Stats and how they work.

[Kaeko]

Regarding Magic Accuracy:

Tested using Lv1 Corroded Coblyns and Lv15 Moiling Moles both outside of Ul'Dah. I would gather ~10 of them together on 50 THM then AoE Blind, Slow, and Paralyze every minute. The goal was to find the "land rate" cap. This cap was 95% in XI, meaning no matter how much more MACC and dLVL you had over a mob, you would still miss 1/20.

Results as follows (ty Tachikoma for the wonderful parser)

http://img402.imageshack.us/img402/4306/macctest1.png

Couple things from this:

(1) There appears to be a cap on "land rate" of roughly 75%.
(2) The cap is not increased by the 8-man party bonus.
(3) Max # of mobs you can target with an AoE is 8.

Because these tests deal with low level mobs at the apparent land rate cap, these tests show nothing regarding how the MACC and MND stats change the land rate (for instance, trying to answer the "which is better?" question).

Overall pretty disappointed at how low the land rate cap is... Unless you can increase this 75% cap or there was some error/variable in my testing, you are essentially unable to create a build that reliably lands enfeebles. If we also consider how bad the duration and long recasts are on most enfeebles, I would say the enfeeble build got really hosed in XIV so far.

EDIT: From testing magic attack spells, I can say from an anecdotal perspective that the partial resist rate is very very low on Lv1 mobs and the full resist rate on the same mobs is near non-existent. Unsure of why the caps are so different.

[tachikoma]

Disappointed to see this, as RDM was my main FFXI and I always pushed for a monster enfeeble / enfeeble-proc build. I expect though since they are removing enfeebles from both THM/CON they are prepping for a new enfeebling buffing mage class such as arcanist>RDM for 2.0 and I expect them to clean up the enfeeble formulas by then. Or add direct +enfeeble % type gear.

>Glad the parser is helping, I would be happy to add any functionality you wanted kaeko that it doesn't have or if you find any bugs let me know. I have a new version out with more menu driven settings to try and clean it up, as well as some options for users who were unable to use the parser before due to EU log files and NM named enemies matching the players name.

[tachikoma]

Kaeko can you expand on this a little bit?

Also from testing MATK and INT, INT seems to do more than MATK 1 for 1 at all level differences

I don't need hard proof with thousands of casts but a basic breakdown would suffice. I know a guy in my ls was saying 1matk=3int which is the opposite of what you are suggesting.

[TacoTaru]

Has there been any test to check the ratio of Healing Potency to MND? I have my Wand of Tides melded with +16 Healing Potency, but I'm wondering how a Jade Crook with 16MND would fair against it.

[Drako]

Kaeko can you expand on this a little bit?



I don't need hard proof with thousands of casts but a basic breakdown would suffice. I know a guy in my ls was saying 1matk=3int which is the opposite of what you are suggesting.

even if thats true it doesnt mean that you gonna do more dmg stacking m.potency, from what i have seen (have feet double melded with m.potency IV and hat with double int IV) you kinda need both to really go up in dmg.

Tested mostly on lvl 40-42 mobs

[Neoscrilla]

Its probably like XI where as Int increased the overall damage cap, and MATK helped you hit that cap. Therefore, having MATK over the cap would be wasted or diminish horribly, and you wouldnt see noticeable damage increase until you added more int to raise that cap.

[dfdfpp]

Kaeko can you expand on this a little bit?



I don't need hard proof with thousands of casts but a basic breakdown would suffice. I know a guy in my ls was saying 1matk=3int which is the opposite of what you are suggesting.

I asked the same question over PM; Kaeko said that INT is better than potency by a decent margin but exactly how much better INT is than potency depends on the level. He also said he didn't bother doing more since 1.20 is throwing everything out the window. Again.

Your friend who's saying that 3 INT = 1 potency stuff is probably going off the conversion charts that were compiled early in 1.19. Despite the fact that attack and STR were supposed to convert at a similar ratio, all the decent parser results that have been posted here show that STR beats attack one-to-one when you're talking about anything that you care about your damage on. And I certainly trust Kaeko's math and rigor more than some official forum flunkies.

[Spider-Dan]

Its probably like XI where as Int increased the overall damage cap, and MATK helped you hit that cap.

That's nothing like the way XI worked for casting (MATK? I'm guessing you mean MAB, which is a multiplier). It's kind of like the old, flawed physical damage theory for XI (STR to enlarge the bucket, ATK to fill it up), but that theory is... well, old and flawed.

[Neoscrilla]

That's nothing like the way XI worked for casting (MATK? I'm guessing you mean MAB, which is a multiplier). It's kind of like the old, flawed physical damage theory for XI (STR to enlarge the bucket, ATK to fill it up), but that theory is... well, old and flawed.

Yeah I meant MAB, lol. Also your bucket analogy is exactly what I meant, thanks!

[Kaeko]

Sorry for the late reply. A lot of this will likely change in 1.20, but considering there are no enfeebles next patch, I feel like the market for magic attack attributes will probably go up quite a bit. Here are some summary charts that kind of show where I'm coming from. I'll write some conclusions in bold red at the end too.

http://img338.imageshack.us/img338/8004/matkint.png

The first series of tests involved Lv1 Corroded Coblyns outside Ul'Dah. The most significant parts of the first chart I can highlight are that:
(1) Adding MATK did absolutely nothing (suggesting MATK or the MATK/MDEF? ratio can cap)
(2) Adding INT gave a constant increase (suggesting no cap)
(3) The damage varies around +/- 5% of the average (or the MAX = MIN * 10%)
(4) The critical hit bonus is ~74-75% at dLVL=49, though it's likely capped


With these ideas in mind I then went to test on R50 Drubbers since they will give a dLVL=0. The next 3 shots will show the data collection...

http://img88.imageshack.us/img88/7081/matkint2.png

http://img836.imageshack.us/img836/6531/matkint3.png

http://img526.imageshack.us/img526/2375/matkint4.png

I pretty much just spammed spells at 3 different equipment sets. There was a control, one where only MATK was increased, and one where only INT was increased. I focused on the results where the % difference between MIN and MAX from the average neared 5% (because in these cases, I will have the average nearest the true value) and bolded / highlighted these windows. Summary of what I got out of these tests were:

(1) Adding +68 INT (198 to 266) gave about a 30-37% damage increase depending on spell tier.
(2) Adding +47 MATK (385-432) gave about a 9-15% increase depending on spell tier.
(3) Both the Crit % and the damage bonus when Crit procs have been lowered (according to SE this is a product of dLVL only)

There are any number of reasons why this result is the way it is. I cannot really offer a clear explanation as the formula is probably fairly complex. But for me, INT is better than MATK because
(1) Based on the coblyns test, MATK can clearly cap, whereas INT doesn't
(2) Test2 suggests quite a big bonus from INT over MATK. This could be stat/mob dependent, but from eyeballing, near everything I fight will see greater increases from INT stacking. Also remember you get more MATK just by having more INT in a practical setting.

Also from an anecdotal point of view, stacking INT causes your Shadowsear damage to increase significantly. I'm currently using a build around 365 INT / 490 MATK and it will 1-shot some types of R50 mobs. The chainspell double sear will 1 shot anything 54 or below (non-key holder/nms). Some of my youtube videos show damages if you're really interested, though of course this all changes next patch.

[Kaeko]

Also for Tachikoma, I'm really interested in testing evasion / parry / (and hopefully block) next patch, so really looking forward to when you can fix all the issues with evasion you mentioned before. I think another idea I'd like to see is being able to see the % land rates on more than 3 enfeebles. I haven't played in a bit so you may have already updated some of these things.

[Naver]

Anyone know when the next patch is coming? Would be interesting to see the changes for all of your work Kaeko. ><

[CrstyCptSteveWitherspoon]

Patch I think should be around the 10th maybe 15th latest. Getting this from Yoshidas last letter when he said they had to delay patch about 10 days from late november

[Niiro]

Kaeko's MATK/INT tests

Much thanks for this, leveling THM/CON in tandem atm and anything to make them suck less (especially CON) is a godsend -.-;;

[Corrderio]

That chart was a life saver. Looks like I'll just aim for a forbidden INT hat instead of going for the mortal hex mask.

Though I am curious. For the Base/INT builds why did you use Tier 2s but only Tier 1s for the M. Potency?

[Niiro]

Did some painfully boring DEX/ACC tests.

-Weapon used was a weathered gladius, I swung my DEX up/down with gear and conversion traits, unfortunately I didn't have much ACC gear
-All testing was done with defender II and obsess II up, testing on the lvl 54s had protect as well with sentinel/deflection/aegis boon II mixed in (none of this should have effected the results, just being thorough)
-I had the STR->INT traits on in an effort to make the mobs last longer (see above)
-All attacks were regular auto-attack
-Doblyns were chosen because they've only got one TP move which they can only use once (and doesn't effect acc/eva) and they were in the proper level range

Target: LVL 50 Sphene Doblyn
DEX 177
ACC 341
Baseline Hit rate: 84.08%

DEX 242
ACC 341
Hit rate: 86.04%
DEX+65 ~ +1.96%

DEX 177
ACC 366
Hit rate: 88.81%
ACC+25 ~ +4.73%

Target: LVL 54 Sphene Doblyn
DEX 177
ACC 341
Baseline Hit rate: 71.00%

DEX 242
ACC 341
Hit rate: 74.77%
DEX+65 ~ +3.77%

DEX 177
ACC 366
Hit rate: 75.34%
ACC+25 ~ +4.34%

==========================

(Tentative) Conclusions

-DEX appears to scale with dLVL, the higher the mob is the greater effect it has on hit rate
-ACC does not appear to scale with dLVL, and overall has a much more significant effect on hit rate point-for-point
-DEX doesn't effect crit rate for now (we were told as much and this may change in 1.20)
-*Not pictured in the summary images, there was no relevant change in block rate with the changes in DEX (1-2% variance, also told to us, also may change)
-Unless the functions of DEX change in coming updates, ACC appears to be superior in almost all situations unless a cap is found or if the effect of DEX on hit rate increases dramatically with greater dLVL

Summary pics of stats/parses

Spoiler: show

Any comments/questions that might help critique/clarify the info are welcome.

[tachikoma]

Also for Tachikoma, I'm really interested in testing evasion / parry / (and hopefully block) next patch, so really looking forward to when you can fix all the issues with evasion you mentioned before. I think another idea I'd like to see is being able to see the % land rates on more than 3 enfeebles. I haven't played in a bit so you may have already updated some of these things.

I'll get the changes in right after 1.20, I think SE is changing what text is written in the logs when you evade/block etc and I don't wanna redo the same code twice I'll update when this happens.

[tachikoma]

And its already hard to get mage specific materia, after 1.20 and especially 1.21 its only going to get worse. I hope I can get a set of double IV's for my mage done pre 1.21

[Neoscrilla]

I am still thinking that by increasing INT, it allows the cap of MATK to go up. Anyone else coming to this conclusion?

[tachikoma]

^ disagree the lvl 1 test shows steady increases in INT but adding MATK does nothing, would have to test to find a mob that lets you apply gear and hit cap on MATK then raise INT and see new #'s, then add in some MATK and see if it raised I still don't think it will work that way.