Thread: Black Mage Theorycrafting - Best In Slot

Update for patch 2.1
See this post


Hello everybody. I've done some theorycrafting on how to gear up my BLM.

Information
The goal with this study is to find out which endgame items that currently are Best In Slot (BIS) for BLM in normal circumstances. It's hard to figure out how to weigh in determination, spellspeed, intelligence and accuracy. You want to maximize your damage while still maintaining a certain level of accuracy. This is why I've created a program that automatically tries all combinations of items and measures damage and accuracy.

Spellspeed and GCD
One of the harder tasks is to determine how to weigh in spellspeed. First of all, we need to know much of the dps that is affected by spellspeed in a standard black mage rotation. I've tested a simplified version of the rotation with FFXIVAPP:

Thunder II -> Fire III -> Fire Loop with Raging Strikes -> Blizzard III -> Restart. Use proc spells when possible and cast raging strikes when off cooldown. Scathe when reasonable.

The thunder dot is the only element in the rotation not affected by increased spellspeed. This due to the fact that dots have fixed tick rates of 3 seconds. In my rotation the thunder dot was approximately 2.2% of the total damage. Based on that, we should weigh spellspeed with a factor of 0.978. But let's give us some margin for numerical errors, human errors while playing, positioning errors etc. I will use 0.95 as spellspeed weight (0 = none, 1 = full). So what does that mean? It means that I will only use 95% of the percentual damage boost that the decreased GCD will give us.

There are studies done on GCD and how much it's affected by spellspeed. But I haven't found anything that's 100% accurate and last time I checked people are still debating this on forums. I will keep it simple and use 2.5s as GCD and say that 100 spellspeed equals 0.1s decreased GCD. This will do for now.

Damage formula
I use the formula for WHM Healing as damage with potency 100. When I tried it with various weapons and gear I found it be 99.5%+ accurate when determining base damage. I think this is the most accurate one out there and this is the one I'll stick with until proven otherwise. Please prove me wrong if you believe otherwise since this is the most crucial part of this study. Credit goes to valk and volpes for creating this formula.

Weapons and items tested
I'll focus on the Relic+1 weapon, since most people will use that, but I have looked into allaganScepter + ancientBuckler as well. Itemwise, it's sorcerers and heros vs. allagan. The rings are unique and therefore I'll lock them in place. That gives us a total of 9 swappable gear positions, resulting in 512 different item sets that need to be tested.

Food buffs
All serious players will probably use some food buff while raiding. I've included a couple of buffs as well in the calculations.

A = No food buff
B = Lava Toad Legs
C = Pineapple Ponzecake
D = Apkallu Omelette
E = Black Truffle Risotto

Accuracy
This will determine how much misses you will have. The ideal is of course to have zero misses when raiding coil. I haven't found a 100% accurate limit for this yet, but I have somewhat of an understanding on where I want to be. 430 will generate some misses occasionally, but 440 is probably safe.

Results
The results from my program is outputted into text files. Each text file represents a different threshold of accuracy. Within each file you find results for relic+1 with various food buffs, and allaganScepter + ancientBuckler with various food buffs. See the summary table blow for all possible variations.

Code:

Maximum damge with accuracy >= 440
    Weapon  A       B       C       D       E       
   Relic+1  288.74  290.47  290.64  290.28  290.25  
   Allagan  288.76  291.06  290.70  290.97  290.28  

Maximum damge with accuracy >= 435
    Weapon  A       B       C       D       E       
   Relic+1  289.06  291.85  290.97  292.23  290.57  
   Allagan  289.66  291.06  291.60  291.44  291.18  

Maximum damge with accuracy >= 430
    Weapon  A       B       C       D       E       
   Relic+1  289.62  291.85  291.55  292.23  291.24  
   Allagan  290.21  291.96  292.16  291.44  291.73  

Maximum damge with accuracy >= 0
    Weapon  A       B       C       D       E       
   Relic+1  295.83  296.69  297.83  297.17  297.48  
   Allagan  292.34  293.20  294.30  293.67  293.98

I will only paste the results for accuracy limit 440 and 435. You have to dig into the text files to find out more

Relic+1, Accuracy >= 440, No food buff

sorcerersHead
sorcerersChest
sorcerersHands
allaganWaist
allaganLegs
allaganFeet
allaganNecklace
allaganEarrings
allaganBracelets
herosRing
allaganRing

Relic+1, Accuracy >= 440, Pineapple Ponzecake

allaganHead
sorcerersChest
sorcerersHands
herosWaist
sorcerersLegs
allaganFeet
allaganNecklace
allaganEarrings
allaganBracelets
herosRing
allaganRing

Relic+1, Accuracy >= 435, No food buff

sorcerersHead
sorcerersChest
allaganHands
herosWaist
allaganLegs
allaganFeet
allaganNecklace
allaganEarrings
allaganBracelets
herosRing
allaganRing

Relic+1, Accuracy >= 435, Apkallu Omelette

sorcerersHead
sorcerersChest
sorcerersHands
herosWaist
sorcerersLegs
allaganFeet
allaganNecklace
allaganEarrings
allaganBracelets
herosRing
allaganRing

Final words
This study was originally done for my own private use only. But I decided that it might be interesting to someone, so I gathered all the data and posted it here. If you use any of the data in any way, please think of these things:

• I can't guarantee it's accurate in any way. The damage formula is unofficial and I may have missed a lot of factors. My assumptions may be way off, and my code might be buggy
• Parts of the formula, like how critical damage is factored in, might change in the future. Therefore the information in this post might change over time. Nothing is written in stone here
• Always test things yourself and make your own decisions.

Thanks for reading this far I'll try to update this post whenever I have new information.

Originally Posted by FCrowles

You can edit your post and have unlimited character space.

Weird system. Makes no sense. Sounds like an exploit. And SE usually bans for exploits Sounds not worth it

Update: I merged all posts. Seems to be the standard on this forum to edit to make posts larger, so I did that as well. If an admin can delete the deleted posts, that would be nice

I don't think Critical Hit Rate should be the focus of BiS for BLM. If my testing results are correct, DET >> CHR =/> SS

I tested Determination vs Critical Hit Rate in a mixture of iLVL70 + iLVL90 gear for BLM by attacking training dummies 3,500 times per test. The two tests I did had the same equipment except for the pants: a non-materia melded Vanya Breeches of Casting vs Darklight Breeches of Casting, the only thing different being +18 DET on the Vanya against +27 CHR on the Darklight. Using Blizzard II for the attacks against 3 training dummies, which is a burst attack and where 90% of a BLM's DPS comes from, I came up with the following:

Vanya Leggings
305 DPS
240 average non-critical damage
12.69% critical chance

Darklight Leggings
303 DPS
238 average non-critical damage
14.36% critical chance

As you can see, the +27 CHR definitely manifested, but criticals are so weak in this game, even something like +2 average damage to a 100 Potency attack will give more DPS, and that was +18 DET vs +27 CHR. It takes an average of 2% actual critical hit rate to increase your base DPS by 1% (1.5X damage on criticals), which is just not a strong increase compared to Determination (though both are minor increases). Even in top level gear, CHR shouldn't surpass DET, the gap is just too large. See, the thing about criticals is, it actually gets worse compared to Determination since it only multiplies your base DPS. If you add 2% CHR and increase your base DPS by 1%, you're not actually multiplying your current DPS by 1.01. You're just multiplying your original damage, independent of criticals, by 1.01.

We'll need a lot more base DPS before Critical Hit Rate can topple Determination, from the looks of things.