don't seem to understand why you keep on insisting that the frequency of Holy Light casts is independent to the mana regen you have.

Indeed, Flash of Light lowers the effectiveness of crit versus mp5. However, while more mp5 does increase the number of Holy Lights you can cast, the mana return from mp5 does not change depending on how many spells you cast, whereas that of crit does.

 

Abeness

So far, the only point I want to get across is that crit can be represented as a Mp5 value, and that the scaling properties of crit are so minimal that there is no way it breaks my point. And that tiny amount of mana from scaling is easily made up by the fact that crit does not give mana back for Beacon, Sacred Shield, Cleanse, etc. Also, that tiny amount of regen that scaling crit gives can easily be factored in (with very minimal reprocussions on my conclusions) if fight length is considered. Again, my first goal was just to prove my "crit as Mp5" math.

Agreed. The slower you cast, the less Mp5 you get from crit. The more you cast, the more Mp5 you get. And when mana is spent on Flash of Light rather than Holy Light, crit loses Mp5 value.

 

gcbirzan

Going from 30% to 65% doubles your mana return per point of crit, I don't really see how you can call that minimal. Your crit as mp5 math is, probably, correct, your interpretation of it is wrong.
Ironically, your last statement proves our point. Yes, mana spent on Flash of Light makes crit lose value, but you should only do that if you don't have enough mana to cast Holy Light. So the more mana regen you have, the more Holy Light you can cast and the better crit becomes.

 

burghy

Each spell gives you mp5/5*cast time mana back. The more spells you cast, the higher the return from mp5 is.

 

gcbirzan

No... It doesn't matter how many spells you cast, it doesn't matter how much time you spend casting, the mana return from mp5 is only dependant on the fight's length.

 

Endoscient

You really really don't get how this theorycrafting stuff works, and no matter how much or simple we explain it, it isn't helping anything. I am going to post one last time, then stop bothering. Almost all of these statements I am about to make you have refuted in one of your posts, I can dig up to quotes if don't believe me.

The amount of effective mana you get from mp5 does not change as you gain more mana. The amount of effective mana you gain from crit increases as you gain more effective mana. Those are flat out true statements that you do not seem to understand. You seem to think as you gain more mana from mp5 it takes longer to cast all of your spells, so mp5 gives you more in return. In the real world though fight durations are fixed, and having more mana won't change how long it takes to kill the boss. Crit restores a percentage amount of mana on each spell cast, so obviously the more spells you cast (so more mana spent) in a given time period the more mana it restore.

Fight duration is fixed no matter how much mana you have. Fight duration changes only based on the fight itself, and how fast your dps is able to kill the boss. Just because you can last with your current spell "rotation" for another 20 seconds, doesn't mean you will get the chance to. Instead you should have cast more Holy Lights during the fight so you wouldn't have that extra mana at the end.

Gains from Illumination can be accurately represented by reducing average spell cost by base_cost * 0.6 * crit_chance. You can also represent it as increasing your effective mana pool by 1 / (1 - 0.6 * crit_chance) if there are no cost reduction effects, you can modify that formula for cost reduction effects but it isn't as simple then. These are two 100% true and accurate statements, say if you don't believe them to be true or accurate and why, and I'll prove it wrong.

You also don't seem to understand it is possible to gain more mana and use it for additional healing, while keeping the fight duration that same. You can easily do this by replacing FoL casts with HL casts. This is only not true if you are spamming HL at its cast time for the fight duration. But this is incredibly hard to achieve in any fight that is remotely difficult, and if you somehow achieve it then extra mana regen stats aren't worth much so the point is moot.

Crit scales with itself and mp5 scales with crit, but crit scales with itself at a much higher rate then mp5 scales with crit, but starts out lower. These can easily be seen using the equations I posted two paragraphs back. Just because they both scale with crit doesn't mean you can disregard the importance of initial crit rate, which is exactly what you are doing.

Your initial formulas were incorrect as well (on top of fault premises) because they only considered the initial gains of crit. They didn't calculate the advantage that spells casted with the gained mana from crit will have a higher chance to crit then the spells casted with the gained mana from mp5.

Also use Rawr.Healadin if you want to explore the value of stats. It is the most accurate tool at the moment for modeling a Holy Paladin healing a fight.

 

Ranjurm

Abe this is what's happening:

You're essentially assuming that you cast HL at a static rate until you run dry and thus have variable fight lengths. What others are saying is that, that information is not useful (generally) since fight lengths are more or less fixed while HL/other ratios are variable when it comes to lasting the whole fight. If you exceed the fight length then you should either use HL more or need less regen gear. If you fall short of the fight length you should use HL less or get more regen.

While what you are testing is not generally useful it can be used to judge if mp5 or crit could allow you to use DP less or to just enough that you only have to use it under low danger phases.


Assuming a static fight length (you of course could test different values for that number) you could have a maximum of 37 HL casts per minute assuming a 1 second GCD for spells, 1.5 second HL cast, 1 BoL/minute(m), 2 SS/m, 1 Judgement/m (1.5s GCD). Now you would find the highest value of HLs/m are allowed (up to 37) given your starting conditions that leaves you at the end of the fight with as close to 0 mana or a slightly higher value if you prefer to account for an actual healer's dislike of being without even a thousand or so mana. Those are the conditions that MP5 and crit must be pitted against. There are still a lot of variables to take into account such starting mana and crit levels but those are variables not the basic modeling method.

 

Endoscient

If for some reason you wanted to use a healing until oom model for fights, which isn't really a good idea for previously mentioned reasons, then the benefit of stats should not be given as equivalent mp5 values. Since that value doesn't represent at all that it made you heal for longer then another mp5 value of the same amount. The value of mana stats in that model is only to increase fight duration. You should give values that are over the total fight duration, like total mana, total healing, or just the duration itself. It will very seriously skew and misrepresent the results otherwise.

Changing the times you DP is very difficult in these equations, since it is very dependent on the fight itself. For example on Algalon no matter how much mana I have, I am going to gain no benefit by having to use DP less. Since you always use it while going down during Big Bang, and there is no healing to be done anyway. On a fight like Freya+3 when there isn't really a defined low damage phase, you can get a lot more mileage out of using DP less.

 

DiamondTear

He's not ignoring the fact that mana regen from crit allows for more regen. It is represented in his first post which shows that the mp5 value of crit increases as amount of casts goes up. Assuming regen is what is limiting the amount of casts, having enough regen makes crit eventually better. This includes the regen granted by crit.

You seem stuck on the idea that the cast interval is set before the fight, when it's actually calculated from the average cast interval you had during the fight. The interval doesn't have to be the same throughout the fight. You cast like you need to and after the fight you can see if you cast enough to make crit better than mp5.

I found his way of modeling crit as mp5 useful, but I can't say if it's better than the established way.

 

• Mex

This was something that wasn't stressed, and compounded by an insistance on using a wildly impractical interval. That said, it's still largely irrelevant because the assumption was that it was a static value, rather than the core value that holy paladins should be striving to improve (HLs per minute or HL:FoL ratio, however you want to represent it).

Modelling crit as mp5 is not helpful, because the answer, invariably, without fail, will be "it depends." Holy paladins have been trying to quantify the relationship between the two stats since TBC and the end result is always those same two words. This is why pallies started looking at it as a reduction in spell cost or increase in effective mana, because it gave them the ability to take crit rating as a means to improve HLs per minute, instead of viewing HLs/minute as a variable in calculating crit rating's value.

At the end of the day, crit and mp5 are different stats and function in different ways. This is obvious at even a cursory glance. Trying to bend and twist and distort one of them to fit into the model of the other is never going to be a useful exercise. It is far more helpful to ensure that we understand how these stats work and in which situations they're useful (for instance, you wouldn't stack crit in a fight where you used 70% of your GCDs on cleanse, nor would you favour MP5 for a slowish start fight leading into a feverish DPS race burn, a la Steelbreaker / Thorim), rather than trying to force them into general rules of thumb, or worse, overly specific and complex relationships better handled by spreadsheets / RAWR.

 

Palados

@ Abeness

While your math seems to be OK per se, as other people mentioned it is not relevant for real fights. Let's look at the toy model with 'easy' numbers and if you think about each step (I will number them) you will see why 'conventional' math is better.

Model: Mana pool is 70000, there is only 1 healing spell (imagine FoL having no cost and being just a 'filler') that costs 1000 mana and can be casted once per second, crit restores 600 mana, fight duration is 200 seconds.

1. With 0% crit you can cast 70 spells.

2. With 50% crit you would in average restore 30k mana and cast 100 spells (70 spells will give mana for 21 more, those 21 spells will give mana for 6.3 more casts, etc)

3. With 100% crit you would restore 105k mana and will be able to cast 175 spells (70 spells will give mana for 42 more spells, 42 will give mana for 25.2 more spells and so on)

As you can see going from 0 to 50% crit gives you only 30 extra spell casts, while going from 50% to 100% crit will give you 75 extra spellcasts. I think you agree with it, right?

Now closer to the bit that you fail to realise - both numbers (100 casts with 50% crit and 175 casts with 100% crit) can be casted within 200second fight without a problem. What you are missing is that for your model to be more relevant, the extra mana should not only increase the number of possible casts but also decrease the time between casts. Because your averaged time per cast that you list in the table will depend on the amount of mana that you have available.

I'll stress once more, extra mana WILL let you cast more HLs in usual circumstances. Just not because of speeding the casts (that you were rightfully questioning), but because of shortening the intervals between the casts. When you take this into acount then 'magically' your math will give the same increasing value for a crit.

In one of your posts you said that extra mana from crit at bigger crit % will require more time to cast HLs. What we are trying to say is that unless you spam HL already that extra mana will NOT require more time to cast HLs but instead it will allow to idle less within the very same time frame.

P.S. Previous poster said it nicely. Comparison between crit and mp5 is more misleading than helpfull at the moment.

 

EvadDeWahr

Correct me if I am wrong but is this not what RAWR does just with more sophistication?
Does it not look at the fight length that the user selects in the option section and work out the maximum amount of healing that can be done with the other parameters set?

If that is the case then it must assume that the user is OOM at the end of the fight???

And even in this case fight length is only an estimate.

 

Endoscient

When I said healing until OOM model, I meant modeling casting a predetermined rotation of spells (ex: 50% HL, 50% FoL) until you are out of mana and then stopping. So how long the modeled fight duration is would depend on your mana. Rawr models what spells you can cast in a fixed time period, using all of your mana.

 

Abeness

To be honest, I'm a little confused as to what the argument is about anymore. From what I read, the point made against me is this: If the player stacks more crit, they get more mana back; because they have more mana, they can cast more Holy Lights in the same amount of time thereby increasing the gain from crit. Assuming that is the point being made (correct me if I'm wrong), the variables I used change. Yeah, the paladin can cast more often, but that means he moves on up the chart and the Mp5 gain from crit goes up because the paladin is casting more often. So you are correct in assuming that more crit can allow for more casts, but if you increase cast speed, the equations change. In order to continue and make clearer conclusions, I need to know people understand and are on board with the idea that crit can be compared to Mp5. Hopefully we can clear up our misunderstandings on this to determine how the comparison can be used.

Although there is another point I'm confused about. If the paladin can heal the fight and the tank is kept up/healed enough to avoid unlucky avoidance problems, why would the player want to increase the cast rate? Wouldn't that be a waste? If so, wouldn't that mean the paladin has too much regen on his hands? To those I've had the discomfort of debating with: This is an actual question, not me making assertations on how to play.

The ideas you say that I've refuted are things that I either did not refute or corrected myself in a later post. I can only assume you thought I was refuting a lot of those statements through a misunderstanding of what I was saying. And arguments like in the second paragraph quoted lead me to believe that you did in fact skim over my posts. In a later post, I admitted leaving out crit's scaling abilities on the effective mana pool in my original posts, but showed that the conversion to Mp5 accomodates crit's increased cast amounts at a given cast interval. If you increase the interval, you can gain mana back faster and use more of it, but then your spot on the chart moves up and crit is shown to be worth more Mp5.

 

Endoscient

The argument is that your numbers for mp5 value of crit are not accurate for various reasons. The logic you use to back up your formulas is inaccurate and misrepresents the value of stats. The largest logical fallacy is that you use variable length fight durations, and then still represent gains as mp5 over different duration fights. You also use your margin error comparison (which was your justification for ignoring initial crit chance) compared to initial mana pool, when it should be comparing your computed gain to what the actual gain is. You only consider the first order returns of crit, and ignore the mana gained by crit has a higher chance of criting and restoring mana then the mana gained by mp5. There are a few more that I highlighted in my previous posts as well.

If the Paladin can heal the fight with no problem, then you have enough gear for it and you should start planning for the next difficult fight.

As for quotes about you refuting those statements, here are a few quickly found from your most recent 2 posts (besides the one I am responding to).

"So far, the only point I want to get across is that crit can be represented as a Mp5 value, and that the scaling properties of crit are so minimal that there is no way it breaks my point" (They are far from minimal, and will drastically undervalue crit if you ignore them.)

"It does scale in that it gives more casts per crit percent stacked, but those extra casts take time to get off. The extra time it took to cast the extra casts gained, the Mp5 equivalent was able to catch up." (No they don't, you can easily use more mana in the same fight duration.)

"It does give ~50% more casts, but so does the Mp5 equivalent." (Mp5 scales at a worse rate with base crit chance. I believe in this particular example the gain to mp5 casts is only ~25%)

"When I run your numbers, they rarely match up with what they should be." (Uses the equations I posted, and you have yet to refute their accuracy besides saying they don't match up to yours so they must be wrong.)

 

Abeness

At this point, I'd like to repeat my original process in a way that should put to rest some of these accusations that I've ignored crit's scaling abilites. No matter what initial mana pool the player has, a certain crit percent always gives the same average cost per cast. At 20% crit for instance, the average cost per cast when you have 30,000 mana is the same as the average cost at 100,000 mana. I just took the average saved per cast, as it is a constant as well (baseCost - averageCost) and used that to find the average savings per cast. This amount saved per cast is the same at any initial mana pool amount at a given crit percent. All I did was take that amount one saves per cast, and converted it to Mp5. So no matter how much mana you have, the Mp5 value from crit does not matter.

The more crit you have, the more mana you gain per point of crit. You gain more mana, so why would the Mp5 value still be the same per point of crit? Because you are still casting at the same speed. The paladin gained x more casts, but at the given cast speed (which, again, is an average for the fight). So it takes time to get those extra casts off (x casts gained multiplied by seconds between casts gives you how long it takes to use those casts). Your argument is that the paladin can get more casts off because they have extra mana. As I said in my last post, once you cast more, you move up on the chart and crit becomes better.

That is what I'm confused about. Up to this point, I've been told that if the paladin has extra mana at the end of the fight, he wasn't casting fast enough. But why cast more if they don't need to?

As I said above, the scaling properties are actually included in the original equations, just not clearly explained. The margin of error I spoke of earlier was not actually an error. I had forgotten to account for the time it took to get off the gained casts, so there is no error. As far as your numbers go, they aren't as wrong as they are misleading. You continue to leave out cast speed which is a key point in valuing crit. You always assume the paladin is using all of their mana in the given time period (even when they have 200,000+ mana), and doing so would require the paladin to cast faster. Casting faster makes crit better (in terms of regen/mana saved). For instance, one of the charts you posted a while back showed mana gained in a 7 minute fight from crit and Mp5.

The only way mana gained from crit would continue to increase as initial mana pool does is if the paladin was always using all of his mana in the 7 minute window. But to use more mana, you have to cast faster. And casting faster and faster increases the mana back from crit, as I have said. You're not so much arguing with me; you're actually trying to prove the same point as me. But my numbers do account for crit's scaling.

 

◊ madsushi

Let me step in and see if I can explain concisely what (I believe) Abeness is saying.

In a 60 second fight, a Paladin who is chain-casting Holy Light (at a 1.5s cast speed) will be able to cast a total of 40 Holy Lights. Therefor, the amount of casts in the amount of time is fixed: the Paladin cannot cast Holy Light faster than 1.5s without changing the Haste variable.

If a Paladin gets to cast 40 times in this fixed-duration 60 second fight, he will save: 40x, where X is the average mana saved per Holy Light critical strike. In that same 60 second fight, a Paladin will gain 12 effective ticks of their MP5 values. This value is fixed based on fight length, and can be referred to as: 12y, where Y is the MP5 value.

Abeness is trying to find the ratio between 40x and 12y. The question becomes whether this ratio is even relevant, considering the fact that very rarely are fights "fixed length" and that a Paladin CAN generally exhaust all of their mana in a fight if they wanted to. Because running out of mana, the number of casts per interval, and amount of healing necessary can change drastically between attempts, the ratio loses its accuracy on any non-specific fight. Add in the ideas of casting non-Illumination proc'ing spells and Flash of Light, and there's really no fight for which the model makes sense.

Author of the Rogue column on WoW.com: Encrypted Text

 

Endoscient

I know Abeness was saying exactly that. But one of the mains points me and others have been trying to make is that those numbers are meaningless to real situations and fights. Since in real fights the Y mp5 value from crit that is computed that way will actually give you more extra casts then having Y mp5 on gear. Due to reasons that have have been stated very many times, and its useless to keep repeating them.

 

Palados

double post

 

Palados

Because we can?

On a more serious note - you never know how the fight will go. Maybe tank will have unlucky string of no avoidance? Or someone doesn't move out of fire and you will have to heal them, etc. The goal of theorycrafting for healers is to maximize something (HPS, total healing done, effectiveness, etc) in the worst case scenario. For us the worst case scenario is when all your mana is gone at the end of the fight. And in this case you can't ignore the fact, that adding more mana will let you cast more spells in the very same fight. Maybe that extra few HL that you are able to produce from 'crit' mana will save the day.

In principle you agree, that being allowed to cast more within the same time frame will shift the crit value up (since it lowers the average cast time in your model). Our point it that this particular scenario is what we have to look into. That is why crit value for palading grows up non linearly. It's not that your math is bad. Your assumption about fight are not really relevant for worst case scenario. And I stress once again - we gear and theorycraft for those, since for 'usual' fights you may as well go for spirit and be fine. As soon as you look at spending all your mana within a priori fixed fight duration that is too long for really spam HL non stop (aka most fights), then your math gives the same results as graphs posted by Endo.

Really, we are arguing about the model to apply your math to. Fixed fight duration with fully going oom in the end is the best model for 'worst case scenario'. Compare it with the real life problem - you have some device with certain output power. To calculate the amount of energy it is able to produce, you will have to choose the time first and then you can do it. In our case you have a paladin with certain amount of power (crit power ). To calculate how much mana he will produce, you will have to fix the fight duration first and make an assumption about amount of mana he will have at the end (0 for the worst case). You are saying that we have to fix the amount of damage tank is getting and thus look only at cast rate as a fixed parameter, but that is not relevant. Because this rate in the worst case of going oom at the end depends on amount of mana available (thus depends on crit per se).

If you don't like 'oom at the end' idea you can go further - you can say that paladin will have x% of mana not used at the end of the fight. And this is still better than looking into the average cast rate. Because we theorycraft not for a specific fight where amount of healing needed is determined by incoming tank damage, but for the worst case where we barely have mana to do our task, no matter how big this mana is (realistically we really rarely have enough mana to not be able to spend it within real fights).

Another argument for 'cast till oom at the end' case is that if you have mana at the end and no one died then all your math is void, you would better go stamina for your own survivability or haste for better reaction instead of effectiveness stats. Thus to compare crit and mp5 we have to look at the case where they matter - casting till oom no matter how much mana you have. Otherwise such comparison is scholastic.

 

Jackinthegreen

Apologies for this being a bit unorganized, I'm not always so good at putting things in the right order. All the pertinent info is in here though, and it is presented as best I know and have found. Please feel free to look for and post errors.


To find the value of crit compared to Mp5, we need to look at everything that affects the mana returns on crit in the first place.

Mp5 of course is the constant by which we're comparing crit to, since it is constant and reliable no matter the fight length, cast speed, or anything else save for gimmicks like Vezax that disable it which aren't counted here anyway. The only thing that varies is how much we have and how long it is useful to us.

Of note is that crit scales from Mp5 based solely on the fact that you need mana in the first place to cast the spell that can crit. Mp5 does -not- scale from anything else save for how much you have, and how long it is useful. Therefore, to compare crit to Mp5 we have to take the returns from crit that -don't- scale from minimum or maximum mana. We either can cast, or we can't.

Now we have to look at crit: Critical strikes give us back 60% of the base cost of the healing spell in 3.1. Thus, crit scales with itself so long as we are casting/capable of casting. The main issue with actually turning crit into Mp5 is the fact that Crit is burst return versus Mp5's constant regen. The following list is what affects crit's mana return.


Base cost of spell(s) (and hence the return from them)
Effective chance to actually get the return (crit itself)
Speed of casting and therefore proccing the mana return
Length of time casting is actually effective (fight length)


The last two in there can be consolidated to simply seconds per cast (or casts per minute if you prefer) so it can be better dealt with due to average cast speed being easier to place than the final fight length. The main reason to compare various actual lengths of fights is for the use of total mana gained, lost, et cetera. Useful in the overall scheme of things, but to solely compare Crit and Mp5, final fight length isn't needed.



To calculate the possible returns from Holy Light, we need to factor in reduction effects first. 1274 is the base cost, and so we get 764 mana return from it on a crit. Libram lowers HL to 1161. Each of the possible 5% reductions are calculated after the libram, so they are worth 58.5 mana off each lowering HL to 1102 then 1044. For end game comparison and easier calculations, let's stick with 1102 per HL. Our return is still 764 if we crit, so that's 338 we "spent" on the HL. Let's just call it 764 mana every time we crit though. If cast every 2.5 seconds for 250 seconds (100 casts) at 50% crit, we can expect to get on average 38200 mana back, so this technically equates to 764 mana every five seconds. This can be simplified to say that for an average cast time of 2.5 seconds, and a 50% chance to crit on HL, we can expect to get the equivalent of 764 Mp5 if only casting HL.

For Flash of Light, the numbers of course change to a 307 base and lowest possible 291. The mana return from Flash is 184, so equivalent to 184 Mp5 at 50% crit if casting every 2.5 seconds. Obviously using a mix of HL and FoL will result in a number between 764 and 184 (assuming 50% crit of course).

This shows that crit is indeed an amazing stat, but the real question is how crit -rating- and Mp5 compare to each other. 100 spell crit at 80 for us gives +2.17% crit chance, so each point of crit is .0217%.


And my brain just died.. The follow-through on this requires more math than I can put out right now. Hopefully it has made points and helped though.


After reading Palados's post though, I guess other things have to be added. Mana expenditure is a given on all our spells, and even if we get a crit on an HL with 4pT7, we still used 276 mana. Not huge, but when you're also casting Beacon, SS, and all our other spells, it adds up.

In short, the calculations are possible, but it would take a lot more than just making a graph to compare them. An app that can show the numbers based on the gear we have (and could have) that shows the changes dynamically would be the most useful, although possibly not practical to create given everything involved.

 

Palados

Such program exists - RAWR does exactly that, using simulations and taking everything into account.

Also if you think about it, the gain from mp5 is not fixed and depends on your crit. Since crit effectively increases the gain from mp5 by 1/(1-0.6crit). The best way to check it - add a prismatic socket in RAWR and insert the 8 mp5 gem there. You will see that actual mana gain during for example 5min fight will be bigger than 480 mana that you would expect from 8mp5. The more crit you have, the more extra mana you will get. So maybe even in 3.2 - with huge int and mp5 - crit items might be not that bad after all.

It was already said a few times - for paladins the best way to treat crit is to represent it as an increase of effective mana pool. It is so fight dependant, that comparing crit and mp5 as effectiveness stats using some general formulas is almost equivalent to comparing apples and bananas. Just use rawr with specific fight settings and you will see how much mana will be returned from extra mp5 and extra crit gem.

To everyone who still thinks their math is relevant - answer a simple question using your math:

A paladin is going to try Mimiron hard mode with his guild (for first kill fight length is usually really close to 10min). He just got a new belt and wants to use prismatic socket from buckle for extra mana, either mp5 or crit (let's ignore int gem for the moment). Which gem should he choose - mp5 or crit one?

As you can see, to answer this question you need to take into acount so many things, that any simple crit-to-mp5 conversion table will fail. Unless you fully simulate the fight as rawr does. But then why would you need a simple table if in order to use it you have to make complicated calculations taking into account pretty much everything (like using mana for non crit spells, etc)?

@ Abeness

after corrections you made your math is OK for answering the question how good is mana return from crit assuming a fixed cast rate. However no one cares about this value. What people really care about is how to maximize their potential mana gain for a specific fight or set of fights. To answer this question you will need a complex simulation that will find the cast rate first for given fight parameters. And this simulation will take into account all those things that Endoscient and other people are trying to explain. Thus your table while not being wrong per se is not relevant for making a choise between mp5 and crit in real situations.

 

Jackinthegreen

Technically speaking, in 3.1 ALL mana returns for us can be increased by 1/(1-0.6crit). Crit thus scales will all mana period, which basically shows why int is so strong for us. The more mana we can use to cast, the more we can regenerate. In 3.2 of course it'll drop to 1/(1-0.3crit) but we'll find ways to compensate.

Now I need to download RAWR again...