Talk:Kick Attacks

Kick Attack Proc Rate[edit]

Kick Attacks II at 71 does increase the frequency of kicks as opposed to the 51 trait.

Do + to kick attacks or items enhancing them grant Kick Attacks if the current job doesn't have it? Tahngarthor 02:29, 28 November 2006 (EST)

Perhaps, if you're using H2H weapons, it acts just like items that enhance Double Attack. It's just you'll have a very low proc rate. If all you have is Kick Attacks +2, you only have a 2% chance of Kick Attacks going off if you're not a MNK. --Chrisjander 02:53, 28 November 2006 (EST)
I'm a 75 PUP; I just recently got Bahamut's Hose, and it does indeed grant me kick attacks. Tahngarthor 15:52, 11 March 2007 (EDT)

What's the base Kick Attacks rate with Kick Attacks I and II? --Divisortheory 02:27, 17 March 2007 (EDT)

KAII at 75 is theorized to be: base 10%
+1% per merit
+1% per "+1 Kick Attacks" Verification Needed (this is disputed by some, and generating scientifically significant data on a precision of 1% or better is really really time consuming)

Not sure about KAI, would need some quality time in Pso'Xja-60 to figure that one out. Be my guest :P --Aurikasura 06:06, 17 March 2007 (EDT)

my +6 Kick Attacks from the bahamut's hose (on PUP) does *seem* to proc more than 6% of the time. Tahngarthor


There is currently no data in support of a +1 = +1% for kicks and tests suggest +5 gives closer to just 2% or 3%. If you want to believe in this or convince others of it, find and reference some data that supports it or do some tests yourself.

Believing something just because people believed it for a long time without any evidence, then locking a page with what all available evidence suggests is the wrong information labeled as a "fact" is not a good solution and is the opposite of what a wikipedia should present.

A good wikipedia entry would at least present both claims and discuss the amount of support (or lack of) for each claim, letting people make up their own mind given the lack of a conclusive answer. If your only reason for stating something as fact even in the absence of supporting data and presence of evidence to the contrary is that one was the "status quo" for a long time, you should state that clearly and explicitly.

The tests showing AF2 increasing kicks by 3% followed the correct methodology and had appropriate controls and a fairly modest size, they just lacked a large enough size to conclusively disprove the alternative assuming a Bernoulli distribution (which is a big assumption already as this depends on how attack events are coded in FFXI). --Genome 17:25, 23 April 2007 (EDT)

First, what makes a good wikipedia entry is one that cites data, not one that is arbitrarily changed based on data hidden on lesser-frequented forums in order to avoid peer review which might show that the analysis of the data is wrong. Speaking of, please post your data here so that people may perform their own analysis instead of relying the unknown competency of one person for a conclusion.

Second, there is data supporting +1 = 1%. Your data supports it. Furthermore, considering many other mechanics in this game operate under a "+1 -> 1% increase" mode, it's a reasonable conclusion. Finally, it's also been partially supported by data generated by me (I tested 2 merits + melee hose to be in the 17% range a long time ago, I don't know if I still have the raw data) and most likely by other monks as well. To this date, I have yet to see a data set which FAILS to support the theory that +1 -> 1% increase.

The article as it stands now is 100% correct, because it does not state the conclusion that +1 -> 1% as fact. The article as you wish it to be is NOT 100% correct, because of two reasons: 1) "actual tests show that the rate of increase ... is much lower" is false. The actual tests that you are citing fail to show that the rate of increase is lower that 1->1% with any reasonable amount of confidence. 2) "It was previously believed that ... "Kick Attacks +" would increase the rate of kicks by that many percentage. However there is no evidence to support that claim" is false. My tests long ago provided data that supported the +1 -> 1% theory, under the assumption that base rate is 10% and each merit adds 1%.

Third, if you're discarding the idea that a series of attack rounds fits the necessary requirements to be treated as a Bernoulli Trial, you should stop pushing theories which assume that very fact.

--Aurikasura 02:03, 24 April 2007 (EDT)

- The data has been public for over half a year, and I referenced it repeatedly with full access to the whole data with a publicly accessible link, and again recently in the wiki discussion forum to give a chance to whoever might have any desire to further analyze it or present additional data to do so, for the purpose of correcting this entry. It doesn't support the +1=+1% idea, it just can't conclusively eliminate the possibility that the value is low due to a combination of random chance and error margin due to the sample size. On a side note, if you have a suggestion on how to reference or present data and tests on this wiki I have no problem doing so. I am not sure this wiki has done much in the way of embedding data and the general approach is to reference it anyways. Also, you should avoid using loaded sentences that insinuate malicious intent in these types of discussion (from your statement above: "..pushing theories.." or "..in order to avoid peer review.."), as these are irrelevant to the topic of activation rate of AF2 in the game.

- Your own experiments also showed a kick rate lower than expected from AF2, but suffered from several problems: an even smaller sample size (I think less than half of what I used) and no negative control (you only measured your kick attack rate while wearing AF2 plus a couple merits at 75, never comparing it with a similar run without AF2). At the end, you obtained a value that was fairly low, started by assuming the conclusion that AF2 MUST give +5%, and worked back to conclude that kick attacks I and II combined would be 10% instead of the currently accepted 15% (10% + 5% which is by the way also stated in the wiki entry and has been supported by multiple tests, you can see that even my control shows 20% base, 15%+5% from merits).

- All the data we have essentially shows a much lower than expected kick attack rate from items like AF2, yet it's very consistent when looking at just kick attacks traits and merits. I think it's hand-waving to say that since the sample sizes are not large enough it's just a big coincidence the numbers always end up low and never end up higher than expected, and that if you assume a large enough error margin all data supports all theories in the sense that you can discount all results that don't fit your favorite theory as error margins...

- On your third point, I simply would not encourage people to rely too heavily on an assumption that could easily be false depending on how attack rounds are coded in FFXI. Choosing to discard the idea of bernoulli trials completely would be just as wrong, in fact what you propose is a false-dicotomy (you seem to imply that if you question whether bernoulli trials can be applied since events might not be independent, you would have to reject other unrelated experimental results). You shouldn't assume a distribution before testing that all your assumptions for choosing it are true.

- A simple example would be this: on top of rnd and probability code, for all we know FFXI might have a couple of lines of code that check and force attack events to fit caps at least every say 40 rounds, so a counter checks if you ever hit 38 times in a row, it WILL force the next 2 hits to be misses (to keep max acc to near 95% even on smaller sample numbers) so even if a few times bad rnd or %chances don't work themselves out, they might cover their back this way to get fairly consistent caps. This is just an example, but you should already see that in this case you couldn't assume a bernoulli distribution as the attack rounds would not be independent events at all, and in fact in that specific example a much smaller sample size would be enough to get reasonably repeatable and accurate readings, much smaller than what you would expect from a true series of bernoulli trials.

--Genome 06:40, 24 April 2007 (EDT)

Could both of you (Genome and Aurikasura) please post links to your sources here on the discussion page so that others can see them and participate in the conversation as well if they wish. Even if you posted them in the forums, please repost links here as well, I don't want anyone coming in from a blind perspective. Thank you both. --Chrisjander 08:16, 24 April 2007 (EDT)

Here is a link to my data, it shows the data and explains how the test was set up: http://genomeffxi.livejournal.com/12872.html

--Genome 16:06, 24 April 2007 (EDT)

First, can you please copy/paste the relevant raw data to this page, I don't think that wiki is a good place for you to advertise your livejournal, and I don't feel like digging around in your personal diary which is loaded with your own biases and conclusions. Thanks.

Second, this data DOES support the +1 = 1% idea, if you do the statistical analysis. If you construct a null hypothesis and calculate the likelihood of results such as yours, you'll see that it's not very unlikely at all. Like it or not, your data supports the +1 = 1% theory under conventional, accepted statistics.

Third, you silently changed this subject three times without so much as engaging in discussing after placing your data in nonconventional venues when you knew that your theories and data were far from being accepted -- on the contrary, people have already pointed out to you several times that your data fails to support your claim that +1 != 1% several times. If the page hadn't been protected, I assume you would have simply done it again, so I'm glad that your hand has been forced and now you must defend your ideas. However, don't act this way in the future, as it's even more counterproductive than someone spending the time calling you out for it.

Fourth, my ancient data (which I don't have a link to anymore, if you can find it be my guest and repost it here) didn't show that my kick rate was abnormally low given my assumptions. In fact, my results were extremely likely if the null hypothesis "+1 -> 1%" were true. Furthermore, 15% is not the accepted value for base kick attacks with no merits, and 20% is not the accepted value for kick attacks with 5 merits and no gear, and your data does not support rejecting the long-standing belief (and anecdotally tested many times) that it is 10 and 15%, respectively. Your belief that the values written for KAI and II on Wiki are meant to stack is wrong: trait values are presented in a way such that you only gain the benefit from the highest trait obtained. You don't get the benefit of both Attack Bonus I and Attack Bonus II from being 30+ DRK.

Fifth, you say All the data we have essentially shows a much lower than expected kick attack rate from items like AF2, yet it's very consistent when looking at just kick attacks traits and merits. The problem is that we fundamentally disagree that your data shows a "much lower than expected kick attack rate". Your data shows that you are within an reasonable amount of error from the predicted value, given the number of trials performed and with a reasonable amount of confidence.

Sixth, if you're rejecting the idea that a sequence of attacks don't qualify for analysis as a Bernoulli trial, then you need to propose an alternate model, describe the impact of this change, and show us evidence that would support your incredible claim. Personally, I think it's hand-waving to say that basic, first-year statistics should not apply to your data and simply because the result of a short test gave some result, that this result must be the exact value. I await your evidence that suggests anything but Bernoulli should be the case.

--Aurikasura 16:56, 24 April 2007 (EDT)

  • I think most of those points are not productive areas of discussion as they do not address the topic at hand, which is kick attacks activation rate.
    • If you really want to discuss your theories about how many times you think I change subjects, what you believe I think, the differences between adding a reference to test data and advertising, what I should or shouldn't do, etc, maybe you can send me PMs about it or discuss it somewhere else, as those are all irrelevant to determining the actual kick attack rates in FFXI.
  • Back on the topic, if I understand you correctly you are saying that:
    • you refuse to click and look at the data I presented (there's no searching or sorting through involved).
    • you also refuse to present your own data, or anybody else's data that might be relevant to answer this question, and related questions like whether kick attacks II trait gives 10% or 15% (since you can't get the 2nd trait by itself, calling it stacking or just higher value is just semantics), for which again no data whatsoever is presented in support of a 10% claim.
    • your language suggests you believe some data in support of your position must exist somewhere, even in abundance (suggested by your use of terms like "anecdotally tested many times", let's skip over the fact that anecdotal tests on this sort of things are very unreliable...) yet again even that is missing.
  • I think to address the topic properly, if someone has more data it can only help us find an actual answer and possibly remove any claim from the kick attacks page that is not standing on anything other than "longterm belief", wishful thinking, and the likes, and replace them with things that are actually reflected in the game, tested and have data to support them, to which people can add to.
    • If for example some people came forward with well-setup tests showing kick attack increases from wearing melee hose even higher than 5%, that would give an actual approximate error margin for a certain sample number and even give some more support of what the actual value would be, i.e. you would get a distribution of actual results and could look at its width and read its center value to get your answer. Otherwise it would be reasonable to believe the results supported by the most data until more data is available.
  • As a side point, if you want to save some work and try to infer or reject things based on what you refer to as first-year statistics, you have to support your claims that the distribution you assumed can actually be applied to your data. For example a bernoulli distribution can ONLY be used if you have a series of independent events. I already gave you a clear example of how events can be coded in a way that makes them not independent of each other and gives much narrower distributions than would be assumed in the bernoulli trials case.
    • To verify this, you could simply point to any data that shows the distribution of values from repeated tests from any attack event, for example misses vs hits with capped accuracy. If it really did follow a bernoulli distribution, it should scale up to the cap accordingly. Once you or someone else can show fairly conclusively that attack rounds in FFXI are independent and can thus be assumed to fit such a distribution, then you could use it to estimate possible error margins and acceptable sample sizes for all types of attack events.
    • However I think you would find that this is not the case at all in FFXI, since you might have noticed people's error margins from parsing even small numbers of fights seem to converge to near 95% much faster and with smaller variability than could be explained with that type of distribution or assuming independent attack events in general.

--Genome 03:02, 25 April 2007 (EDT)

First, you absolutely must cross-post the data here, as I've said now several times. Post the data. A link to your LJ does not cut it. How many times and how many different ways do I need to tell you this before you grasp what I'm saying? Post the data here. Not link it, post it. Post it here. Do you understand yet?

Second, I have not refused to present data. The word "refuse" is a loaded word. For someone who complains about other people using loaded language, you seem to have no issue in employing it yourself. I have no responsibility to provide data to support the prevailing theory, as no evidence has come to light that casts doubt on the prevailing theory; there is no refusal involved.

Third, do not read what my language "implies". Read what it says and respond accordingly, instead of constructing straw-men to push over. I never said there was an abundant amount of data that supported any theory. Therefore pointing out that this data which should be abundant is missing is a straw-man and I reject your characterization of my argument. Please try again, this time without your creative interpretation of what I am saying.

Fourth, there does exist some data predating your tests which support the prevailing theory. This is my ancient kick attack tests, which show at least in one configuration, the actual kick attack rate was within a reasonable distance from the theorized kick attack rate. The test is buried on Killing Ifrit monk forums, I do not know where.

Fifth, I am pointing out that your interpretation of what wiki is presenting is incorrect if you think it was presenting 15% as the actual final kick attack rate with no merits. The wiki claims and has claimed for a long time that KA2 is 10% final rate, not a 10% increase from KA1, and your misreading of this does not support your claim that the ancient kick attack tests I performed showed that KA rate with 2 merits and melee hose was low. On a tangent, you mention that "multiple tests" have shown base KA with no merits to be 15%. Please cite.

Sixth, the KA page must either remain uninformative or be based on conjecture because no data exists to firmly establish the exact details of the game mechanic. If any conjecture were to be made, parallels can be drawn to traits such as Conserve MP, Subtle Blow, Store TP, and Magic Attack Bonus which all follow the pattern of +X = X% increase. Since no data exists to show contrary, this is a reasonable conjecture to make, and the current revision of the page already uses language to indicate that this is a conjecture, not a proven fact.

Seventh, if you wish to reject the assumption that a set of trials does not fit the parameters for Bernoulli, that is your perogative. However, there is no evidence I know of that indicates this is true, so Occam's Razor suggests that I should reject your unnecessary muddying of the waters until you show us good reason. On that note, you state: "parsing ... seem[s] to converge...much faster and with smaller variability than could be explained...assuming independent...events". Please cite the data you are referencing, and show the method you used to arrive at this conclusion.

--Aurikasura 06:05, 25 April 2007 (EDT)


Refer back to my previous answer as all points still apply to your latest reply.

  • You raise a few side discussions that are again irrelevant to the question of kick attacks rate, I will address them as briefly as possible so that the discussion is not sidetracked.
    • "refuse" is not a loaded word here (you were asked to provide some data and you expressed clearly and explicitly that you are unwilling to do so, that's near verbatim the definition of the word refuse).
    • there was no straw-man argument. Your choice of sentences in regards to anecdotal tests you believe exist, but do not provide, specifically pointed to that conclusion. The only argument I 'pushed over' is that you mention these anecdotal tests and then fail to provide them. You may wish to retract or clarify your statements if that was not what you meant.
    • your notion of what you think people 'must' do is not pertinent to this discussion.


  • Now, back to the topic:
    • Present some data. Making long-winded arguments only gets so far, actual test results would be a lot more useful and would spare you from all this adversarial discussions and unrelated side arguments, shifting back the topic to analyzing and putting together all the data, tests, and results collected by people instead of just refusing to look at ones available and presented here (only mine so far) or producing additional ones of your own.
    • Test data doesn't even have to be yours. If you do not have ANY data to back up your beliefs, however, you should not have a reason to believe them and should be reasonably persuaded by actual data like the one I presented, until even more conclusive data is taken.
    • If you mention something like what you keep referring to as your "ancient data" (which might not be the best way to describe it since it was just from about a year ago, 6 months or so before my tests but long after kick merits were introduced or many NA players obtained their AF2s) and even try to use it to justify claims on the kick attack pages, you should present it so people can analyze it.
    • As I mentioned, the problem with that test is that it was just one test with smaller sample size that never even compared AF2 to no-AF2, reading a total kick value generated by traits+AF2+merits and then assumed the conclusion by proceeding to guess how these 3 factors are distributed in the final sum value. You can't do that, for example there are many different solutions to something like X + Y = 4, you can't use just that to try to prove that X is 1 and Y is 3 any more than you can prove the opposite is true. If you also get a value with and without AF2 as I did in my tests, you get something like X+Y=4 AND X=3 and can now justifiably conclude that Y=1. That's why I can claim that data actually supports a 15% rate from kick attacks II trait (or I&II, again semantics), because my data does support it explicitly. From that I can also use the base value I read directly and apply it to your sum-only result and also show that your data shows a low value of kick attack rate from AF2 assuming the experimentally determined base kick attack rate. If you wanted to disprove it or challenge it, you could repeat your tests this time comparing both AF2 and no AF2 to get a base reading and the amount added by AF2. If those numbers were actually different from mine I would be surprised but it would be more convincing than some appeal to "longheld beliefs", and probably reason enough to perform further tests.
    • The argument on analogies to other stats can work both ways, there are also other stats that are not a direct conversion to %, and the same equally valid argument could be made for the opposite. In addition, the fact that SE chose to explicitly include the % for kick attack merits and just a generic +1 (without the %) in AF2 or any of the newer kicks+ gear (like they did for triple attack+1 in THF AF2 vs something like indra katars) is also another analogy argument that can be made in favor of kicks+1 being different than +1%. Either way, when results from actual tests casts doubts on something that was assumed just by virtue of analogy, you should strongly consider revising your beliefs (or conduct further testing to at least try to generate some evidence to support your position), rather than remaining entrenched in your position and refusing actual evidence to the contrary.
    • In regards to evidence suggesting that using a bernoulli distribution and assuming truly independent attack events is likely incorrect, there is evidence in almost any parse and in the general consistency and smaller than expected variance between parses: for example, consider the data I discussed when comparing Togi, Osode, Meat and Sushi in KRT, data generated and parsed by other monks on a different server, it compared 4 different conditions and with a fairly small sample size each time, 100 fights:
    • The sample number for something like counters, which is clearly much lower than something like attack rounds and distributed between monks is actually fairly small, so a bernoulli assumption would predict a large error margin and require a large confidence interval. Yet the variance here is less than half a percent point on counters between those sets (e.g., between 43.1% and 43.5% and between 45.1% and 44.4%). You can repeat this with other parses and other parameters (like the example of capped accuracy I discussed in my previous answer) and see similar results, that the distribution is actually much narrower and converges faster than would be predicted by a bernoulli distribution assumption.

--Genome 08:23, 25 April 2007 (EDT)

Let's stop this nonsense. You have a lot of explaining to do, and your voluminous text does little to close the issues which I am raising, and instead raises new ones.

1) refuse is definitely a loaded word. It has the connotation that I have the ability to do something and yet decline to do that thing. You suggest that I am refusing to provide data when in reality I am simply not providing data. There is an important distinction. If you think I am refusing, cite where I am refusing, and explain why you consider it refusing. You must either support your accusation or leave it unsaid, otherwise it is a baseless accusation and not welcome here.

2) There was indeed a straw-man argument you constructed, and I quite clearly explained where it was and why I thought it was a straw-man. Again, you have taken what I said and "interpreted" it to mean something that I did not say, and then pushed over that thing. That's a straw-man argument.

3) What I think people "must do" is entirely relevant to the dispute at hand. I'm outlining what I think is imperative in order to resolve the issue. Do you disagree these imperatives? Which ones and why?

4) I have presented my case already, you just missed it. I repeat myself from point 6 of my previous reply: the KA page must either remain uninformative or be based on conjecture because no data exists to firmly establish the exact details of the game mechanic. If any conjecture were to be made, parallels can be drawn to traits such as Conserve MP, Subtle Blow, Store TP, and Magic Attack Bonus which all follow the pattern of +X = X% increase. Since no data exists to show contrary, this is a reasonable conjecture to make, and the current revision of the page already uses language to indicate that this is a conjecture, not a proven fact

5) I have asked you repeatedly to copy/paste your data here so that we can work with it inline. I reject your claim that posting a link is sufficient, because I want to work with the numbers inline on this page. Why have you not done this? Why will you not comply with this request?

6) I have already stated several times that I no longer know where my so-called 'ancient' data is. (However, since none of my claims rely on this data, this does not impact my argument) However you seem to indicate that you know where or know of this data. Do you know where this data is?

7) Some of your criticisms of the so-called ancient data is accurate: the sample size is small, and it only tests one configuration. However, despite that, the results do support the predicted KA value of the 10/+1->1% theory, hereafter referred as the 'prevailing theory' despite the fact it does not contain a before-and-after relationship. To borrow your language, discovering that "X+Y=4" supports the theory that X is 3 and Y is 1. Since my data showed that X+Y+Z is close to 16, it supports the prevailing theory which predicted X+Y+Z=17 with my given setup. Thus it fails to cast doubt on the prevailing theory.

8) The claim that base KA2 base rate is not 10% and KA2+5merit is not 15% is a radical claim and thus requires radical proof. Your data does not satisfy the need for proof as your result was not unreasonably distant from the predicted frequencies in either of your tests. Do you have any data which casts doubt on the base KA rate of the prevailing theory?

9) You state that there exists "other stats" where +1 does not increase by 1%. However, KA is a job trait, not a stat. Are there any job traits where +1 does not increase by 1%? If so, please cite them.

10) Reliance on SE's differentiation between +1 and +1% is a shaky argument, as can be seen on Goliard cuffs which say "+4%" but recent tests seem to show they act no different than "+4". Also to note are the nontrivial amount of translation errors, such as the one that was recently fixed with respect to defense on the AF2 Paladin body piece. Are there any examples where there is an actual, functional difference between +1 and +1% on a piece of gear?

11) As yet, no results from any actual tests cast doubt on the prevailing theory. As I've mentioned several times, your data does not cast doubt on the prevailing theory. Therefore I see no reason to revise my beliefs. In addition, I have not 'refused' (there's that word again) any evidence to the contrary because no evidence to the contrary exists. If you have evidence to the contrary, please post it and show how it is "to the contrary".

12) In regards to evidence concerning the validity of using a bernoulli distribution, while you cited some data which may support your position, I do not see any statistical analysis that leads to your conclusion. You need to show the work you did that made you arrive at this conclusion. In particular, you need to show mathematically that the results are abnormally clustered -- simply asserting that they're abnormally clustered does not make the grade.

If you would please respond to each point in turn and clearly label which point you are responding to in each paragraph so that we can reduce the sprawling it would be appreciated.

--Aurikasura 17:26, 25 April 2007 (EDT)


As I did already in my previous answer, I will try to be as concise as possible in addressing points that are not relevant to the question kick attack rates to avoid sidetracking:

  1. It's refusing. (see explanation in my previous answer)
  2. there was no straw man argument, you suggested there were anecdotal tests and didn't even reference or cite those (see more detailed explanation in my previous answer)
  3. what you think people must do is again not relevant (as I explained earlier). Those are irrelevant imperatives, data is cited, referenced and accessible, your suggestion to inline it is both unnecessary and a bad practice, tests need to be referenced and cited back to the original source just as it's done in all peer-reviewed scientific literature to better discuss them.
  4. You have not presented any data, nobody has shown any type of support to what you call a prevailing theory. The current kick attack pages contains information that does not match the currently available tests and results and should be changed. You can still leave your assumptions of a base kick rate of 10% and a 5% from melee hose, but from all you have presented here you MUST specify that you believe this just from word of mouth superstition, and that when actually tested these values were shown to be 15% and 2-3%.
  5. (see point 3 and my previous answer).
  6. it's somewhere on KI, you had pointed it out to me and I looked at it back then. It was from about a year ago so I don't think it would have been erased. You should use the search function in that forum if you wish to find it.
  7. There is no support ever presented for what you call a 'prevailing theory'. Just repeating what might be a wild guess from years past does not prove anything. Where does a value of 10% for base kick rate comes from? Was there ever any test done to arrive at this number? In fact the only data we have points to a 15% rate, which is also consistent with your data. (see my previous answer for more discussion on this point).
  8. All data available so far supports a 15% base kick rate. No data whatsoever has been put forward to support a 10% base rate. A 10% rate is what we call superstition, while a 15% rate is the best current theory supported by all available data. The data casts doubts on a 10% rate by showing explicitly a 15% rate in the absence of kick+ gear.
  9. see my previous answer again. Triple attack is a job trait, and again SE chose to use % vs + in different gear with different results.
  10. all of these analogies are fairly shaky, especially in contrast to actual tests. That was the point I was making. Refer back to my previous answer for a longer explanation.
  11. I posted evidence already. What you call "prevailing theory" has no evidence to back it up. Who first made this claim, when and why? Without any data to support it and with data showing differently, it's just a superstition that survived probably longer than it should have, like CHR+ = enmity+ or other longheld beliefs that were shown to be false. You don't have to change your beliefs, the wiki entry however should be changed to reflect actual game mechanics.
  12. The data is there for anybody to use, it was just one of many examples but feel free to do it. In that particular case you will have to make some estimates on the number of hits received, as I only have the number of fights and % counters.

Now, to reiterate:

  • As things stand, all the data presented here shows a 15% base kick rate and melee hose adding just 2-3%.
  • The current wiki entry here is incorrect, and should at the very least explicitly state that there is NO factual support for the incorrect claims it makes (a base kick rate of 10% and additional +5% from AF2). So far these appear to be just longstanding superstitions that are not reflected in the actual game.
  • Unless actual evidence is presented or produced through valid tests, the entry should be corrected to reflect actual game data as I presented here.
  • Most of the 12 points we just discussed are side arguments that have no bearing on the actual subject of discussion here, which rests on the 3 points above.

--Genome 01:35, 26 April 2007 (EDT)

1) You have not cited where I am refusing. Cite where I am refusing, or stop repeating this accusation baselessly.

2a) There was a straw man argument: you defeated the argument that there should be a great volume of data supporting the prevailing theory which is an argument I never presented. This is the very definition of a straw-man tactic.

2b) I can cite support for "anecdotally tested many times". You conclude that I can't simply because I haven't bothered to. If you want to contend that I haven't, ask me to cite support. Don't jump to conclusions.

3a) What I think is imperative is extremely important if we want this issue to be resolved to the satisfaction of all parties: I'm telling you what you need to do in order to resolve this to my satisfaction. I think you need to post the data here for statistical analysis.

3b) Now your argument that "there is no need to copypaste the data here" has actual meat on it. However, I disagree for several reasons, and I think the data should be posted here. i) Your site is ephemeral. You can delete the post at any time and we simply would not have the data anymore. ii) We should strive to keep all the data in a neutral arena. This page is neutral. Your site contains thinly-veiled personal attacks. See for example, entry 15432. iii) Your site has no auditable history. You can modify the post without any review or even any history showing the the change. iv) Without the original source data reproduced, further analysis on this page would be harder to follow. v) It is neither difficult nor time-consuming for you to repost the data here. Therefore the impact upon you is nearly negligible.

4) I have presented data that supports the prevailing theory: see my "conjecture" argument. Store TP, subtle blow, magic attack bonus and conserve MP all work on a +1 = 1% scheme. Therefore this data supports the argument that logically kick attacks should follow this pattern as well. Therefore your claim that I have not presented data is false.

5) You are refusing to copy/paste your data here? Hypocrisy...

6) http://ffxi.killvoid.com/forums.php?m=posts&p=2133675 (took me a bunch of searching):
"In the spirit of the great brutal earring parse, here's the number of rounds that it took for each kick-attack round:

13,14,4,7,10,3,6,2,11,2,22,3,7,3,3,1,1,7,3,7,2,10,2,2,1,6, 1,3,2,4,11,4,6,7,2,1,1,7,1,6,12,2,13,3,10,4,20,5,12,10,3,12,2,1,1,15

Total of 325 attack rounds
Total of 56 kick attacks
Rate: 17.23%"

This was with 1 merit and melee hose equipped. The prevailing theory would predict a 16% overall kick rate. This is close to that, therefore this tests supports the prevailing theory.

7) The only data we have points at a base rate somewhere in a very wide range, a range that easily includes 10%. Your dataset simply isn't large enough to narrow down the range to a single number. The longstanding belief has been 10% and as of yet no data has been presented which would cause us to question this belief.

8) No data whatsoever has been put forward that suggests 10% base rate is incorrect. 2 sets of data have been mentioned that support a 10% base rate: the data you generated, and my data posted above. In addition, people have anecdotally confirmed the rate as well.

9) You missed what I was asking: is there a difference between Triple Attack +1 gear and Triple Attack +1% gear? Can you cite data which shows that they have a different effect? Because as far as I'm aware, Triple Attack +1 gives the same bonus to Triple Attack that Triple Attack +1% gives--which would give even more credence to the prevailing theory.

10) The parallel between Kick Attack and Conserve MP is actually quite solid. The others are slightly different. You still haven't cited any job traits where +1 gives anything BUT a 1% bonus.

11a) You haven't posted any evidence here. See point 3b about the necessity of posting the data here and point 5 about your hypocrisy

11b) See point 15.

12) You say this data supports your rejection of a Bernoulli distribution, however you won't show us the math that leads you to that particular, radical conclusion. Not only that, but you admit that the data you cited is insufficient to come to that very conclusion without outside assumptions. So I reiterate my stance: there is no evidence I know of that indicates this is true, so Occam's Razor suggests that I should reject your unnecessary muddying of the waters until you show us good reason.

13) No data posted here suggests that the main article is incorrect.

14a) See point 13. 14b) The statement you suggest is meaningless because you haven't identified which incorrect claims to which you refer. 14c) The main article already uses language that indicates the information presented is conjecture rather than solidly proven.

15) The article is already consistent with your data.

--Aurikasura 03:28, 26 April 2007 (EDT)


I added some links for those of us that may get confused with complicated terminology, aurik, so don't be alarmed, I didn't change anything of what you said. We are going to assume that data posted is posted in good faith, and that no one has altered it to support their argument. The fact that genome's data is on another site is irrelevant, unless that site allows for someone other than the original poster to change the information without data-history. However, Genome, posting a summary of the data (like the example aurik gave of his data) could greatly enhance your argument. Many people don't want to have to go to an external link (although giving them the option is good) to sift through raw data. Both of you are very passionate about this issue, and I don't want to see this turn into a an argument where we're questioning whether someone is doctoring their data. --Chrisjander 09:54, 26 April 2007 (EDT)

Wikified those external links. --Charitwo 13:57, 26 April 2007 (EDT)

Let me add my $.02. First, my MNK is level 0. Second, I know nothing about KA, rates, etc., so all I have is what is presented above (I have read this page, read the KI thread, and read the livejournal, and examined the history of the main article here). As CJ said, you are both very passionate about the issue which is fine, but please try to keep the comments on topic and factual. There is a lot of unnecessary and counterproductive back and forth here. If you want to have this extensive back and forth take it to the forums where it is far easier to quote and respond. These talk pages are great for smaller discussions but I think at this point the forums should take over.
Based on what I've seen I don't see how this can be resolved without additional testing - preferably with more than just melee hose (although I recognize that only 3 pieces give KA+). There are two sets of somewhat conflicting data, plus some anecdotal reports. Perhaps you can each perform the same test and report your independent results? Rather than go back and forth, why not try to work together to test and document a new test. As it stands now I think the entry on the main page can be changed somewhat to:
  • Equipment with the effect of "Kick Attacks +" is generally believed to increase the rate of kicks by that many percentage (i.e. +5 returns a 5% increase), although there is some evidence to suggest that a lower percentage is returned.
  • Equipment with the effect of "Kick Attacks +" will grant the trait if the job wearing the equipment doesn't already have the trait.

--Gahoo 10:19, 26 April 2007 (EDT)


  • I think getting more data is definitely the way to resolve this (if possible by more people and not just me as getting results that agree with each other from multiple sources would be even more conclusive).
  • Otherwise if we just wanted to leave the entry as something like +5=+5% it would have to state that this was not based on any test, and it would be important to figure out who made such a claim first and why (and the same for the base kick rate), or state clearly that the origin of that belief is currently unknown and that it has been left mostly unchallenged and assumed true even in the absence of proof probably just because kick attack rates are hard to parse.
  • I can repeat my tests with maybe 2000 attacks (it's quite a lot of work but I guess it wouldn't be the end of the world) or even just again with 1000 attacks just to see how much variance I get for both the base and AF2, and I think aurikasura could repeat his tests maybe increasing his sample size to get up to 1000+ as well, but more importantly performing 2 sets of tests, one with AF2 legs and one without it, to get both values separately.
  • These would already give us an idea of the variance and if all results agreed fairly well with little variance (e.g., all showing near 15% base and say 2% or 3% kicks from melee hose) these would be pretty conclusive results (the odds of all those tests plus the previous ones all being off by ~+5% on the base rate and by -2-3% on the added rate from AF2 would be very low). If the data didn't agree we would still have some better ideas of the variability and range in the data which is always good and would help draw further conclusions (or figure out what other tests would be required to actually find the correct value). Then we wouldn't have to worry about where the previous ideas for values came from (other than as a historical curiosity) or have to rely on arguments by analogy.
  • Other people can do the test as well hopefully, but these in particular are a bit of a pain because normal parsers don't detect kick attacks (they are not labeled differently on the chatlog), so you have to keep your eyes opened the whole time to count the animations for kicks, or use additional strategies to mark them in the chat log (like my use of boreas cesti). If someone like divisortheory could modify his parser to detect them (since his parser reads things from memory instead of the log like I do) it would be easier for more people to perform these tests. This is a quick summary of my older results:
    • MeleeHose: Attacks: 1020, Kicks 104, kicks% 22.7%
    • B.Haidate: Attacks: 1074, Kicks 96, kicks% 19.6%
  • Also if anybody else can contribute more data, it will help to have a sample size of at least 500-1000 attack rounds (an attack round is 2 punches or 2 punches + 1 kick), and also it's very important to kill each monster with a WS, so that no partial attack round is generated (for example, you might have been about to punch,punch,kick but if the monster died after the first punch, you will never see the kick that was scheduled to occur). Removing any double-attack gear (brutal earring) or subjob (war) also avoids other problems in the results.

--Genome 06:51, 27 April 2007 (EDT)

  • Regarding the comments about my parser, even though it doesn't currently display the number of additional effect procs in a column, if you have any Sql database tools you can easily view the raw text of the parse and run a sql query SELET COUNT (*) FROM ChatLog WHERE Text LIKE %Additional Effect%. This will tell you exactly how many additional effects happened during the fight. If you don't have any Sql database tools to allow you to do this, take a large parse with around 2,000 attacks using boreas cesti, then send me the resulting file and I can tell you the answer. Make sure the button was selected to include raw chat when the parse is first created. --Divisortheory 23:38, 4 May 2007 (CDT)


I'm happy with the above resolution, though I doubt I will get time to generate new data.

To re-encode my old data matching Genome's format:
MeleeHose/1merit: Attacks: 650, kicks 56, kicks% 17.23

--Aurikasura 14:20, 30 April 2007 (EDT)


  • Here's more tests and data (methodology, gear and other details in my entry, http://genomeffxi.livejournal.com/15696.html )
    • Older (BHaidate)+5merits: Attacks: 1074, kicks 96, kicks% 19.63%
    • New (BHaidate)+5merits: Attacks: 1096, kicks 93, kicks% 18.54%
    • Older (MHose) +5merits: Attacks: 1020, kicks 104, kicks% 22.71%
    • New (MHose) +5merits: Attacks: 519, kicks 53, kicks% 22.75%
  • I also noticed that I was wearing AGI+25 on all my tests so I decided to split what I had planned as 1000 attacks MeleeHose tests in the 500 I show above (same conditions as I had used previously) and 500 more tests removing the +25 AGI, and to avoid adding other parameters based on my gear choice I took off completely body, hands, and headgear, and left them empty, then started another 500+ attacks and got:
    • New(-eq)(MHose) +5merits: Attacks: 515, kicks 39, kicks% 13.15%
  • Which could indicate there might be an effect there (although it would need a lot more testing to check this it's a pretty drastic change when all other repeats were so close to each other).
  • Anyways you can combine the data I have already into a big 2k sample, it would help if we had a lot more data so if anybody still has doubts go and generate some too (possibly with a method that marks kick attacks in the parser logs as I did, so it's easier to analyze and confirm data again anytime). (I'm still hoping Divisortheory can make a few additions to his parser to make it easier for anybody to do more kick tests =P)
  • On a side note, I think just to match and compare numbers, Aurikasura's kick values should be computed not as 17.23% (e.g. 56/(650/2))
    but as 18.8% (e.g. 56/((650-56)/2))
    (removing the kick attacks from the total attacks to get the # of single punches alone, and dividing by 2 to get the number of attack rounds, and then finding the ratio of kick events over attack round events)
    Or if the original number of attack rounds was correct, the total attacks converted in my same format for comparison would be 650+56=706 (I used attacks this way instead of attack rounds since it was easier to use in the context of parsers)
  • About the current kick attacks page wording, I still really think we need to add that we don't know where the original assumed values comes from, also because someone going to that page might read that and might be able to share more and shed some light on the origin of those values, or even reference some tests if they existed.

--Genome 04:14, 1 May 2007 (EDT)

I'm a PUP with Bahamut's hose. Since I don't naturally have kick attacks, would it help you if I were to post my rates? Since my KAs are only coming from items, it might help you settle this dispute. Ask and ye shall recieve. Tahngarthor 20:48, 4 May 2007 (CDT)


this is a bit trifling, but i suggest someone add a reminder that KA will not proc if you're not using hand-to-hand after "*Equipment with the effect of "Kick Attacks +" will grant the trait if the job wearing the equipment doesn't already have the trait." while that sentence is of course true, and the page already states that KA only happen if you're using h2h, the "wow other jobs get KA!" seems to be a source of confusion for many people (if topics and posts about it on forums are any indication). even more trifling, this is a bit odd:

Job Traits are always active. Further Notes: This trait is only active using hand-to-hand.

anyway, regardless of the proc rate of KA or how +X KA influences it, apparently some people need it to be very clearly stated that they can't, say, wear bahamut's hose on their DRG and expect to ever kick while using a weapon.Bowser

Damage bonus feet[edit]

I disagree with the statements that these items are like having a weapon with DMG:30 or DMG:whatever; if they did, my kick attacks would do more damage than regular attacks, and they dont. The damage does increase vs. not having the item equipped, but not as much as these pages would imply. Tahngarthor 12:43, 26 March 2007 (EDT)

The damage "rating" on these items is derived from capped critical hit values as tested on extremely low level monsters. It's fact, not speculation. --Aurikasura 14:18, 26 March 2007 (EDT)

To add to what he said, you can perform the same test by equipping said boots and killing Wild Rabbits for a while. After you see the same value enough times, you can be pretty confident that that value is the cap. Then, plug the cap that you determined empirically into the formula for hand to hand damage and work backwards, solving for D. The numbers you get should exactly match with the numbers on the main article. --Divisortheory 14:23, 26 March 2007 (EDT)

I still dont understand why the kick attack is never exceeding the damage of a normal hit when the kick attack weapon damage (+30) is so much higher than the weapon I do normal attacks with (+18). I would expect that on average, the kick attack hits would be higher than normal ones, so why am I not seeing this? Now I understand better how the damage is calculated; is it just because my attack is low (Where the damage of individual hits varies more) and the damage of individual hits is too random for me to get accurate calculations in a reasonable time? Tahngarthor 01:13, 27 March 2007 (EDT)

Either PUP doesn't gain a bonus from kick enhancing shoes or you're not calculating it right. My dune boots kicks exceed the damage of my spharai punches. Test on level 30ish monsters using attack food; the peak crit punch and peak crit kick damage should show that kicks have a higher base damage. --Aurikasura 14:07, 27 March 2007 (EDT)

It does, because the damage increases. It just isnt increasing as much as you would expect from a weapon damage that is almost double your actual weapon. I probably just don't have enough data. Tahngarthor 15:06, 27 March 2007 (EDT)

The DMG number on a H2H weapon is misleading. Keep in mind MNKs have a "base H2H" damage which involves their Category:Hand-to-Hand combat skill level. In particular, it's (your combat skill) * 0.11 + 3 + DMG you see on the weapon. In this case, you get:

  • Destroyers: 276*0.11 + 3 + 18 = 51.36
  • Dune Boots: 276*0.11 + 3 + 30 = 63.36

A bonus of about 23%

--Divisortheory 15:18, 27 March 2007 (EDT)

To reiterate on what people already mentioned, the damage rating from dune boots is equivalent to a h2h weapon with D+30. It's easy to see even on monsters that are just EP- to 75. Look for your highest critical hit damage, you will see it caps at 219 for kick attacks with dune boots if you have full h2h merits, while a punch with destroyers (D+18) will cap at 189. You can ask a monk to show you in PT as well to be convinced (a lower lvl monster will make it easier to get the highest critical sooner). You can use the H2H damage formula to get D+18 from 189 and D+30 from 219.

--Genome 16:58, 23 April 2007 (EDT)

Weapon skills[edit]

Can a kick attack happen during a WS? Tahngarthor 22:08, 2 April 2007 (EDT)

No --Aurikasura 02:03, 24 April 2007 (EDT)

I believe they can on Dragon Kick with Footwork active. It's easy to tell because you'll get ~27% TP return instead of the normal ~13%. --Emeryk 23:48, May 6, 2010 (UTC)

Moved to Damage[edit]

The 9/08 update altered the in-game description of the various Jen/Sen Hibaki, and they now have the trait "Increases Kick Attacks Damage". Moved them from the "Activation" part of the page to the "Damage" part of the page. --Baroness 14:33, 10 September 2008 (UTC)


If any of the restless mnks who complain about never getting an update will test out the new boots in comparison to others it would be quite nice. Im not quite high enough on mnk to parse the difference with higher boots to lower. it seems like kung fu boots outdmg junkenshi habaki but once again (lol) thats simply speculation --GodsBlackArm 20:02, 22 November 2008 (UTC)

This article uses material from the "Talk:Kick_Attacks" article on FFXIclopedia and is licensed under the CC-BY-SA License.