Sea Gorgets

[Amele]

WS gorgets PDIF Damage Modifier seem to be applied to the first hit of a multiple hit weaponskill.

The question is does the Accuracy apply to the weapon skill, or is the WS Accuracy bonus from the gorget also only to the first hit. This is significantly more difficult to discern...

first, it's an fTP modifier, not a PDIF modifier.

It affects both fTP and PDIF. This was already part of said testing and is not what we're wondering about

I'm sorry, I'm going to need to see capped damage pics and math behind it to back this up before I'll agree with your statement.

nothing I've seen, experienced, or read, about gorgets has even hinted that there was a [non-trivial] PDIF mod.

and I was fully aware of what you were wondering about. if you had bothered to read past my first sentence before deciding to reply, you'd see that I addressed that too. (as did, like, 19 other people in this thread.)

[Yutuyu]

It seems the nature of the fTP bonus needs to be figured out before the pDIF thing can be argued.

[Amele]

It seems the nature of the fTP bonus needs to be figured out before the pDIF thing can be argued.

not really.

since the fTP bonus will be static (either way) for a given weaponskill, use a single hit weaponskill versus a mob with known stats that gives you a PDIF below cap.

weaponskill a capped mob with the gorget and determine the observed fTP.

weaponskill both with and without the gorget versus the 'test mob'; compare damage versus theoretical for the calculated fTP bonus based on weaponskilling a capped mob - see if there is a damage discrepency, consider if it is larger than could be accounted for via flooring errors.

report findings here.

[Yutuyu]

I really have no idea what the hell you said.

The logical, scientific way of doing things is to determine the damage bonus (if it is done by fTP or not) on a mob with a capped out pDIF so that any bonus pDIF may give will be irrelevent because it'll cap out (unless of course the pDIF bonus can go over the cap much like the pDIF bonus granted on X's knife).

Once the damage bonus is figured out, then WS a higher level mob where defense *does* matter, and if the difference between gorget and no gorget remains consistent with the damage bonus from the gorget, then there is no pDIF bonus.

[Amele]

that's exactly what I said, just stepped out.

[VZX]

It seems the nature of the fTP bonus needs to be figured out before the pDIF thing can be argued.

fTP modifier is already confirmed +25/256 in several different kind of WSes.

About pDIF bonus: from Nagamaki's test, it didn't show anything significant on pDIF

[Yutuyu]

It seems the nature of the fTP bonus needs to be figured out before the pDIF thing can be argued.

fTP modifier is already confirmed +25/256 in several different kind of WSes.

About pDIF bonus: from Nagamaki's test, it didn't show anything significant on pDIF

Tests never "confirm" anything. They corrobrate it. And the test I just did does not agree with that conclusion, admittedly 5 is a rather small sample size, I'll do some more tests when I have the time, but the +.1 fTP on the first hit does NOT agree with my findings.



Using the 8.2% I gained from my findings I did math on the minimum values. I multiplied that 8.2 by 256, (in order to find what the top would be /256) and it turned out to be 20.992. Very close to 21.

Then I took the minimum of the non-gorget set, and multiplied it by 277/256 (assuming that my 21/256 ratio is correct) and got the minimum of the second set. Coincidence? More data needs to be taken, but it seems fairly compelling.

[VZX]

Sure
Test more, and post the result here.

[aurik]

Or you could just use the fact that pDIF maxes out at 3 and reverse engineer what your fTP must have been with and without the gorget.

[Yutuyu]

Tests done WSing LV1 Bees from behind in Gustaberg using a Hagun at the listed TP.

100% TP Gekko
498 - 457 - 583 (crit?) - 464 - 461 = 492.6 AVG

108% TP Gekko
441 - 440 - 488 - 636 (crit?) - 551 = 511.2 AVG

100% TP Gorget Gekko
631 - 650 - 510 - 649 - 602 = 608.4 AVG

108% TP Gorget Gekko
656 - 633 - 519 - 584 - 652 = 608.8 AVG



The AVG on the non-gorget 100 and 108% seem to reflect the formula well. Every 1% increase of TP with Gekko is a .005 fTP increase. Thus with the 8% increase, it'd be a +.04 fTP increase. 492.6 (first avg) * 1.04 = 412.304, extremely close to the 108% average. This seems to imply that the samples were relatively stable for the first two.

Obviously the tests on the 2nd bit weren't as stable as the results with the extra 8% TP weren't very good.

But one thing is clear, the difference in damage is much larger than 5%, which is what it *would* be if the modifier on these gorgets was +.1 fTP as people claim it is. All of the tests I've done have, in my opinion, suggested nothing of the sort at all.

EDIT: The only thing that my numbers from all of my tests show is that the +.1 fTP theory is obviously not true, unless there's something I'm failing to take into account (I.E. if it gives a pdif bonus too which doesn't cap out).

[Suiram]

You don't get it, you don't take the average. You SA-WS until you repeatedly get the same highest number. Then you divide by 3, then you ascertain the fTP by plugging in the damage from your weapon and your fSTR (possibly capped at 16) and the 75% of your current STR (which you don't even state here) since you're using Gekko. Undoubtedly that's something along the lines of what the first person did who figured out it was a fTP+0.1 bonus. Also I wouldn't use a Hagun but that's just because I'm anal about not using an item that's going to screw with TP bonus to the weaponskill when I'm testing, you know, the nature of the TP bonus to the weaponskill.

[Amele]

although it would be interesting to find out if maybe there's something about hagun we don't understand yet - since they should be linear, but perhaps there's an interaction between them (which would imply that hagun is not a straight +100TP addition.)

[edit] and gorgets might be a good way to check this[/edit]

[Yutuyu]

You don't get it, you don't take the average. You SA-WS until you repeatedly get the same highest number. Then you divide by 3, then you ascertain the fTP by plugging in the damage from your weapon and your fSTR (possibly capped at 16) and the 75% of your current STR (which you don't even state here) since you're using Gekko. Undoubtedly that's something along the lines of what the first person did who figured out it was a fTP+0.1 bonus. Also I wouldn't use a Hagun but that's just because I'm anal about not using an item that's going to screw with TP bonus to the weaponskill when I'm testing, you know, the nature of the TP bonus to the weaponskill.

No, actually, you clearly don't get it or understand statistics or testing at all. You're suggesting doing enough weapon skills to find the max damage and then apply an equation, yet what you don't realize is that, that method of testing has an additional set of assumptions. My method is far more scientific and accurate.

The only difference between the two samples is having the gorget equipped or not. Thus it is the variable, the only factor which changes. With gorget on, damage shoots up by a lot more than it should if the only thing the gorget did was give +.1 fTP.

The simple fact is, unless my numbers are randomly skewed by chance (and I've WS'd a total of 30 times now and collected the data), the +1 fTP thing is wrong, plain and simple.

EDIT: Or as the above poster stated Hagun could be screwing something up. I'll do some tests without it to remove that possibility.

Either way, it's incredibly unprofessional to keep adding in your little ad-hoc hypothesis in order to preserve the conventional wisdom which is quite clearly not correct.

[Suiram]

It's pretty clear now that you have no idea what you're doing and instead simply throwing around terms like "pDIF" and "fTP" without any insight into the mechanics of the formulas you purport to be testing, so rather than dignify what you just wrote with a response, instead I'll just sit back laugh at you with the rest of the people who know what the fuck they're talking about. I too look forward to reading more of your tests, honestly.

[Amele]

all the formulas on the wiki are written under the assumption that you are attempting to calculate the mode of the damage you are doing.

the random number portions have been excluded so using the average is slightly flawed. all that said, .23 is probably within the margin of error of a sample size of 5 once the random variations are added back in. (would need someone more versed with statistics than I am currently; I don't generally use it in my day to day work.)

[Yutuyu]

My methodology is fairly simple.

If the gorgets truly are a fTP +.1 effect, then we should notice an increased damage percentage on average of 5% with Tachi Gekko at 200% TP.

Tachi Gekko's fTP at 200% is 2.00, so.
Assume (D+fSTR+WSC)*pDIF = x.
Then without the gorget
WS Damage = x*fTP, and fTP = 2, therefore WS Damage = 2x
With the gorget WS Damage = x(2+.1) = 2.1x.
Damage difference is = 2.1x - 2x = .1x.
.1x/2x = .05, or 5%

Then I did tests to find the averages. They were apart by more than 5%, so the +1 fTP thing probably isn't true (unless one of my assumptions is incorrect, i.e. Hagun works differently than +100% TP).

[Amele]

you're also assuming the random values are normally distributed, which isn't necessarily the case. (since you are substituting average for mode)

[Yutuyu]

With a large enough sample size, anything can be approximated with a normal distribution. I.E. 30 (by law of large numbers). I can't remember what this is exactly... it's close to testing the two means of a population (been awhile since I've taken my statistics classes). Either way, my numbers aren't "conclusive" but they're certainly compelling enough to challenge the previous conclusion. More data needs to be collected, obviously.

EDIT: Check out my data too. It actually does look pretty normally distributed. Lots of data around the mean, couple points off stray. But if I plotted it, I'd get a bell-curve.

[Ferien]

First, you can't assume normal distribution when you're dealing with numbers with upper and/or lower caps. Both caps will skew the data. PDIF has both an upper AND a lower cap, and on the mobs your fighting to test this, you WILL more often hit at the upper cap than the lower cap.

Second, if you really want to do a "logical, scientific" test of the data, you must isolate ONE variable at a time. If you're testing for fTP, you MUST keep PDIF constant, which is why we test using the consistently highest number. If you're testing for PDIF, you MUST keep fTP constant, which is why you use the same amount of TP for every attack.

You're making an assumption about fTP based on tests where you're varying both fTP and PDIF, therefore your tests are invalid. To test two different variables, you need two different pools of test data. In short, stop arguing with those who know more about game mechanics than you do, and do your research before you try to post flawed tests.

[Amele]

sample sizes under 10 aren't even worth trying to plot.

and no. you can approximate any sample as a normal sample within a certain margin of error which must then be carried through to the final term.

presupposing that the distribution is not normal (it isn't because there's a bound), and given your small sample size; I think the error you're showing is pretty reasonable (i.e. I don't find this compelling.)