The 320/120 rule for Blms is still applicable, right?

[Tenjou]

No, your numbers are inferior, 330/130 is the bull god number. 95% capped magical accuracy with 330/130.

Hello all, I'm very interested in magic accuracy. I would like to be able to accurately model resists, but to do so I need to know a few things:

1. Is there a base magic accuracy, and if so, is it something like 50%? Is macc a direct analogue to physical accuracy?
   
   2. How does skill play into the equation? Is it something like every x skill adds another 1% to macc? Same thing for INT?
   
   3. Since 95% seems to be the cap, is there a minimum cap of 5%, or at all?

I invite anyone who can understand a little LUA to check my model so far:

Spoiler: show

http://pxi.pastebin.com/f1277774e

Any help and information is greatly appreciated, as we aim to provide the most accurate models. Once we have an accurate model, we could even improve on it later

[heavensword]

For NIN:

FFXIGEAR.com

Magic Accuracy 320 (269 base skill + 16 merits + 20 gear skill + 15 gear Magic accuracy [Mythic])

INT 120 (Taru +5 merits, Cream puff, +36 from gear)

Might be interesting to see what you could do solo, and still melee (even though the melee stats are horrible). Better to use staves with grips & obis, but ugh @ the inventory.


I'd rather duo/trio with a COR/RDM and get a MAcc/MAtt roll+Haste.

[SephYuyX]

Fun facts:
Kanican
1 INT = .75 MACC/Skill (on average, 1 and .5 depending)
2 INT = 1 MAB
1 MACC = 1 Skill
1 Skill = 1 MACC (.9)
1 INT ~= 3-4% Magic Critical Hit

[SephYuyX]

Upon further review, nvm.

But,

Someone needs to figure out an equip calculator that takes care of all those equations above.

[Tenjou]

Thank you for your help! Those links are very useful. But still, it leaves me very confused...

If 1 skill corresponds to 1 acc, wouldn't a BLM at 75 be 276% accurate? (or 268.4%) Also, what about the poor level 1 BLMs, they have no skill yet, and not too much INT. I guess there must be some base macc...

[Demosthenes11]

http://users.telenet.be/jandebruyn/EnePud.png

fuckin hate you tarus. My nuke setup is a little better than yours (morrigan body, novio, mahatma legs since full ele merits) but in the gear I only have 600 mp...

bullshit!

[Enedin]

Lol :D

[Izildur]

320/120 rule I'm sure every endgame Blm should aim for as a base - but many times you have to look at what your fighting and change your gear balance accordingly and see if the driving of regular and bigger numbers will be more skill based or int based etc

ie

the "320/120" build wont get an efficient/effective result from lolJormy

[Argettio]

Thank you for your help! Those links are very useful. But still, it leaves me very confused...

If 1 skill corresponds to 1 acc, wouldn't a BLM at 75 be 276% accurate? (or 268.4%) Also, what about the poor level 1 BLMs, they have no skill yet, and not too much INT. I guess there must be some base macc...

The target has a 'magic evasion', which is compared to your total magic accuracy, that gives you a 'hit rate', which is you chance for your nukes not being resisted.

Adding 1 skill will increase this 'hit rate' by 1%.

So if you went and nuked the bees in a starter zone then effectively you have 268% accuracy (which is obviously impossible, so it would be truncated to 100%).

[pchan]

1 skill is not always 1% hitrate. The lower the base rate the lower the effect.

[Argettio]

1 skill is not always 1% hitrate. The lower the base rate the lower the effect.

True, but I was trying to simplify it for him (although I should have mentioned it)

[Tenjou]

Ah of course, magic evasion. I wonder how we can determine that though...

Alright, I've decided to try using magic evasion for mobs based on a C-level skill for now. It seems to correlate about right with everything else that's been said. If you give a level 88 HNM (I tried on Byakko) a B-level skill then 330 skill gives you an 89% to not resist. This seems about right.


http://pxi.pastebin.com/f4021a8a3 Let me know what you think, and thanks everyone for all the helpful insights and ideas.

[Stumps]

I judge the 320/120 rule as a starting point for BLMs and a goal for SCH's, given that for one job getting to 320/120 isn't all that difficult, but for an SCH it can be more +1 intensive.

[Argettio]

Ah of course, magic evasion. I wonder how we can determine that though...

Alright, I've decided to try using magic evasion for mobs based on a C-level skill for now. It seems to correlate about right with everything else that's been said. If you give a level 88 HNM (I tried on Byakko) a B-level skill then 330 skill gives you an 89% to not resist. This seems about right.


pxi private pastebin - collaborative debugging tool Let me know what you think, and thanks everyone for all the helpful insights and ideas.

I skimmed through it, but didn't notice any level correction. (possibly because I just missed it?)

As melee accuracy is level corrected, I think it is safe to say magic would be too.

other than that, interesting maths.

[Kaeko]

Simple, yet very amazing program ;o Considering modeling is such a powerful tool in real world research, I feel so dumb reading this thread for not thinking that it could be used for this as well.

Here are a couple pointers...

1) Major Flaw - the 0.9 skill above 200 base skill has been proven FALSE within alpha=0.05 by Robonsto using Lodeguy's data. Check and adjust lines 25-29 of code.

Unfortunately this was found by Robonosto after I did my dumbed down version of this results so most people don't know this. He was able to show that the 0.9=1 thing was false later on. For all intents and purposes, I think you can consider 1 skill = 1 MACC at all skill ranges. If you would like the formal postings I can dig those up for you.

Also, it should be noted that the 0.9 skill = 1 ACC for melee was adjusted in an old PATCH, clearly stating that something was changed in terms of skill and acc at higher levels, but for MELEE. This was in some archaic patch but someone was able to dig up the patch notes. It's late so I don't feel like digging but I can find that if you'd like too. This could explain away why melee has the 0.9 = 1 but elemental magic does not (also note we only know about elemental magic, not other forms, but just assume them to be the same or similar).

2) Level Correction - statistically shown to likely exist, but unknown values.

As others have stated, there should be a level correction for magic as well (Lodeguy suggests this, but cannot give precise values on it). This would be very difficult to show without running another grueling study like Lodeguy did, but with a well tuned modeling program like the one you started it should be easier to back or debunk the melee level correction for accuracy. That seems like a good place to start.

3) Elemental Resistances - we basically don't know anything about them

Not sure what to say about this other than we don't know anything and it almost certainly affects the outcome. Do you get negative changes for being 'weak'? Is it a percentage or is it a raw decrease? etc.

4) INT Difference Tiers - Lodeguy only shows that a tier at dINT=10 exists. He does not exclude the chance of others existing.

At the moment your code only checks for the dINT=10 while other tiers can exists. Of course we don't know if they do and where they would occur but it's likely another kink in the formula. One way to show others exist is to give yourself extremely high INT and compare simulated vs. actual on a high resist but low INT mob (so the dINT is extremely large). Alternatively I'm thinking you could do a low resist mob with a very negative dINT...

I need to think a bit more on how to use this but really if anyone actually does end up finding the MACC formula, it would probably rely heavily on a simulator, so this is big.

[Tenjou]

Hey all, thanks for the input! Here's my new model with the new advice. Some changes are that it uses the same alpha (level correction) as blue magic apparantly does, and that I fixed magic skill above 200. It also uses C-level magic evasion, and B-level for mobs above level 75. At 276 magic skill I only have a 40.5538% chance to not resist against a level 88 Byakko

As for elemental resist, I plan to handle that separately. And finally, I invite anyone to follow the link in my sig to learn more about how to help test this model. I'm usually on the webchat if you want to try it now, I have the model running. Pretty soon, it will be possible to easily edit the model yourself and try it out, and compare to retail.

Oops, here's the link lol: http://pxi.pastebin.com/f1f403419

[Snuts]

Wow ... nevermind ... maybe I should try scrolling down.

:facepalm:



I think I missed the original post for this.
Its not a straight 320 skill and 120 INT, right?
I'm sure theres some conversion for MAB and Macc.
Anyone have that info?

[SephYuyX]

Its not a straight 320 skill and 120 INT, right?

No, it's straight INT and EleSkill, so you have to have a total of 120 int from base and + green numbers, and blue 320 skill. No conversions can be taken place to achieve these numbers, aside from magic accuracy, but thats not really a conversion. So you can have 297 elemental, and then have on an omega ring, and still fulfill 320.

And if youre looking for other conversions, look at the first page.

[Pindle]

So you can have 197 elemental, and then have on an omega ring, and still fulfill 320.

Uh?

[Snuts]

So having a MAB setup is completely fail except for soloing?

I had been using MAB mostly because it was easier to get and I had been saving my gils for my SAM.

Ele. Staves MAcc +20
Grips Macc +2
Sorcerer's Petasos Skill +10
Elemental Torque Skill +7
Goliard Cuffs MAB +4 MAcc +4
Igqira Weskit Skill +5 MAB +6
Penitent's Rope INT +5
Cobra Trews MAB +3
Cobra Crackows MAB +2
Solitaire Cape MAB +2
Sorcerer's Ring MAB +10
Snow Ring INT +5

Now while this setup gives me the necessary 320 I need and also allows me to break 2k pretty easy on puddings, its pretty garbage for anything that counts due to the horrible INT. Does MAB just play absolutely no role?