Black Mage Sanctuary - A Guide to DPS

[Neophyter]

My math-fu isn't the strongest, but that seems quite a lot lower than DRG, BRD, MCH, which is DET/7290+1 or 74(?) for 1% increase.

Yeah, it's the thing I noticed while I was crunching my numbers and I don't have much gear in order to verify. However, I don't agree with the formula used for DRG, BRD, and MCH because it means that even at 0 DET from gear, you get a 3% increase in DPS, which doesn't make sense to me. Also, I'm under the assumption that DET is a linear increase in DPS but that could change once we get higher ilvl gear.

[drizzle234]

Took replacing 2 healers and 2 dps to finally get a group that works. Would be that we clear on my sloppiest run ever lol XD

A1 Savage clear: https://youtu.be/i46G2CZeGiI

[Neophyter]

Update: I dun goofed on some formulas and after some more number crunching, I came out with new ones.

Spell Damage Avg = (Potency * INT * (WD / 25 + 1) * ((DET - 218) / 5946.91 + 1) / 891 * Buffs
DET Dmg Multiplier = [(DET rating - 218) / 5946.91] + 1
Crit and Spell Speed remain the same.

I updated the spreadsheet to reflect this.
This gives DET more weight than I previously gave it.

[Kenji1134]

Neophyter, I used your formula form with my numbers, and got some slightly different coefficients, though the idea is roughly the same. Note that my equation is basically yours but in an expanded form.

Base Potency = [ Int*WD*(Det-218)/350,000 + Int*WD/22 + Int*(WD-218)/2200 ] / 970. This somewhat approximated set of coefficients gave me fairly good results, under 0.5% error on most terms, at most 1.5% over.

Looking at your spreadsheet, I like the idea of examining how much of a stat it takes to gain X% more damage. So I extrapolated the above formula by changing one stat at a time to see how much it takes to get 1% to 10% potency increase:

Here is the raw chart:

%... Int WD Det
1% 835 99 372
2% 844 100 447
3% 852 101 521
4% 860 102 596
5% 869 103 671
6% 877 105 745
7% 885 106 820
8% 894 107 895
9% 902 108 967
10% 910 109 1044

And the associated changes between each %:

Change Int WD Det
1% 8 2 75
2% 9 1 75
3% 8 1 74
4% 8 1 75
5% 9 1 75
6% 8 2 74
7% 8 1 75
8% 9 1 75
9% 8 1 72
10% 8 1 77

Giving us the following average stat change per 1% damage increase... Considering Int, WD, and Det...

8.3 Int/%
1.2 WD/%
74.7 Det/%

Based on this we can say that 1 WD is worth 6.917 Int, and 1 Det is worth .1111 Int, or 9 Det = 1 Int in terms of % increases.
This also says that if you were to somehow have the choice between 2 weapons, where one has 1 more WD, and magically the same Int, the other one needs to have 62 more Det to make up for it.

I will add crit to these results sometime in the future, when I am in a mathematical mood.

[Viuiji]

For endgame raiding (static is working on A1S atm) is it ok to stay Elezen DW? FFXIV is my first ever MMO and while I love it I do see many of the other BLM's are Lala??

Is there an advantage to being one over the other for raiding? I know that Lala have better PIE whilst the Elezen have higher Int, but am unsure if I should change race. I have averaged 1106 on Faust over the last 10 Faust clears and understand I need to keep practicing, but will a race change help that at all?

Thank you

p.s. Thank you Aikaal for this great guide! has helped me alot to love my BLM even more than 2.0 I enjoy that it is more challenging now!

EDIT:
Scoured through quite a few pages and saw Aikaal mention he had moved to Elezen for the Int. SO I guess I'll stay as is for now.
Although, from what I read, it doesn't seem to matter much?

[Dervy]

And those values are also dead-on the values I achieve with my Dragoon Model also.

73 DET = 1% (Accurate)
11 STR = 1% (Rough)
1.4 WD = 1% (Rough)

So we're all roughly similar with our damage models.

[Xenitan]

Is there a particular reason we're expressing DPS increases as a percentage of base DPS? Seems like that'll be confusing later. I'm not a fan of expressing it in terms of your primary stat, but at least it makes sense that way. A 1% increase in damage can change in terms of "value" very quickly and we look at DPS as inherently linear anyway.

My numbers were:
9.29 Det = 1 Int
5.07 Int = 1 WD

Which seems reasonably in line with Det and I'd say the Weapon Damage is just a product of your interpolation which could explain a bit. I also only looked at 100 and 102 magic damage items which could indicate I'm inaccurate. Could go either way and in all honesty because WD is so important you really could get away with just assigning each WD value its own det/int forumla.

[Neophyter]

One more update regarding Spell Speed and its effects on DoT damage:

Spell Speed DoT Damage Multiplier: (SS - 354) / 5544 + 1

So the average spell damage for DoTs (AKA Thunder) is:

Average Spell Damage = (Potency * INT * (WD / 25 + 1) * ((DET - 218) / 5947 + 1) * ((SS - 354) / 5544 + 1)) / 891 * Buffs

Also, I revised the formula to make look better. This formula has a 1% - 2% error.

Lastly, as for Xenitan's remark, while I do prefer the old method of doing stat weights, percentage-based stat weights does have its uses. It also isn't that difficult to convert the percentage-based stat weights into the old method. Given Kenji's weights of:

8.3 Int/%
1.2 WD/%
74.7 Det/%

In order to get WD in relation of INT, you just divide the 8.3 INT by 1.2 to get:

6.91667 INT = 1 WD

And do the same with DET and INT to get:

1 INT = 9 DET

[Kenji1134]

The reason expressing potency changes as a % of some previous base would be to provide a common metric between classes.

For example, using Dervy's data, a DRG using the 190 weapon and stats for a similar ilvl has a base potency of about 4.08.
A BLM using the 190 weapon w/ an overall 180 gear level (me), has a base potency of 4.68.
A WAR w/ the 190 weapon and full 190 gear has a potency of 3.08.

This is due to the fact that each class has a different amount of weapon damage, stats, buffs, and rotational potency.
For the most part we assume that the rotation maintains all buffs, and outputs the optimal potency/sec, which we cannot really optimize further. That just leaves us with the base potency for optimization.

So say we compare BLM and WAR.
BLM has higher WD, higher stats, and base potency than WAR. So one would say that for a WAR, 1 WD (for example) is worth 10 STR. Okay... but how does that affect the actual potency value? Does 1 WD add .05 to the 3.08 that I have, or does it add .1? More, less?
You can apply the same logic to BLM, unless you KNOW exactly what you have now, it's kind of difficult to say how much a stat will affect things.

We can circumvent that a bit by dividing out the unknown quantity and considering how much of a given stat we need to elicit a X% increase. Surprisingly (at least to me) that was fairly linear across 1 variable. Still because the variable are linked, I am certain that changing multiple variables over time will bring forth some nonlinearity, at which point the model needs to be fed new data using the new gear, but atm I think we can use these models to estimate out 5% on each stat. And given that we have 5 stats, including crit and speed, thats 1.05^5 = 1.276, a "calculable" 27.6% max dps increase without updating the model.

Either way, if we say that 6.92 Int = 1 WD, or 1.2 WD = 1%, we can still use them to gauge the value of changing stats.
Though I will say that IF someone were interested in comparing complete gear sets as a whole, it would be cumbersome to use the weights. Instead it may be simpler to plug the resulting stats for each gear set into the complete model (w/ crit and speed modifications), to have a finalized Base Potency Value for a specific gear set on a specific class, and repeat this process for other gear sets in order to compare the Base Potency Values.

[Dervy]

As everything is a multiplier, we assume. So always, X Amount of DET will be 1%. It's also kinda built into the Damage Formula, well, mine at least. DET/7290 is a function in my damage formula for Determination.

1/7290 = 0.000137
1%/72.90 = 0.000137

Scales exactly the same.

One other thing, is that for Summoners, my model actually does WD/23. So, it could be vastly different for jobs, or somewhere, each jobs scale X stat differently.

What we really, really need to check is get a linear regression going of X STR with base DET + 0 WD... But, I'll leave that for the SimFF/Damage Formula Thread.