Test Resuluts On Stats

[Fiddle Sticks]

Seen people were discussing the effects of stats, here is some rather good testing with parse results.

Click here for the Forum is was posted on and for links for more in depth info and parse results.

http://ffxiv.zam.com/forum.html?foru...7176564&page=1

. POST UPDATES
II. MAJOR FFXIV PATCHES
III. CURRENT PROJECTS
IV. INTRODUCTION
V. EMPIRICAL TESTING GUIDELINES
VI. GENERAL POINT MECHANICS
VII. STATUS POINTS
VIII. ELEMENTAL POINTS
IX. DAMAGE FORMULA THEORY
X. ADVANCED DAMAGE FORMULA
XI. CRAFTING LOG ANALYSIS


=========== SECTION I: POST UPDATES ================================================== =

5/12 TEST RESULTS: Effect of Magic Potency and MND on Cure Potency by Ami Trubbles of Besaid and Lifestream LS.

4/29 TEST RESULTS: Log Parse Results! Effect of STR and DEX on Damage and Accuracy

4/25 TEST RESULTS: Log Parse Results! Effect of New NM Gear Stat Bonus on 30MRD

3/25 FINALIZED TEST RESULTS: Fully analyzed data and put in final recommendations for Stat Point Allocation and Appeal to Dev Team. This test case is now 100% complete!
* The Effects of INT PTY RANK and ELEMENTAL POINTS on Scourge Scourge II Shadowsear

3/24 FINALIZED TEST RESULTS: Fully analyzed the data and added several new items on the spreadsheet to illustrate the effects of the variables. This is the methodology I will be using in my tests from now on. This test case is now 100% complete!
* The Effects of INT PTY and RANK on Phantom Dart and Damnation

3/21 MAJOR POST UPDATE
*Massive content updates to every category in the entire post, added new categories
*Major changes to post format, organization, etc.
*FFXIVCore Exclusive Naoki Yoshida Interview, hints major damage calculation changes are underway in Upcoming Patches

2/7 NEW TEST RESULTS: My First Parsed Test Results:
* The Effects of INT PTY and RANK on Phantom Dart and Damnation


=========== SECTION II: MAJOR FFXIV PATCHES ================================================== =

Appeal to Naoki Yoshida and the Dev Team
Please produce a combat system where increase of player stats will directly impact damage output, and lower misses and resist rates. This is the only way gear with stat bonuses will have any impact. Even after the November patch where spells became usable, spells still give inferior SP/hr over using basic attacks, and are not at all affected by increasing your stats.

Further, results like I observed in both tests, where the damage was dramatically reduced going from the R58 to R59 mob show inconsistent formulas are being applied. Consider smooth curve functions, in this case, when mob rank changes. Dramatic inconsistent differences with just a small variable change will only confuse players.

Get rid of physical level, it will diminish any value from having gear with stat bonuses, in addition to providing the players too many variables for things they don't need to change, which just confuses them. Too many custom variables also keeps you from having a balanced system. Fix stats with respect to each class and rank, just like it works with Mobs and in FFXI. FFXI worked. If it wasn't broke, don't fix it.


UPCOMING PATCHES
In a recent interview with Eorzeapedia, thanks to these test results (and thousands of complaints), the developers have finally understood the need to rewrite all the game's damage calculations. They cited that they want the calculations to be more transparent and logical, citing that essentially accuracy is broken with respect to DEX and PTY.
FFXIVCore Exclusive Naoki Yoshida Interview
Naoki Yoshida: "They want to make [the Calculations] simpler, not just so the players know what's going on--having to have some super program going on in the background to understand the calculations--but also so when the devs do balancing, it's easier for them to balance."

Also see:
FFXIV ZAM's Exclusive Interview with Naoki Yoshida & Sage Sundi

03/03/2011 PATCH 1.16
Lodestone Notes
Changed the % that durability of an item affects its output from 50% to 35%

02/02/2011 PATCH 1.15a
Lodestone Notes
Major changes increasing the amount of SP for all monsters in Eorzea

11/25/2010 PATCH
Lodestone Notes
Notably "Physical and magic accuracy calculations have been adjusted. Attacks are now easier to land and more difficult to resist." The effect of VIT on HP gain and MND and MP gain was drastically reduced. After logging in, many characters had an extreme reduction in maximum HP and MP. Most extreme, the action-based SP system was scrapped so SP gain works just like Physical Level EXP gain. The good news, now it is actually beneficial to kill a mob quickly and increase your defensive and offensive capabilities and value as an intelligent player. However, beware of future patches which may invalidate old testing and renew the need for more testing.


=========== SECTION III: MY CURRENT PROJECTS STATUS ================================================== =

**3/21: It has also been suggested by the second Official FFXIV Players Poll, and the interviews linked above, that the entire status point allocation system will be scrapped, eliminating Physical Level and benefits, rescaling base status points to be fixed to each Class and Rank. This will require an entire overhaul to the damage calculation system, and will be done in phases. After the main update, the values will need to be tested and tweaked by the Dev team.

Any further testing of mine related to the battle system is on hold until this is implemented. I am interested in testing certain effects and calculations related to crafting, see section below. As usual, if you have some thoughts or requests, let me know by replying to this post.


=========== SECTION IV: INTRODUCTION ================================================== =

PURPOSE
To provide a concise, global picture of FFXIV battle system mechanics and status allocation. To cut down on chatter around badly executed tests, emphasizing true empirical testing. To set baseline requirements for testing to include automatic log capture and parsing scripts. New theories, tests, and results will be documented here. Existing test posts in forums will be linked.

AUDIENCE
Advanced players who are interested in damage optimization, who can contribute theories and/or empirical test results. Read Section II for details how to automate your testing. It is VERY EASY to test using Crabby as an external log parser. No more pen and paper or Excel spreadsheets!

CONTRIBUTION
If you would like to contribute by performing your own test, capturing your logs as explained in this thread (submit Crabby's *RAW* data), please PM me and I would be happy to analyze your data similar to the results below.

You would just need to define the constants and variables as shown in the first page of the spreadsheet in the following thread:
The Effects of INT PTY RANK and ELEMENTAL POINTS on Scourge Scourge II Shadowsear

I can tweak your output and/or modify my script to make sure it only captures data conforming to the test. Be sure to not use Buffs or Job Traits for maximum consistency, even though I can make a rule in the logs to discard data, much like I did with the coblyn's Cluster Geyser (broken mob throws off damage values). Just conform to the experiment definitions format and maintain those constants, and its a cinch. Be aware you will want to be able to repair your weapon of choice once it hits 35%.

WORK IN PROGRESS
This post is a growing framework, that identifies what needs testing, and where it fits in the big picture. The format may change as test data is verified and integrated, and game updates are made. Please help out by posting links to any related testing results.

Enjoy and I look forward to your feedback and your own experience testing game mechanics.
With collaboration, we can get somewhere.


=========== SECTION V: EMPIRICAL TESTING GUIDELINES ================================================== =

1) Maintain the same rank. Testing at Level 50 physical and class rank is preferred. Most MMO balancing mechanics are scaled to the maximum level.
2) Fight the exact same monster and rank. Most mobs vary 5 levels in the same spawn area. Green mobs are the best target equal to your level. Physical experience will always be the same on the same rank mob. (Edit: Mob Rank variance is no longer an issue because my Perl script automatically sorts out data specific to each mob rank based on EXP awarded)
3) Log every single hit, miss, resist, and critical value. Resisted values should be averaged separately. Calculate miss rate, resist rate, average non-resisted critical damage, average non-resisted non-critical damage. (Edit: my perl script now automates this. See one of my tests for an example)
4) Keep your equipment durability topped off. Assume that durability has impact before the item turns yellow. (Edit: Recent FFXIV updates reduced the % that durability affects output from 50% to 35%)
5) If a mob part breaks you will do more damage with subsequent hits. Those hits should be removed from averages. (Edit: my script automatically ignores values after Cluster Geyser is used until the mob dies)
6) Ensure only a single variable exists, and that all other factors are constant. Consider all equipped gear, status allocation, mob type and level, durability, etc. Use the experiment definitions in my own tests as a guide.

Automation
Crabby Thread, by golfandsurf
DOWNLOAD CRABBY HERE
**NOTE THIS IS A THIRD PARTY PROGRAM. USE AT YOUR OWN RISK. I DO NOT CONDONE THE USE OF THIRD PARTY PROGRAMS TO GAIN AN UNFAIR ADVANTAGE OVER OTHER PLAYERS**
1) Download Crabby.rar from the link above and extract the EXE to your own directory.
2) Start FFXIV and log in using windowed mode.
3) Run Crabby.exe. Check if it has synced up by looking at the bottom of the window. It will say "Attached". If you need to resync, just double click the icon at the lower left of the window.
4) You can create separate window feeds for each chat type. Click "Settings > Chat Tabs" and select one or more types. They correspond to what you "Show" or "Hide" In-Game under "Configuration > Chat Log > Settings". Best of all, there is no limit to the history or number of log entries Crabby remembers.
5) Execute your test with the guidelines above, and copy out the timestamped entries into an external text editor. Please be aware that logging out of the game may crash Crabby and prevent you from copying out the logs. Be on the lookout for bugs and report in the Crabby Thread. Try a more stable version of Crabby if you experience problems. The newer built-in parsing features are not necessary for our testing purposes.
6) Parse these text files to determine your test conclusions. I have completed customized perl scripts which automatically evaluate your average non-critical damage, average critical damage, etc.

****3/21 Update: I have requested the application developer put out a "lite" release that just lets us grab the logs, and may reduce the amount of crashing due to his 'extra' parsing features. A new version of crabby was required after the Feb update, because of the inherent changes in the logs. The app developer is putting out release fixes very timely.

****If you would like to send me log data for any test, I am willing to parse your test with my Perl scripts. Please message me through the forum. It must be in Crabby's "RAW" format to include the chat codes. Please take out any personal information such as linkshell chat.


=========== SECTION VI: GENERAL POINT MECHANICS ================================================== =

STAT POINT REASSIGN
At Physical Level 50, it will take exactly 3 hours for Reassign to regenerate after use. Tested late January 2011.

STAT POINT CAP
The hidden 'cap', or the point where you no longer receive benefit from high stats based solely on your current rank.
** The Effects of INT PTY and RANK on Phantom Dart and Damnation
My own tests in this thread conclude that the proposed hidden soft cap either does not exist for INT, or exists at values higher than what was tested.
**"Stat Caps Per Rank [WIP]" by Beregren. Last updated 1/22/2011
This thread tracks each rank where adding VIT stat no longer increases your HP. It does not claim directly, but proposes through inference that all other stats might cap similarly.
STILL NEEDS TESTING

STAT POINTS ON GEAR
TEST RESULTS: Log Parse Results! Effect of New NM Gear Stat Bonus on 30MRD
This test showed a correlation where a decrease in "Attack AND STR" at the same time, resulted in a decrease in average damage and average critical damage.
Discussion: There is a distinction between adding stats in your stat allocation and stats in your equipped gear. The difference is clear that you do not receive HP from adding VIT gear, but unclear how else equipment stats are ignored. I.E. If you are under the suggested cap and add +20 STR to your gear, do you do more average damage? Adding status points via gear may or may not actually contribute to defensive or offensive capabilities.

SP GRINDING "FATIGUE"
Your log will tell you you have received "bonus skill points" when you are under fatigue. When fatigue starts or gets more severe, your log will tell you your "skill bonus rank has increased." Each step will reduce your skill point gain -10%.

It's easiest to think of fatigue as a slider that slowly resets to origin, and slowly increases based on the AMOUNT of SP you earn, and the maximum time the slider will take to fully reset is 1 week.

**3/21 Update: I've noticed while getting a consistent 50-60k SP per leve reset that it is enough to maintain a 1x level of fatigue (-10%) through the whole week, where at the beginning and middle of the week there is a short period of time where there is no fatigue. I never hit 2x fatigue. Since there is 4.6666 leve resets per week, this means earning roughly 234-280K SP per week will keep you safely within 1x fatigue. The amount of SP needed to hit fatigue from the origin may vary depending on your rank. Also some perspective, if you zerg any job from Rank 1 to 18 in a day, you will not hit 1x fatigue until you hit Rank 18. I tested this on 5 jobs 1 to 18.

"FFXIV Tactics: Understanding the Fatigue System" VIDEO


=========== SECTION VII: STATUS POINTS ================================================== =

A Final Recommendation for stat balancing for SP grind to accommodate DoW AND DoM jobs
STR: Max it as high as possible to affect basic attack damage. May cap out effect sooner for weapon skills (untested)
VIT: Maximize after you are happy with your other parameters. Affects Max HP and may also contribute to reducing raw damage.
DEX: In large volume, DEX will result in a decrease miss rate.
INT: Max it as high as possible to affect basic wand attack damage. Will cap out or not be relevant for weapon skills and nukes.
MND: Don't increase it past 40. You shouldn't ever have issues getting MP with Siphon MP. Also affects Cure potency.
PTY: Minimize, it's utterly useless. May still affect enfeeble accuracy and duration, but this is untested and would only matter for NMs.

STR
Log Parse Results! Effect of STR and DEX on Damage and Accuracy by myself, Deus Bahamut.
STR behaves much like INT such that an increase in STR will result in a steady and consistent increase in damage.

INT
The Effects of INT PTY and RANK on Phantom Dart and Damnation by myself, Deus Bahamut.
The Effects of INT PTY RANK and ELEMENTAL POINTS on Scourge Scourge II Shadowsear by myself, Deus Bahamut.
**See each thread for my conclusions.
**Increasing INT values does steadily increase Phantom Dart average damage. However may not affect Weapon Skill or Nuke (Scourge/Scourge II) damage. Either the cap is met at a lower value, or INT does not affect Weapon Skill or Spell damage. Phantom Dart does not yet appear to cap at all. Stick to using Phantom Dart on SP grind mobs with a maximized INT and minimized PTY.

DEX
Log Parse Results! Effect of STR and DEX on Damage and Accuracy by myself, Deus Bahamut.
Increasing DEX will result in a decreased Miss Rate. This test shows a correlation between increasing DEX by 120 points resulted in a consistent Miss Rate decrease in Light Swing, Skull Sunder II, and Trunksplitter II.

PTY
The Effects of INT PTY and RANK on Phantom Dart and Damnation by myself, Deus Bahamut.
The Effects of INT PTY RANK and ELEMENTAL POINTS on Scourge Scourge II Shadowsear by myself, Deus Bahamut.
**See each thread for my conclusions.
**Increasing PTY has ABSOLUTELY NO EFFECT OR CORRELATION TO INCREASING ACCURACY OR LOWERING RESIST RATES for Phantom Dart, Damnation, Scourge/II, Shadowsear
Stick to using Phantom Dart on SP grind mobs with a maximized INT and minimized PTY.

VIT
"Stat Caps Per Rank [WIP]" by Beregren. Last updated 1/22/2011
**NOTE: The following test was completed before the 11/25/2010 Patch.
This thread tracks each rank where adding VIT stat no longer increases your HP.
VIT may also reduce the raw damage you take, in addition to increasing your HP value.

MND
Should behave similar to VIT. Keep your MND at 40 unless youre increasing cure potency. You only need a small amount with abilities like Siphon MP.

Effect of Magic Potency and MND on Cure Potency by Ami Trubbles of Besaid and Lifestream LS.
Ami was gracious enough to do this test and give me permission to post her results here. You will see in the following spreadsheet data for Cure II HP values at variable MND and variable Magic Potency. From these values she was able to derive the formula to conclude:
Conclusion: Raising MND+2 will increase Cure II Potency by +1.
Conclusion: Raising Magic Potency+2 will increase Cure II Potency by +1.
Conclusion: Cure II base cure amount is 312.92


=========== SECTION VIII: ELEMENTAL POINTS ================================================== =

ELEMENTAL DAMAGE
How greatly does increasing elemental affinity affect elemental damage?
STILL NEEDS TESTING
How greatly does increasing elemental affinity affect rate of resistance?
STILL NEEDS TESTING

SCOURGE AND BANISH DAMAGE
Is there a correlation between the increasing elemental affinity and light and dark damage?
Water, Ice, Earth = SCOURGE
Fire, Wind, Lighting = BANISH
STILL NEEDS TESTING

ELEMENTAL BUFFS
Is there a correlation between buff power and elemental affinity?
STILL NEEDS TESTING
**An easy empirical test would be to test Stoneskin I+II power with lowest and highest earth against a Cactuar's 1000
needles.
Determine Elemental affinity:
Stoneskin = Earth?
etc

ELEMENTAL DEBUFFS
Is there a correlation between debuff resistance and duration and elemental affinity?
STILL NEEDS TESTING
**Logging/parsing is nearly required in order to use time stamps to determine how long the debuffs last. This will make for some interesting Perl script enhancements.
Determine Elemental affinity:
Sleep = Water?
etc


=========== SECTION IX: DAMAGE FORMULA THEORY ================================================== =

NOTE
As stated above, fansite interviews with FFXIV developers indicate the damage formulas are broken and need to be completely revamped.

DEFINITIONS
PC = Player Character
MOB = Monster
RANK = Current rank
STR/VIT = As reported on Attribute screen
STR_BONUS/VIT_BONUS = Equipment attribute bonus total
Attack/Defense = Total Physical Attack/Defense rating on your character status screen
PC_ATK_OBJ/PC_DEF_OBJ = Object rating. Physical Attack/Defense rating on PC's Primary Weapon or PC's Armor item description tab.
MOB_ATK_OBJ/MOB_DEF_OBJ= Object rating. Hidden rating on 'MOB's Claw' or targeted 'Body Part'.
DUR = Object Durability. OBJ value is scaled by DUR based on the object's condition' or 'durability %'

FORMULA
The formula is merely demonstrating a basic relationship between combat variables. The structure of the formula is loosely organized around the notion that the core damage is decided by subtracting the STR and VIT stats (Group B), scaled by the ratio of Attack and Defense values (Group C).

A,B,C,D,X = unknown constant values (may even be zero). Also denotes groups referred to below.

A*( MOB_Level / PC_Level ) *
B*( MOB_STR - (PC_VIT + X*PC_VIT_BONUS) ) *
C*( MOB_Attack / PC_Defense ) *
D*( MOB_ATK_OBJ*MOB_DUR - PC_DEF_OBJ*PC_DUR ) *
= Physical damage taken by PC

A*( PC_Level / MOB_Level ) *
B*( (PC_STR + X*PC_STR_BONUS) - MOB_VIT ) *
C*( PC_Attack / MOB_Defense ) *
D*( PC_ATK_OBJ*PC_DUR - MOB_DEF_OBJ*MOB_DUR ) *
= Physical damage taken by MOB

Group A:
This value is significant to soloing in FFXIV as opposed to FFXI. Damage seems to drastically reduce after a mere 2-3 levels whether it is Level 15-18 or Level 37-40 (i.e. MOB turns from Red to Green). The reduction rate appears to be similar in both cases. In FFXI, your level did not seem to directly influence damage much at all, as you could get killed by easy prey at a high level, while at low level you could kill mobs several levels above you in 2 or 3 hits.
**See my test results for INT and PTY to infer the effect (DELTA)Player/MobRank has on output.

Group B:
This value carries the most weight, and can be heavily scaled by Group C. Against mobs equal or greater to your level, increasing STR will scale your damage better than increasing Attack. Inference from FFXI.
STILL NEEDS TESTING

Group C:
Can heavily scale Group B. Against mobs lower than your level, increasing Attack will scale your damage better than increasing your STR. Inference from FFXI.
STILL NEEDS TESTING


=========== SECTION X: ADVANCED DAMAGE FORMULA THEORY ===============================================
GROUP D: ITEM VALUE, TARGETING, AND DURABILITY FACTORS

INNATE ITEM ATTACK/DEFENSE VALUE AS AN INDEPENDENT FACTOR
Group D operates under the inference that the innate item Attack/Defense characteristic is an element in the equation separate from the character's total Attack or Defense value. In other words, the a weapons' Attack rating may contribute to the end damage even more than the character's total Attack. In FFXI, this group translates to the Equipment's DMG and DEF values on the item description. However, in FFXIV, the item description is not labeled DMG or DEF, but named the same as the character's total Attack or Defense rating. The weapon may contribute to the overall Character Attack AND as an additional element in the equation.

I suspect Group D is more closely related mathematically to the core damage factor, Group B, than I've indicated. It may be more than just a scalar on the overall damage. I separate it to Group D for clarity. In FFXI, DMG and DEF on the item seemed to be even more influential than Group B or C. The bottom line is, items with higher DMG were sought over those that offer more Attack.
STILL NEEDS TESTING

DIRECTION AND BODY TARGETING
It's clear from the logs what part of the mob you are facing when attacking is noticed by the battle system.
Does it increase damage or critical rate?
Does it make it more likely to target certain body parts?
**3/21 Update: I have confirmed from fighting various open-world NMs that the direction you are facing the mob directly impacts which mob part you are likely to hit. For example, standing in front of Uraeus and using Flashfreeze II will sometimes result in breaking his eyes. Standing in front of Tarbh Usige's legs will make him fall if Incapacitation is triggered. Breaking his horn purportedly increases the drop rate of his horn.

DURABILITY
Group D is tied to item durability; a broken weapon of 0% durability will result in a damage penalty.
Conversely, when an armor part breaks, if it is hit again, it will result in increased damage (confirmed).
Durability only affects Attack or Defense based on the %, NOT based on the status color change of the bar indicator, or the status color change of the equipment box on the Gear screen. The logs may prematurely indicate the item is damaged, when it has not yet crossed the 35% threshold.
STILL NEEDS TESTING

SPECIFIC TO PC BODY ARMOR
When the monster attacks you, it's probable specific gear is targeted each hit. The damage may just be evaluated on the whole, or may also be independently influenced by that targeted armor's Defense value. The mob might even target more than 1 armor parts per hit.
STILL NEEDS TESTING
**3/21 Update: Without numerical durability indicator (pre-November 2010 patch), we can only go by %, which isn't as telling as to the rate durability decreases between gears, so this would be hard to test. Since 99% of the time players face the mob, I'm betting its pretty randomized which areas the mob hits. I'm sure it differs based on each mob attack.

SPECIFIC TO MOB BODY PARTS
This relates to Group D. Many weapon skills target a single specific body part, but it seems any attack will target at least one body part per strike, as mobs can be incapacitated without using weapon skills. There could be a separate HP counter assigned to each part, such as in "Valkyrie Profile 2" (PS2). The probability of which part is hit may depend on several factors. Breaking a mob part will result in increased damage to that part with subsequent hits.
**3/21 Update: A hidden HP bar seems unlikely, because only be executing the appropriate weapon skill in the mob incapacitated. However, there still might be a hidden HP value that must ALSO be met for incapacitation. This theory may be disproven by incapacitating on the first hit of battle.


=========== SECTION XI: CRAFTING LOG ANALYSIS ================================================== =

MEASUREMENT
Full automated log measurement is not straightforward primarily because the COLOR of the orb is only reflected visually, not in the logs. This may have been done to combat crafting bots, although if they are sniffing packets the data is still probably intercept-able.

Back on topic, this means that during a crafting test, crafts should only be done on WHITE orbs or only on YELLOW orbs, etc. when observing the success of the craft. The PROGRESS, QUALITY, and DURABILITY can be directly recorded with no issues. Hidden attributes seem to be a RISK value, which increases the probability of sparking or element destabilization. Sparking is rare and usually from doing too many bold synthesis, but sparking does not appear in the logs, but only visually. Element destabilization does appear in the logs, so this can be measured.

So assume a scenario where an item is being crafted equal to your rank (for a more telling distribution of passes/failures). You would exclusively do a standard synthesis ONLY when the orb is white. If the orb is not white, you can WAIT or selectively use abilities to turn it white (Preserve) but NOT abilities that MIGHT affect SUCCESS or RISK rates. Until these abilities are tested they might just throw off the results.

The cool thing is it doesn't matter if you fail the synthesis, each individual 'standard synthesis' attempt is all that matters. If you can optimize each individual synthesis attempt, imagine the impact that will have on your overall item success rate. Only by parsing will you spot the most obvious trends.

HELP
I'd really appreciate some assistance testing in this fashion, giving me your logs to analyze. I am currently stuck on 2 crafting professions doing high risk expensive synthesis that I can't afford to exclusively craft on white orbs just yet. I've come to expect nobody will help me do anything though.

EFFECTS OF CRAFTSMANSHIP VS MAGIC CRAFTSMANSHIP
From my research online and personal experience, this is still a great unknown. It is unknown which items and trends apply as to which specific recipes have higher success rate with one stat higher than the other.
STILL NEEDS TESTING

EFFECTS OF CONTROL
The hidden risk value seems to effect element destabilization and aether sparking.
STILL NEEDS TESTING

EFFECT OF DESTABILIZATION
The presence of destabilization seems to dramatically reduce standard synthesis effects (increased durability loss and reduced progress). This, however, needs to be more accurately measured.
STILL NEEDS TESTING

EFFECT OF AETHER SPARKING
I'm guessing the sparks have a similar effect as destabilization.
STILL NEEDS TESTING

[Celeras]

This is all going to change in like 3 weeks.

[Drako]

PTY increases spell DoT dmg.

base + 1 dmg / 10 PTY i think.

[Fiddle Sticks]

This is all going to change in like 3 weeks.

More than likely, but seen people discussing it so I thought I would post.

[Hyan]

Here's a test resulut on HQ/quality ratio:

http://static2.finalfantasyxiv.com/a...KzM1CS-naU.png

[Kaeko]

Great find Hyan. Do you happen to know the lodestone blog this is from? looking for more info about how the chart was made / data collected.

[Celeras]

Its pretty self explanatory. For those who can't read it, the header across the top is simply Quality>HQ rate %>total attempts>HQs. TLDR is get above 300.

Dunno how relevant this is to the thread though since Quality isn't a stat.

[Kaeko]

It's not about it being self-explanatory its how the data was collected that interests me. Depending how they did it the chart could have too many holes in it to be useful.

[Dooom]

It's the 350-399 results which interest me, compared to the rest of the 300+ ones.

[Celeras]

It's not about it being self-explanatory its how the data was collected that interests me. Depending how they did it the chart could have too many holes in it to be useful.

How could the data in any way have holes? Total crafts, crafts that HQ'd. The only way possible would be if they used different recipes, and someone that stupid wouldnt bother parsing over 3000 synths and making a graph.

It's the 350-399 results which interest me, compared to the rest of the 300+ ones.

A 2% fluctutation in a sample size under 100 hits isn't really anything eye-popping. The trends pretty obvious overall.

[Kaeko]

There are always confounding variables which you have to at least acknowledge. For a video game I think the vast majority of players would just take the graph and apply it without much thought, but maybe I'm just too big a stickler for methods. Various things you could argue for just off the top of my head:

(1) moon phase
(2) level of synth
(3) level of crafter,
(4) dLVL between synth and crafter
(5) are there any sub-crafts involved in the synth (and you can apply the former questions to this as well)
(6) are all these trials with the same synth recipe (because the former questions will vary if it isnt)
(7) does he have all the raw data available so you can do a simple standard deviation - can get confidence intervals this way
(8) touch-ups?
(9) stat distribution?
(10) control?

I'm not saying the graph isn't very useful, I just asked for a source if Hyan had one. I think making a comment that "how could this possibly have any holes" is going a bit far though. Plus, it's always possible to find even more relevant results not specified in the graph if there's something else to go along with it, so it's not just about poking holes, but also about gathering potentially more information.

[Celeras]

1 would be the only actual variable I guess, if you want to go ahead and call moon phase a concern (lol). Everything else goes away if you assume its a R50 doing the same synth over and over, which I didn't think was a stretch. Touchups wouldnt matter at all as you could just count every one as its own synth at the same quality.. since all it is is a re-roll.

Not suggesting a source is irrelevant, but theres not much to screw up here.

[Hyan]

I'll look it up, but keep in mind it's in JP.

Edit: there's tons of graphs in there, dunno what they relate to, but the crafting post should be the 2nd most recent one.

http://lodestone.finalfantasyxiv.com...4558&tag=11552

[Spider-Dan]

1 would be the only actual variable I guess, if you want to go ahead and call moon phase a concern (lol). Everything else goes away if you assume its a R50 doing the same synth over and over, which I didn't think was a stretch. Touchups wouldnt matter at all as you could just count every one as its own synth at the same quality.. since all it is is a re-roll.

Not suggesting a source is irrelevant, but theres not much to screw up here.

Even if you are R50, it's silly to simply dismiss dLVL between crafter and synth.

Unless you are presuming that 300 quality at dLVL=+2 is the same as 300 quality at dLVL=+42?

[FireCrystal]

There are always confounding variables which you have to at least acknowledge. For a video game I think the vast majority of players would just take the graph and apply it without much thought, but maybe I'm just too big a stickler for methods. Various things you could argue for just off the top of my head:

(1) moon phase
(2) level of synth
(3) level of crafter,
(4) dLVL between synth and crafter
(5) are there any sub-crafts involved in the synth (and you can apply the former questions to this as well)
(6) are all these trials with the same synth recipe (because the former questions will vary if it isnt)
(7) does he have all the raw data available so you can do a simple standard deviation - can get confidence intervals this way
(8) touch-ups?
(9) stat distribution?
(10) control?

I'm not saying the graph isn't very useful, I just asked for a source if Hyan had one. I think making a comment that "how could this possibly have any holes" is going a bit far though. Plus, it's always possible to find even more relevant results not specified in the graph if there's something else to go along with it, so it's not just about poking holes, but also about gathering potentially more information.

From same blog:
https://spreadsheets.google.com/spre...hl=en_US#gid=0

Moon phase seems irrelevant..

[Celeras]

Even if you are R50, it's silly to simply dismiss dLVL between crafter and synth.

Unless you are presuming that 300 quality at dLVL=+2 is the same as 300 quality at dLVL=+42?

Only if youre trying to pinpoint the actual percent chance, rather than the sweet spot for quality. 300+ would be that spot regardless of what level the synth is, which is all I care about. You're talking about whether or not its a 20% or 50% chance based on what leve/etc the synth is.. which I find to be largely irrelevant.

[Spider-Dan]

Only if youre trying to pinpoint the actual percent chance, rather than the sweet spot for quality. 300+ would be that spot regardless of what level the synth is, which is all I care about. You're talking about whether or not its a 20% or 50% chance based on what leve/etc the synth is.. which I find to be largely irrelevant.

How can you possibly know that the quality "sweet spot" is the same across all dLVLs, without any data?

[Jamison]

How can you possibly know that the quality "sweet spot" is the same across all dLVLs, without any data?

You can't, but it is possible dLVL isn't a specific factor in the HQ formula; but it still shows itself because with a small dLVL it is significantly harder to get more quality.

[Celeras]

How can you possibly know that the quality "sweet spot" is the same across all dLVLs, without any data?

What are you expecting? Getting 300 quality shows a huge jump in this trial. Now up the dLVL of the synth. You expect not only the HQ % to be lower... but for the benefit of this threshold to also decrease NON-proportionately? That doesn't make any sense logically or from a design standpoint. SE would have to be.....

[Spider-Dan]

If your point is that logic is a substitute for data, particularly in an FF Online game, that is your prerogative.

I find it perplexing that anyone would seriously argue, "well, it just wouldn't make any sense for this to work differently" in a game where the only relevant stat is virtually always dLVL, but that's me.