What's Better?

[Fottiti]

Regarding claiming it as mnk/rng, you won't last long regardless of chakra gear and it would be easy to keep 5-10 pieces of sam gear in satchel and swap it with chakra when running to change jobs. There really is no practical justification for going mnk over sam to ixion that I can think of assuming both jobs are equally geared (which is generally a given as they share so much gear).

[Dooom]

I have no movement speed on sam so i scan on mnk/rng.

No ASA legs? Sad to say, but I pretty much paid $10 for it for +movement for samurai... which I still haven't finished

[Motenten]

Motenten: the rounds per ws I have take into account all possible retrurns of asuran fists, and also take into account TP overflow, it is an exact formula similar to what you can find for 5 or 6 hit builds for 2handers. I did not simply divide the hits needed till 100 by the average number of hits per rounds... ( although this may look like it for 21 STP builds, but that's not the case, because you need 14 hits, not 15, starting from max asuran return).

Assuming this is in reference to:

with 95% acc, 15% kick, 5% DA, 21 STP during melee, 12 during asuran: 7.018 ( compared to his 9.09 that's +30% WS damage)

Ok, I did misinterpret that slightly. You need 14 hits on 8/8, 15 hits on most other returns.

Make a list of all sequences of DA and KA leading to 14 hits ( and 16 hits), there is quite a lot, then list the probability of each possibility ( its tedious, ive got it on a spread sheet); with 5% DA and 15% KA rate, I've got 7.41 rounds needed to get 16 hits, and 6.52 rounds needed to get 14 hits; That's a bit less than one round differences, as expected, since you have more than 2 hits per rounds on average.

Of course in practice it is more complicated as acc max is 95% so there can be infinite sequences, but the maths work out; and you do not just divide 6.52 by 0.95 to get the rounds, because you just never have the same number of hits per round, and that's what the alla guy is doing ( another approximation). For instance here, with 95% accuracy, and starting from a 8 hits asuran fists, with 95% acc, 15% KA, 5% DA, you need 7.78 rounds to get 16 hits, and 6.84 to get 14 hits. 7.78 is NOT 7.41/0.95, and 6.84 is NOT 6.52/0.95, the first digit generally coincide, but not the second, so it is a first order approximation (1% ish error).

I'm not entirely sure I can agree with the objection. Some simple points:

~~ 7.78 is NOT 7.41/0.95

7.41/0.95 = 7.8, which is 0.26% off from your target value, not 1%.

~~ 6.84 is NOT 6.52/0.95

6.52/0.95 = 6.863, which is 0.34% off; again nowhere near a 1% error.

Not to mention your own values are presumably rounded to the nearest 0.01.

You also admitted that you don't include the full sum of infinite sequences in your calculations. If you're truncating the sequence at some point, that's implicitly indicating that the 'true' average is going to be slightly higher than your result. Is there not the possibility that your equations drop the last 'insignificant' 0.25% to 0.33%?


And finally, your equations still don't account for the frequency of an additional round of attacks after reaching 100 TP, which can occur for a variety of reasons (mob 1 attack round away from dead, recasting Utsu, in the middle of adjusting position to line up for next mob, mob getting killed just before you weaponskill and you get an attack round in on next mob before you can use yours, macro lag, etc), and is more likely to happen as haste increases.

[Appie]

For mnk on ixion, would it be better to go full str build for DK or would it be wise to toss some acc in there to make sure the second hit lands?

Was looking to swap genbu for 25 ws acc anwig and osode for usu body

Edit:
Also, Dune boots vs Kyoshu Kyahan

This is what I used for DK on Ixion:

input /equip Head "Genbu's Kabuto";
input /equip Ammo "Bibiki Seashell";
input /equip Neck "Thunder Gorget";
input /equip Body "Kirin's Osode";
input /equip Hands "Alkyoneus's Brc.";
input /equip Ring1 "Rajas Ring";
input /equip Ring2 "Omicron Ring";
input /equip Ear1 "Brutal Earring";
input /equip Ear2 "Hollow Earring";
input /equip Back "Cerberus Mantle +1";
input /equip waist "Warwolf Belt";
input /equip Legs "Shura Haidate";
input /equip Feet "Dune Boots";
input /ws "Dragon Kick" <t>;

Haven't done Ixion in ages actually. The only reason Hollow is in there is because I have no Triumph and I didn't care enough to just get one for Ixion. I honestly don't remember getting upset about missing a second hit, so that either means it happened rarely or I was happy enough with my 2k dmg DK (think I have peaks to 2500 somewhere in my screenshots, that was a Brutal Earring proc iirc).

[noodles355]

mdkuser thinks rounds/100TP 1 / rounds/100TP 2 = increase in WS frequency.
This, with dual wielding multi-hit WS jobs is complete trash ans should be ignored.

Sadly, he has never seen the reasoning why he is an idiot because he put me on his ignore list as soon as I questioned his pchan methods

[Cream Soda]

Regarding claiming it as mnk/rng, you won't last long regardless of chakra gear and it would be easy to keep 5-10 pieces of sam gear in satchel and swap it with chakra when running to change jobs. There really is no practical justification for going mnk over sam to ixion that I can think of assuming both jobs are equally geared (which is generally a given as they share so much gear).

Talk to them about it. I'd prefer to stay in town and wait on pop, lol.

[Cream Soda]

No ASA legs? Sad to say, but I pretty much paid $10 for it for +movement for samurai... which I still haven't finished

negative

[mdkuser]

Assuming this is in reference to:



Ok, I did misinterpret that slightly. You need 14 hits on 8/8, 15 hits on most other returns.



I'm not entirely sure I can agree with the objection. Some simple points:

~~ 7.78 is NOT 7.41/0.95

7.41/0.95 = 7.8, which is 0.26% off from your target value, not 1%.

~~ 6.84 is NOT 6.52/0.95

6.52/0.95 = 6.863, which is 0.34% off; again nowhere near a 1% error.

I dont mean to calculate the exact error, just the approximate error. I say 1% because its not 0.1% nor 10%.

Not to mention your own values are presumably rounded to the nearest 0.01.

You also admitted that you don't include the full sum of infinite sequences in your calculations. If you're truncating the sequence at some point, that's implicitly indicating that the 'true' average is going to be slightly higher than your result. Is there not the possibility that your equations drop the last 'insignificant' 0.25% to 0.33%?

I take all series into account, but i think its a good approx to just divide by 0.95.

And finally, your equations still don't account for the frequency of an additional round of attacks after reaching 100 TP, which can occur for a variety of reasons (mob 1 attack round away from dead, recasting Utsu, in the middle of adjusting position to line up for next mob, mob getting killed just before you weaponskill and you get an attack round in on next mob before you can use yours, macro lag, etc), and is more likely to happen as haste increases.

I aggree. However this is something you consider after you think the model is correct. Also bosst delay and round excess thing favor destroyers more than footwork, because footwork gets TP faster. I'm not going to start mathing destroyers versus footwork for an afk monk.


Edit: also Im not trying to bash footwork build, I think it has potential. right now im trying to find the best footwork build and I think the build i gave can be improved. for example dune boots will be better than KKs if you swap your accuracy gear around ( usu body versus osode for example), I also need to include all possible TP returns from asuran fists in the model ( I only considered max tp returns, that lead to 5 hits till 100TP), some of the possible returns lead to 6 hit builds, which favors destroyer build even more. But mainly, the relation ship between OAT, DA and KA needs intense testings.

[Motenten]

Also bosst delay and round excess thing favor destroyers more than footwork, because footwork gets TP faster.

And that's why they need to be taken into account.

However this is something you consider after you think the model is correct.

Perhaps, but until you include it you'll end up with notably skewed results. Worse, if you ignore it while others account for it you end up with wildly divergent results, such as your +40% damage total compared to a corrected +17% total.

I'm not going to start mathing destroyers versus footwork for an afk monk.

I'm not sure why you would. No one else has thus far.

Edit: also Im not trying to bash footwork build, I think it has potential. right now im trying to find the best footwork build and I think the build i gave can be improved.

The main issue to be aware of here is the target. Everything calculated so far has shown that Footwork gets worse (relative to standard H2H) the stronger the player's gear when run against a static target. Vs lvl 82 Greater Colibri, full Usukane will be better off using standard h2h; slightly lesser gear has them basicly on par with each other.

The more interesting consideration is that as targets get more difficult, Footwork gets -better- relative to standard H2H. As such the question is not so much "can I make a Footwork build better than standard H2H for colibri meriting?", but "at what point does Footwork start surpassing standard H2H for the gear that I'm using?" With the d17 Wargs, that threshhold seems to be drawing closer to merit-level mobs, which implies a corresponding improvement on higher level NMs and such.

[CDF]

I never understood this trivial objection about failure to account for "wasting" attack rounds beyond achieving 100 TP and various delays because no one ever came close to accounting for it anyway! Take a stand, argue that x% of the time you waste y (y = 1, 2...) attack rounds, add y to the ideal average number of rounds which can be converted to average time, then add whatever delay exists, then do a weighted average.

As for OAT, KA, and DA, I suppose the truth lies somewhere between jailer/virtue weapons and Ridill, but there is no real way to tell because getting only 2 attacks per round (for whatever reason that is impossible to know) means many possibilities reduce to the "degenerate" 1 - (1 - DA)(1 - KA)(1-OAT) effective double attack rate. So, whatever effective rate you get is correct more because of degeneracy than the actual validity of the assumption.

I have seen it claimed that DA has "priority" over KA for normal H2H, which would be easy enough to observe except I don't really take those who assert this with an appeal to common knowledge at their word.

[shepardG]

This has come up in LS....

What's Better for Zerging

Drk/drg or Drk/nin

Same gear, buff's, etc.

[ZhiGaruda]

This has come up in LS....

What's Better for Zerging

Drk/drg or Drk/nin

Same gear, buff's, etc.

Ask yourself these questions:

Is the mob stunnable? Yes? DRK/DRG. No? DRK/NIN
Can you rely on your stunners? Yes? DRK/DRG. No? DRK/NIN
Are we 100% positive this mob will die in ~30 seconds? Yes? DRK/DRG. No? DRK/NIN

If you come up with a different answer for each question, always fallback on DRK/NIN. It's safer and slower, but the best option if your linkshell doesn't work well together.

[Clash Perez]

For SA+Cannonball:

Relic Head (DEF: 24, VIT+3, Skill+5)

or

Anwig Salade (DEF: 21, STR+4, Crit damage +2%)?


Edit: Full Blue Magic merits

[Motenten]

But mainly, the relation ship between OAT, DA and KA needs intense testings.

While I agree in principal, I suspect we're going to have to live with assumptions here.

Starting with my own: 14.5% KA and 5% DA

If they are additive, that's a 19.5% proc rate.

If they are exclusive, that's a 18.775% proc rate.

To distinguish between them, you need two sample sets with a 0.36% margin of error. To reach that you need two sample sets with 50,000 samples -each-.

At 22% haste that's 3.6 days of continuous testing. Unfortunately you can't use the -1 fStr trick with kicks, which means fortifications are probably your best option. That in turn means probably 5 days of testing for each configuration, or a total of ~240 hours of doing nothing but kicking walls.

[Kyte]

For SA+Cannonball:

Relic Head (DEF: 24, VIT+3, Skill+5)

or

Anwig Salade (DEF: 21, STR+4, Crit damage +2%)?


Edit: Full Blue Magic merits

Anwig

[Darkmagi]

If you have any job that isn't blu I wouldn't waste that hat for a small CB damage increase though. Er sorry if that came off as the standard lolblu comment. More so that other jobs would benefit much more from the hat.

[Taint]

This has come up in LS....

What's Better for Zerging

Drk/drg or Drk/nin

Same gear, buff's, etc.

DRK/DRG if no crazy AoE is going to kill you. The jumps not only add damage but the hate shed can be very handy to turn the mob if it looks your way and 5% haste is a large boost with haste and marches.

[CDF]

While I agree in principal, I suspect we're going to have to live with assumptions here.

Starting with my own: 14.5% KA and 5% DA

If they are additive, that's a 19.5% proc rate.

If they are exclusive, that's a 18.775% proc rate.

To distinguish between them, you need two sample sets with a 0.36% margin of error. To reach that you need two sample sets with 50,000 samples -each-.

Has this been checked for H2H? Suppose the following (let me know if any of these are invalid from experience):

1) Double attack can occur on either or both hands. (Counter example: lack of observed 4-attack rounds.)

2) Given 15% DA rate, then the observed rate of no double attacks is 72.25% (.85*.85).

3) With 5 kick attack merits, the KA rate is 17.5%. We want to know if the "effective" KA rate is different from the nominal rate in the presence of DA

We are interested in the idea that if a double attack occurs on any hand, a kick attack can't occur, and if a double attack does not occur on either hand, a kick attack can occur. (Counter example: evidence of a 5-attack round.) Shorthand: "DA > KA"


Now, only one hypothesis is true. Either DA and KA are exclusive (directionally) or something else, so you don't need to obtain two samples. I already stated the proposed direction, so the expected kick rate in the presence of 15% DA will be 12.64375% (.7225*.175), a bit lower than 17.5%. If you have the patience to count kick attacks, you want to see if your observed rate is significantly lower than 17.5%.

For a one-sided test of one proportion (null proportion .175), if "DA > KA" is true, then to achieve ~90% power to reject the null, you need n = 469 attack rounds. If you actually reject, then "DA > KA" has some traction. If you fail to reject, then you would have to explain it.

Yes, it's tedious to have to examine kicks manually, so alternatively just consider this problem in terms of "composite" multi-hit rate (effective double attack rate ignoring the distinction between kicks and punches). Then the null proportion would be .4525 (.2775 + .175) and the alternative would be .4039375 (.2775+.7225*.175, still "DA > KA"). The necessary sample size is then n = 889 (for ~90% power).

I haven't examined whether KA occurs when DA occurs so I write this from a position of ignorance, but if everything above holds you hardly need the results of 10,000+ attack rounds to tell the difference.

[Kyte]

If you have any job that isn't blu I wouldn't waste that hat for a small CB damage increase though. Er sorry if that came off as the standard lolblu comment. More so that other jobs would benefit much more from the hat.

Well, while I personally didn't make my hat that way, it's still a valid choice because there are still other uses for it beyond CB. You're not *just* getting +4 STR +2% critical damage for CB; you're still getting the WS acc +15 and +4 DEX.

[Darkmagi]

Eh true. Still decent for a WS head on a lot of those jobs.