Forum Settings
       
This Forum is Read Only

Current Parse ResultsFollow

#1 Apr 11 2011 at 1:04 PM Rating: Good
**
924 posts
Ok, so I'm almost complete making a version of kparser for FFXIV, and I logged 7 hours of gameplay, mostly on NM's (uraeus, great buffalo and dodore).

Here's my results:

 
Name	Attacks	Hit	Miss	Acc	Crit	Crit%	Damage	Reg Dam	Avg Reg	Crit Dam	Avg Crit	Crit Ratio 
VG	1512	1326	186	87.7%	269	20.29%	72634	58419	44	14215			53	19.57% 
SC	1266	1050	216	82.94%	210	20%	57060	46129	44	10931			52	19.16% 
SR	1949	1704	245	87.43%	254	14.91%	49542	42345	25	7197			28	14.53% 
AO	1684	1454	230	86.34%	197	13.55%	48115	42475	29	5640			29	11.72% 
NN	843	721	122	85.53%	140	19.42%	43953	36306	50	7647			55	17.4% 
MH	1132	1072	60	94.7%	123	11.47%	42581	37591	35	4990			41	11.72% 
DP	244	218	26	89.34%	67	30.73%	40214	26415	121	13799			206	34.31% 
MC	1260	1115	145	88.49%	274	24.57%	29963	23529	21	6434			23	21.47% 
HT	418	372	46	89%	84	22.58%	27567	21008	56	6559			78	23.79% 
SA	437	395	42	90.39%	40	10.13%	19227	17132	43	2095			52	10.9% 
JB	580	514	66	88.62%	52	10.12%	17586	15966	31	1620			31	9.21% 
SK	250	219	31	87.6%	7	3.2%	14673	14072	64	601			86	4.1% 
TB	66	45	21	68.18%	22	48.89%	177	85	2	92			4	51.98% 


Now VB has 100 strength. SC has 174 strength. Both have equal weapon stats. Both are level 50. On NM's why would more strength not equal more damage/hit? At least on average.

EDIT: (updated mob HP based on how many we fought, had miscounted in log)
Goblin (NM Only) HP: 36k~
Goblin (Brags) HP: 35k (for the group of them)
Uraeus HP: 40k~
Great Buffalo HP: 42k~

These are the evasion/damage taken results: (again these are all over 7 hours on 3 uraeus, 1 great buffalo, and 2 gobs):

Uraeus x3:	Damage Taken	# of Attacks	Average Dam/Hit	Evades	Evasion %	STR	VIT	DEX 
AO		34185		31		1102		0	0%		84	84	84 
DP		1860		1		1860		0	0%		84	84	84 
DA		3265		3		1088		0	0%		68	83	60 
HT		4485		5		897		0	0%		120	110	64 
JB		23582		22		1071		0	0%		110	120	110 
MH		60678		122		497		4	3.28%		114	158	85 
MC		17165		19		903		1	5.26%		110	135	115 
NN		26954		25		1078		0	0%		130	118	109 
SC		19216		22		873		0	0%		174	131	26 
SR		45163		33		1368		0	0%		108	108	108 
SK		46265		196		236		1	0.51%		122	174	46 
VG		23183		24		965		0	0%		102	101	101 
								 
Buffalo:	Damage Taken	# of Attacks	Average Dam/Hit	Evades	Evasion %	STR	VIT	DEX 
AO		17508		22		795		0	0%		84	84	84 
DA		781		1		781		0	0%		68	83	60 
MH		53902		108		499		0	0%		114	158	85 
MC		45696		83		550		15	18.07%		110	135	115 
NN		22831		61		374		1	1.64%		130	118	109 
SC		33946		84		404		2	2.38%		174	131	26 
SR		20915		25		836		1	4%		108	108	108 
VG		21557		45		479		0	0%		102	101	101 
								 
Goblin x2:	Damage Taken	# of Attacks	Average Dam/Hit	Evades	Evasion %	STR	VIT	DEX 
AO		95890		154		622		0	0%		84	84	84 
DP		16448		38		432		1	2.63%		84	84	84 
DA		2801		6		466		0	0%		68	83	60 
HT		672		4		168		0	0%		120	110	64 
JB		26324		52		506		0	0%		110	120	110 
MH		82487		216		381		9	4.17%		114	158	85 
NN		32395		145		223		1	0.69%		130	118	109 
SC		86159		122		706		0	0%		174	131	26 
SR		97141		124		783		0	0%		108	108	108 
SK		70051		454		154		3	0.66%		122	174	46 
VG		83558		140		596		0	0%		102	101	101 



Roughly, that's the total damage done to them.
added more detailed spreadsheet by ability:

http://syndicatedlife.com/files/downloads/ffxiv/MeleeData.xlsx


Edited, Apr 12th 2011 9:04pm by Elionara

Edited, Apr 12th 2011 9:55pm by Elionara
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#2 Apr 11 2011 at 1:12 PM Rating: Decent
Scholar
**
844 posts
Elionara wrote:
On NM's why would more strength not equal more damage/hit? At least on average.

:(

Because the battle system is broken.
____________________________
See your face upon the clear water. How dirty! Come! Wash your face!
#3 Apr 11 2011 at 1:14 PM Rating: Decent
**
924 posts
Ilean wrote:
Elionara wrote:
On NM's why would more strength not equal more damage/hit? At least on average.

:(

Because the battle system is broken.


At least curing isn't? :( One of our 50CON's healed a total of 402,000 HP during that time... I sucked ^^; healed for 242,000 @ r43 :P
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#4 Apr 11 2011 at 1:28 PM Rating: Decent
Scholar
**
844 posts
Elionara wrote:
Ilean wrote:
Elionara wrote:
On NM's why would more strength not equal more damage/hit? At least on average.

:(

Because the battle system is broken.


At least curing isn't? :( One of our 50CON's healed a total of 402,000 HP during that time... I sucked ^^; healed for 242,000 @ r43 :P

Just cure the NMs instead then, maybe they'll OD.
____________________________
See your face upon the clear water. How dirty! Come! Wash your face!
#5 Apr 11 2011 at 1:31 PM Rating: Excellent
***
1,636 posts
Could be tons of reasons. Exact same weapon, same damage types? same durability, did someone go yellow or red faster? Exact same gear everywhere else? See what I'm getting at, you need to eliminate every variable.
____________________________


#6 Apr 11 2011 at 1:38 PM Rating: Decent
**
924 posts
KujaKoF wrote:
Could be tons of reasons. Exact same weapon, same damage types? same durability, did someone go yellow or red faster? Exact same gear everywhere else? See what I'm getting at, you need to eliminate every variable.

I can't check them right now since I'm at work but:

VG: (Axe)
Attributes
Strength 102 (102)
Vitality 101 (102)
Dexterity 101 (101)
Intelligence 61 (61)
Mind 60 (62)
Piety 59 (61)

Elements
Fire 84 (84)
Water 84 (84)
Lightning 84 (84)
Wind 84 (84)
Earth 84 (84)
Ice 84 (84)

Axe Stats:
ATK: 132
CRT: 155

SC:(Axe)
Attributes
Strength 174 (198)
Vitality 131 (137)
Dexterity 26 (34)
Intelligence 13 (13)
Mind 15 (19)
Piety 12 (15)

Elements
Fire 90 (90)
Water 13 (13)
Lightning 90 (90)
Wind 174 (174)
Earth 14 (14)
Ice 16 (16)

Axe Stats:
ATK: 146
CRT: 160
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#7 Apr 11 2011 at 1:42 PM Rating: Decent
Scholar
**
844 posts
KujaKoF wrote:
Could be tons of reasons. Exact same weapon, same damage types? same durability, did someone go yellow or red faster? Exact same gear everywhere else? See what I'm getting at, you need to eliminate every variable.

While the lack of variable control makes the test inconclusive, the similarities in every column point to the battle system just ignoring player stats, or having a ridiculously low cap.

Did the 100 STR guy put his points into DEX? No critical rate difference. Of course maybe he had them on VIT instead and every other stat aligned in a way that they ended up parsing very similar, but, by Occam's razor the battle system is broken.
____________________________
See your face upon the clear water. How dirty! Come! Wash your face!
#8 Apr 11 2011 at 2:13 PM Rating: Decent
**
924 posts
Ilean wrote:
KujaKoF wrote:
Could be tons of reasons. Exact same weapon, same damage types? same durability, did someone go yellow or red faster? Exact same gear everywhere else? See what I'm getting at, you need to eliminate every variable.

While the lack of variable control makes the test inconclusive, the similarities in every column point to the battle system just ignoring player stats, or having a ridiculously low cap.

Did the 100 STR guy put his points into DEX? No critical rate difference. Of course maybe he had them on VIT instead and every other stat aligned in a way that they ended up parsing very similar, but, by Occam's razor the battle system is broken.


If the second one swung as much his damage would be about same since their "avg damage/hit" was 44 and crit was pretty dam close as well. Other than not swinging as much I would think the higher STR would yield better. Unless elements plays a role, or there is a cap to how much the str is affected by the monsters defense, maybe other caps in there as well we don't know about?
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#9 Apr 11 2011 at 2:16 PM Rating: Good
Scholar
**
494 posts
could it possibly due to the axe they are using?
for example engraved bhuj has 70% slashing 30% blunt. And a +1 would do more damage then a reg
Barbarian's Bardiche has 60% blunt 40% slashing.
the damage also depends on how worn the axe is. For example if the mrd with the higher str had 0% durability he would be missing a lot and doing considerable less damage than if his axe was fully repaired.

sorry if that sounds stupid, im a noob at parsers
#10 Apr 11 2011 at 2:18 PM Rating: Decent
**
924 posts
Azurymber wrote:
could it possibly due to the axe they are using?
for example engraved bhuj has 70% slashing 30% blunt. And a +1 would do more damage then a reg
Barbarian's Bardiche has 60% blunt 40% slashing.
the damage also depends on how worn the axe is. For example if the mrd with the higher str had 0% durability he would be missing a lot and doing considerable less damage than if his axe was fully repaired.

sorry if that sounds stupid, im a noob at parsers


I think next time we go out, I'll get them with exact same gear, even if it's cheap NQ just to test. lol
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#11 Apr 11 2011 at 2:24 PM Rating: Good
Scholar
**
494 posts
Elionara wrote:
Azurymber wrote:
could it possibly due to the axe they are using?
for example engraved bhuj has 70% slashing 30% blunt. And a +1 would do more damage then a reg
Barbarian's Bardiche has 60% blunt 40% slashing.
the damage also depends on how worn the axe is. For example if the mrd with the higher str had 0% durability he would be missing a lot and doing considerable less damage than if his axe was fully repaired.

sorry if that sounds stupid, im a noob at parsers


I think next time we go out, I'll get them with exact same gear, even if it's cheap NQ just to test. lol


sounds good :] i have a feeling its just that str is broken though lol
#12 Apr 11 2011 at 7:48 PM Rating: Decent
Scholar
**
844 posts
Elionara wrote:
Azurymber wrote:
could it possibly due to the axe they are using?
for example engraved bhuj has 70% slashing 30% blunt. And a +1 would do more damage then a reg
Barbarian's Bardiche has 60% blunt 40% slashing.
the damage also depends on how worn the axe is. For example if the mrd with the higher str had 0% durability he would be missing a lot and doing considerable less damage than if his axe was fully repaired.

sorry if that sounds stupid, im a noob at parsers


I think next time we go out, I'll get them with exact same gear, even if it's cheap NQ just to test. lol

Thanks a lot for taking the time :)
____________________________
See your face upon the clear water. How dirty! Come! Wash your face!
#13 Apr 11 2011 at 7:59 PM Rating: Good
Again, STR is the one stat that does appear to do something for DDs.

Strength testing.
____________________________
#14 Apr 11 2011 at 8:07 PM Rating: Good
***
3,530 posts
Maybe this "ridiculously low cap" was SE's plan: if everything basically caps at ~95, then we may as well just put all stats at ~95 and not worry about reallocation! :D

... they really need to completely rework this game. >_>
____________________________
"... he called to himself a wizard, named Gallery, hoping by this means to escape the paying of the fifteen hundred crowns..." (Machen 15)

"Thus opium is pleasing... on account of the agreeable delirium it produces." (Burke para.6)

"I could only read so much for this paper and the syphilis poem had to go."
#15 Apr 11 2011 at 10:09 PM Rating: Excellent
Avatar
*
235 posts
Well, this game has no auto attack so each person would have to be pressing attack at the same time for the whole fight to get accurate results. The first guy swung their weapon 300 times more then the other person so they obviously have a higher damage total. Average hit damage and critical damage are the same though so, I can see where that wouldn't make sense considering one has a truck load more STR then the other.

I think we all knew that most stats don't really do much yet. I'm not sure how that's even possible and how a Dev team like SE can give us stats(probably the most important thing in RPGs) and let them do nothing. Though, it would follow in suit with a lot of other things in this game and they don't want the "hardcore" players to be better then the casual players.

Edited, Apr 12th 2011 12:11am by Scape13
____________________________


#16 Apr 11 2011 at 11:21 PM Rating: Decent
**
924 posts
I've updated the OP with evasion/damage taken, avg/attack and mob count. I've also included str/vit/dex stats for all people in party :)
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#17 Apr 12 2011 at 12:19 AM Rating: Excellent
1)Were you both standing in the same place?
2)Different skills do different damage, so did each person use the same exact skills in the same proportions?
3)Did either of you eat food?
4)Was all of the gear used identical?

I'm sure there are many other variables that do not seem to be accounted for, but it's late. Please provide a more systematic, controlled set of data with conditions that can be easily replicated by an outside observer.
____________________________
#18 Apr 12 2011 at 6:52 AM Rating: Decent
**
924 posts
Yabusame wrote:
1)Were you both standing in the same place?
2)Different skills do different damage, so did each person use the same exact skills in the same proportions?
3)Did either of you eat food?
4)Was all of the gear used identical?

I'm sure there are many other variables that do not seem to be accounted for, but it's late. Please provide a more systematic, controlled set of data with conditions that can be easily replicated by an outside observer.


I'll be releasing my parsing program along side another early next week. I know it's hard to show these numbers without the answers to your questions but I don't even know if it's possible to know at all times where the ls members are.

However. If anyone has suggestions for how/what the parser should track I'm more than willing to put those in there :)

Also, while I may be the sole developer I can update things on a regular basis. If ever there's something untracked you can tell me what it is and I'll code it in :)

So, Ideas? :) Things anyones noticed in chatlog? :)
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#19 Apr 12 2011 at 6:56 AM Rating: Good
***
1,636 posts
Elionara wrote:
Yabusame wrote:
1)Were you both standing in the same place?
2)Different skills do different damage, so did each person use the same exact skills in the same proportions?
3)Did either of you eat food?
4)Was all of the gear used identical?

I'm sure there are many other variables that do not seem to be accounted for, but it's late. Please provide a more systematic, controlled set of data with conditions that can be easily replicated by an outside observer.


I'll be releasing my parsing program along side another early next week. I know it's hard to show these numbers without the answers to your questions but I don't even know if it's possible to know at all times where the ls members are.

However. If anyone has suggestions for how/what the parser should track I'm more than willing to put those in there :)

Also, while I may be the sole developer I can update things on a regular basis. If ever there's something untracked you can tell me what it is and I'll code it in :)

So, Ideas? :) Things anyones noticed in chatlog? :)


Actually your parser looks great to me. The thing is they only show what actually happened, they cannot be used to prove anything 'scientifically'.
____________________________


#20 Apr 12 2011 at 7:00 AM Rating: Decent
**
924 posts
KujaKoF wrote:
Elionara wrote:
Yabusame wrote:
1)Were you both standing in the same place?
2)Different skills do different damage, so did each person use the same exact skills in the same proportions?
3)Did either of you eat food?
4)Was all of the gear used identical?

I'm sure there are many other variables that do not seem to be accounted for, but it's late. Please provide a more systematic, controlled set of data with conditions that can be easily replicated by an outside observer.


So, Ideas? :) Things anyones noticed in chatlog? :)


Actually your parser looks great to me. The thing is they only show what actually happened, they cannot be used to prove anything 'scientifically'.


Well, later today I'll have it updated to show damage from right/rear and such. I'm using my xml output file of 50,000 lines to test this now, because I don't wanna fight nm's all day lol every day. But it will work just fine with memory after :)
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#21 Apr 12 2011 at 11:37 AM Rating: Decent
**
557 posts
You can certainly estimate the effect of strength even in the absence of an experimentally-controlled design. If you get enough players representing much of the range of the different stats, you can use linear regression to determine the effect of strength independent of the other traits. Sure, things like food and play style will matter, but with enough data that shouldn't matter much (unless players with low strength systematically eat more food that gives them more DPS).

If I had a table with each row containing one player with their stats and a damage measure (whatever measure you want - or use multiple measures), I'd be happy to run a regression.
____________________________


#22 Apr 12 2011 at 5:58 PM Rating: Excellent
If Light Shot does 30 damage, and Heavy Shot does 36, then Archers with identical stats each using the opposite skill will look like they have different damage output on a parse. The Heavy Shot Archer will appear to be doing 20% more damage per hit "because stats are broken", but in reality he's just better at making good parser numbers.

In order for us to actually learn anything real from these numbers the study of them must be scientific. Otherwise, it is only speculation.

What we need is someone to use this parser (which by the way does seem nice and is very impressive) with a strictly laid out set of controlled variables, changing only one variable at a time to test that variable's effectiveness.

My posts on STR and DEX are old, based on early game scenarios, and no longer provide a lot of value to r50 players, but new tests need to have just as strict or more strict guidelines to be factual.

PS - It will be impossible to test STR and remove all other variables on NM fights. Too many people using abilities at random times and skewing the data. Kill something like Raptors, have a tank that does not attack the mob, and have the DDs wear identical gear, identical weapons, and identical stats except for STR. Have those DDs always attack from the same side, do not eat food, do not buff yourselves, and make sure all of them are r50/p50 with 100% repaired gear at the start of the testing. Take an equal amount of swings to make sure weapon degradation is the same, and use the same skills in the same proportion. Or one person can do all the testing themselves solo like I did and just change their STR in increments, but testing with multiple people would be better. This will be a pain in the ****, but anything less isn't all that helpful.


Edited, Apr 12th 2011 8:09pm by Yabusame
____________________________
#23 Apr 12 2011 at 8:28 PM Rating: Good
***
3,530 posts
Scape13 wrote:
I'm not sure how that's even possible and how a Dev team like SE can give us stats(probably the most important thing in RPGs) and let them do nothing. Though, it would follow in suit with a lot of other things in this game and they don't want the "hardcore" players to be better then the casual players.


So even stacking stats is "hardcore" these days, eh? Back in my time, we called that "the most basic, elementary knowledge" which, it would appear, SE lacks. Oh well, looks like it's time distribute random stats while wearing random gear and hit random macros: I'm sure to parse at least in the top 40% that way! XD
____________________________
"... he called to himself a wizard, named Gallery, hoping by this means to escape the paying of the fifteen hundred crowns..." (Machen 15)

"Thus opium is pleasing... on account of the agreeable delirium it produces." (Burke para.6)

"I could only read so much for this paper and the syphilis poem had to go."
#24 Apr 12 2011 at 10:03 PM Rating: Decent
**
924 posts
KaneKitty wrote:
Scape13 wrote:
I'm not sure how that's even possible and how a Dev team like SE can give us stats(probably the most important thing in RPGs) and let them do nothing. Though, it would follow in suit with a lot of other things in this game and they don't want the "hardcore" players to be better then the casual players.


So even stacking stats is "hardcore" these days, eh? Back in my time, we called that "the most basic, elementary knowledge" which, it would appear, SE lacks. Oh well, looks like it's time distribute random stats while wearing random gear and hit random macros: I'm sure to parse at least in the top 40% that way! XD


Updated OP:
Again, this is all from 7 hours of data, Taffer is a Level 4 we had in party :D

Parser is coming along nicely, I have 100+ sets of rules to grab data now from each chatline, I just need to store it in a database like kparser, or xml and call it whenever I need.

You can view a breakdown by ability here:
http://syndicatedlife.com/files/downloads/ffxiv/MeleeData.xlsx

I'd also like to point out, while most of the items/gear they wear I can't attest for, we have alot of +1/2 items on us when playing.

Names have been left in to view stats on lodestone to compare for those willing.

It's not conclusive of course, but I feel 7 hours of data is pretty decent using the same three mobs (I think Dark Poet killed a squirrel at Uraeus hence the high Banish II :P)

Edited, Apr 12th 2011 9:54pm by Elionara
____________________________
http://ffxi.allakhazam.com/story.html?story=18309
Quote:
Like Final Fantasy XI, the game specs will be extremely high for the time, but in about 5 years, an average machine can run it on max settings with little to no issues. Tanaka also expressed interest in making a benchmark program available.

FilthMcNasty wrote:
I endorse this thread.
#25 Apr 13 2011 at 8:14 AM Rating: Excellent
It's not even close to conclusive.

You're pretty good at programming, but terrible at testing theories.
____________________________
#26 Apr 13 2011 at 8:24 AM Rating: Decent
**
557 posts
You don't need more hours, you need more people.

If you can get damage data from 30 people, you can see if there is a relationship between STR and damage dealt while statistically controlling for other traits. You can see if that relationship holds for low STR values or high STR values (i.e., if it's nonlinear, and where the curve starts to bend).

Sure, the players could be playing differently, but I don't think that's a very big deal.
____________________________


#27 Apr 13 2011 at 8:24 AM Rating: Excellent
***
1,636 posts
Again, if you want conclusive data that attempts to prove something, you absolutely cannot do so "in the field", by which I mean on NMs in a group. What you have right now observations with several confounding variables. Its awesome that you're making a parser, (i dont think we have one yet), and I'm sure it will help prove things in the future with the right methods.
____________________________


This forum is read only
This Forum is Read Only!
Recent Visitors: 11 All times are in CST
Anonymous Guests (11)