STR/DEX/PIE/ATK Testing

[Seiken]

UPDATE Patch 1.21: Tested on R52, new WS caps with Ifrit's Bow are DEX 295-300 and PIE 330-335. Both affect WS damage now, but cap really early with light shot.



Hello, after trying out the BG parser, I decided to do some testing on Archer to try and see which stats I should focus on. I wanted to see which of the 4 damage stats (DEX, PIE, STR, ATK) increased Light Shot and Heavy shot the most since in Moogle those 2 skills comprise ~40-50% of ARC damage. I would also like to point out that I have never done any stat testing before, so take my data/advice with a grain of salt. And any feedback would be greatly appreciated!

Materials and Methods

In order to check the stat effects, I first created a “base” stat-line where I was naked. I performed ~100-300 Light Shots and ~70-100 Heavy Shots to get a min/max and average damage. After getting a solid base damage, I increased one stat at a time, usually in 20-30 increments and recorded the new min/max and average. I then repeated parts of this data sampling on R1, R50, R52, and R55 enemies. Due to gear/material limitations, some unwanted stat changes had to occur, ie +4 attack on my +34 DEX gloves, so there is a small degree of error for some data. Also, I generally saw a trend of +/- 8% on the max/min for melee damage. For some reason R50 Drubbers fell outside this norm, so I excluded their data from my analysis. (So focus most of your attention on the R55 Feretrar and R52 Halbedier tabs on the spreadsheet, and the graphs/trendlines didn't transfer to google docs well)

https://docs.google.com/spreadsheet/...hl=en_US#gid=3

Dexterity

Dexterity could be the least or most important stat for Archer depending on several factors. If your primary concern is only instanced content where accuracy is 100%, then Dexterity is probably the worst stat to invest in as an Archer. First of all, DEX appears to have almost no influence on Heavy Shot damage, and has a marginally better return than STR on Light Shot damage. Also, there appears to be some sort of “cap” on Dexterity. After reaching a threshold of 280 Dexterity, adding any more dexterity (up to 325 in my data collection) appeared to add no damage whatsoever to Light Shot or Heavy Shot. However, it should be noted that I experienced a relatively higher accuracy with high Dex gear, so if your focus is on out-of-instance content, then DEX may still be a viable option for stat focus.

Piety

Piety appears to be the most important stat for Archer from my testing. While it has a similar increase in light shot damage as Dexterity, Piety seemed to have, by far, the biggest influence on Heavy Shot. Also, similar to Dexterity, Piety seemed to “cap” at 280. I’m not sure if this is some special number, but it does not appear to be heavily Rank/monster dependent as I tested the cap value on R1 Corroded Coblyns and on the R55 Feretrar and found a similar cap at roughly ~280.

Strength and Attack Power

I’ve read a few posts over STR/ATK recently and there seems to be some controversy over their influence on ARC damage. I’m hesitant to make any definitive statements, but based on my testing of R52 and R55 mobs, I think STR and ATK have a relationship of 2:3 (ATK:STR). So for every 3 points of Strength, you will gain 2 Attack Power. Also, there appears to be a different kind of “cap” for Strength and Attack Power. On R52/55 monsters I never experienced any kind of a “cap” for Strength or Attack Power on Light Shot or Heavy Shot. However, one of my LS mates tested STR/ATK on R1 Coblyns and noted no increased min/max/avg damage on Light Shot. So there appears to be some weird kind of stat checking going on (he suggested perhaps some STR/ATK – VIT check on the enemy).

Vitality

Since Vitality can be equipped to ARC weapons, I decided to test the damage on Light Shot and Heavy Shot. There did not appear to be any increase in min/max/avg.

Discussion/Recommendations

Based off this preliminary testing, I would have to argue that the order of stat preference would be PIE>ATK>STR>DEX (for <280). I would recommend getting PIE to 280 with a double socket body and paragon’s crown, then spend the rest of your resources maxing STR/ATK. I know a lot of people are looking for the “best” ARC gear right now, and this is my best guess as how to outfit yourself with double socketing for instanced content:

Weapon: Ifrit Bow
Head: Paragon Crown
Body: Rainmaker Tunic with 2x PIE (haven’t tested Savage Might, but wasn’t impressed in Instance Fights from eyeballing)
Belt: Doesn’t really matter, but there is a +1 STR belt if you care. I’d rather 2x HP or –Enmity
Hands: Fingerless Raptorskin Gloves of Slaying with 2x ATK Power
Legs: Felt Trousers with 2x STR/DEX material (for economy you can try 2x STR/VIT or +11 ATK pants from GC)
Feet: Cobalt-Plated Caligae with 2x HP (HP is my preference)
Rings: get 280 PIE, then wear STR rings

Summary

Contrary to what most people would expect, Dexterity is, in my opinion, the worst stat for ARC damage wise. Also, there seems to be a “cap” for DEX and PIE at 280; you won’t do any more Light Shot or Heavy Shot damage if you stack higher DEX/PIE. I’m not sure if this is intentional to avoid overpowered material players. I also want to point out that this 280 cap is a rough estimate and I think there is like a 10-20 variation depending on the mob. I still need to test more, but I think 280 is about the right cap for R52-R55 (which is around endgame content rank). I think it’s also important to note that I only tested Light Shot and Heavy Shot. DEX/PIE/STR might affect all the other weapon skills differently.

Future Testing Plans

I’m probably going to work on the rest of the weapon skills to check PIE/DEX/STR influence. I’m not, however, going to test combos and the stat influence on them because honestly that’s too much work for me atm. I’ll also try to test on higher mobs to see how much stat influence diminishes and if the rate of decay is equal across DEX/PIE/STR (probably not equal). After these tests I’ll probably work on LNC or MRD and repeat all my data ; ;

TL;DR

For Light Shot
ATK>PIE>DEX>STR (Up until 272-280 cap)
ATK>STR>>PIE/DEX (Above 272-280 cap)
For Heavy Shot
PIE>ATK>STR>>DEX (Up until 288-290 cap)
ATK>STR>>PIE/DEX (Above 288-290 cap)

Special thanks to Kaeko Leta, Anzu Mazaki, Miko Neversleeps, and Artanis Bluestone for testing and advice on testing.

***

Updated since January 22, 2012

Following a similar methodology I tested the PIE/DEX relationship on the other weaponskills, specifically Piercing Shot, Gloom Arrow, Quick Nock, and Shadowbind. Just like Heavy Shot, these four weaponskills are all PIE based. They are not influenced by the DEX stat at all. Interestingly Heavy Shot and Piercing shot do exactly the same damage and follow the same linear increase with PIE. The same is true for Shadowbind and Gloom Arrow. I think the developers just copy/pasted damage formulas -_-. Also I did more experimenting on the PIE cap for Light Shot and the other WS. My final assessment is that Light shot caps somewhere between 272-280 and all WS cap somewhere between 288-290. I think this might be either completely independent of enemy rank or very slightly altered since roughly the same caps were present on R1, R50, R52, R55, and R65-70 enemies.

Just as a side note, I did a little bit of extrapolation with my data to reach some of my conclusions. I only did ~20-40 shots of any WS and just used the +/-8% damage as a guideline to be within +/- 1-2 of the min/max. It takes a realllly long time to get ~100 of each WS at different stat values. The google doc spreadsheet has been updated as well.


As a final note, I think I’m done testing archer for now. There isn’t much left to test for the class except “confirming” the other WS are PIE and STR/ATK based. That being said, I know the server mergers are coming up and if any testers or players who are interested in stat/class testing want to join with me on Figaro, I’d be more than appreciative. I know I’ve worn my 2 current testing buddies out :3

Seiken Valk
Dancing Mad LS, Figaro

[Seiken]

Thanks for the reply Xianghua. To address your 2 concerns:

1. The reason i did not perform the same # of light shots/heavy shots is because for some tests I was just trying to get an approximation/eyeball the test. For example when I only did 53 Light Shots I was only trying to see if the change in Min/Max was noticeably higher. Having tested over the past 2 weeks, I've noticed that by 50 light shots, you get a pretty good reflection of the Min/Max. You are usually within +/-1. The "average" was not the main number I was concerned with.

2. I actually tested against multiple mob ranks. I'm not sure if you noticed the tabs at the bottom of the google doc, but I tested extensively against both the R52 Zaha'rak Halbedier and the R55 Feretrar. Thanks for the suggestion on fighting mobs with higher Vit/def. I think I'll plan on testing on a R52-55 Gladiator/Marauder type beastmen this weekend (so it should have much higher VIT/def) and on high ranked coblyns to collect more data.

[Abauge_Goga]

Thanks for the work you are doing, at least gives us some better input on stats.

[Meatdawg]

Good Job.

Another bonus to Piety is it should help you land those debuffs like bind/slow/blind etc more often.

I have switched to atk gloves a while ago and never looked back. I could tell from the eyeball test it was just doing better then dex gloves.

For me, the only real question is if 30+ piety on chest is better then 74+ savage might. I would like to see some test with crit damage. (This is a hard test to do solo because a lot of times in ifirt or moogle fights you can position yourself behind the enemy which greatly increases your crits)

[Spider-Dan]

It seems like crit rate (against things that matter) is too low to make +critpower worthwhile. Back when Blindside forced crits on multi-hit WS, it would have been worth it, but now, it just seems like too much trouble to go through just for Wide Volley or Bloodletter.

[Estellios]

Regarding PIE vs Savage might, I feel that I'd rather have a much more reliable Bloodletter/Shadowbind than deal more damage. It would probably help out groups more than extra damage would, anyways.

[Kaeko]

These tests were pretty well controlled in my book. Couple of points to preemptively address common complaints regarding the methodology of his tests. I know most people don't care about anything but the 'result' so you can ignore this post if that's you.

(1) "The number of trials is not even"
If you were to shoot 1000 shots at something, you would quickly realize that the as far as the game is concerned, it's more about the MIN and the MAX damage. You can prove this to yourself by writing down every single shot and then graphing the distribution of damage numbers. You'll quickly find that the distribution follows an "even" or "random" distribution where all values between the MIN and MAX occur at equal frequency. This goes against what most testers default into thinking, which is that everything follows a normal distribution. Because damage follows an even distribution (if you want to argue this point, we can talk more about it), all you actually need to get an accurate representation is enough attempts to get a proper MIN and MAX value. You can really do this in a sample size of 2 if you're lucky and judge accuracy of your current MIN and MAX using variance (which Seiken does). Following the AVE damage is not necessary and likely less accurate.

(2) "The number of trials is too small"
This goes back to point 1. Because the key to testing is the MIN and MAX, the trial number does not matter as long as you have a proper range. You can judge the accuracy of the range by looking at the "variance". For those that do this research enough, you'll find the standard is roughly the MIN x 1.16 or 1.17 to find the proper MAX if both are correct. The trial size is only needed to achieve a proper range and nothing more. Large sample sizes are actually not required, especially on lower damage moves.

(3) "Only tested on 1 mob"
He tested on R55, R52 x2, R50, R35, and R1. You have to use the tabs in the excel file.

[Spider-Dan]

The only real "objection" I would make is that the conclusion seems to lean pretty heavily on PIE over DEX based solely on Heavy Shot, which may not necessarily scale well as a general recommendation; if WSes do have XI-style secondary mods in this game, you'd need to use more discretion in your stat build.

As a point of reference, if a similar test were performed for XI THF and the one dagger WS you happened to pick was Shadowstitch, you would come to the conclusion that +CHR is very important; if the one dagger WS you picked was Viper Bite, you'd come to the conclusion that +ATK isn't very good.

Of course, the argument could be made that Heavy Shot is simply more important than other WSes in all situations (not just Moogle), and I'm certainly open to that. I just want to avoid projection of a "global hierarchy of ARC stats" before we find out the impact of stats on the other WSes. It certainly seems counterintuitive for SE to tout DEX for ARC and then make +DEX the least effective bonus to add.

[Estellios]

Yeah I was also kinda curious as to why you were using Heavy Shot, but I guess it's because it has the lowest recast time?

[Kaeko]

The only real "objection" I would make is that the conclusion seems to lean pretty heavily on PIE over DEX based solely on Heavy Shot, which may not necessarily scale well as a general recommendation; if WSes do have XI-style secondary mods in this game, you'd need to use more discretion in your stat build.

As a point of reference, if a similar test were performed for XI THF and the one dagger WS you happened to pick was Shadowstitch, you would come to the conclusion that +CHR is very important; if the one dagger WS you picked was Viper Bite, you'd come to the conclusion that +ATK isn't very good.

Of course, the argument could be made that Heavy Shot is simply more important than other WSes in all situations (not just Moogle), and I'm certainly open to that. I just want to avoid projection of a "global hierarchy of ARC stats" before we find out the impact of stats on the other WSes. It certainly seems counterintuitive for SE to tout DEX for ARC and then make +DEX the least effective bonus to add.

I agree with the idea that you cannot test 1 WS's modifiers and then extrapolate it globally to all WSs for that job. It certainly is an convenience to use the lowest recast WS though.