Dexterity's impact on critical hits

[CDF]

On tiering, which is your main concern, yes that is a possibility but depends how the flooring actually manifests in terms of the actual crit rates. Actually, who ever showed hit rate takes non-integer values? Departures from piecewise linearity (two ranges) in the lower range may make the flooring explanation more viable. I will take a wait and see approach depending on whether you get more data.

---

As for the other stuff (don't have to read further) on allocating sample size, actually, in my last post I was first replying to Shamaya's comments. I guess I will just be repeating myself but with further clarification.

I am just saying in theory, allocating a fixed number of trials (total hits) between two levels spaced as far apart as desirable for the relevant range of predictor variable is the most optimal way to estimate the slope of the line. (What is "relevant 'is based on the research question.) Conceptually, this should make sense because the possible values of the slope are further constrained the further apart the two levels are. And if you consider three or more levels w/ the same number of total trials, each of the 3+ pointwise estimates will be less precise than the two pointwise estimates.

In practice this is bad because you also need to check the assumption for linearity, unless you know for a fact that the relationship is linear (like a function), which is almost never the case in reality. There is a trade-off between maximizing the precision of your slope estimated (minimizing standard error) and checking the linearity assumption based on deviations of your observed data from the estimated slope. No one should be impressed by you "fitting" your regression line straight through two points.

Key words are "in theory" and "fixed." In theory doesn't apply and neither does fixed.

[Deadgye]

Just to make a quick comment, does anybody know the AGI values for the goldfish in sea? I used them to do my test for homam corazza to prove it was only a 1%TA increase because I was able to start a parser, attack one, and then go afk for a day or two while my character wacked away doing 0 damage.

I'd recommend goldfish for anything where you need a large sample size where you can also test while doing 0 damage.

[krish]

Just to make a quick comment, does anybody know the AGI values for the goldfish in sea?
...
I'd recommend goldfish for anything where you need a large sample size where you can also test while doing 0 damage.

That's a great idea! I might have to look at a zero-damage build again just for doing this.

[Shamaya]

Hey Deadgye. I never looked at your actual Corazza testing. I agree that that is a great concept/method. But how did you produce 0 dmg btw? I'm drawing a blank on this. Level1 dagger and -attack gear is enough? As for Agi; as you probably well know, the values are unknown for most mobs in the game. I'd bet goldfish are no exception. I can't remember offhand the method in which one could figure out the job/family factors which determine attribute progression by mob level. At least, if we were sure that crit% cap was based on a certain amount of dDex (50-51?), we could work backwards and find mobAGI, albeit painfully.
Been doing some sampling but it's slow. Work, lethargy, people stealing my EM crabs.. etc. Don't wanna see this effort go into the dark again, though.

[Masamune]

Wow excellent idea Deadgye !

and quite a coincidence that i went back to this topic, for something related to AGI determination and a new camp i just discovered (tough less interesting than Goldfishes): Colibris Bhaflau.

Lil story short: i was testing their stats using correlation between the known formulas and checks, to obtain perfect matching written @ wiki Talk:Colibri - FFXIclopedia, the Final Fantasy XI wiki - Characters, items, jobs, and more
those calculations/tests shown all 3 levels of Colis available have SAME agi=60 (and vit=60 also) which is quite interesting in terms of parsing (wipes the risk of mixing different lvls of mob and thus agis).

If Deadgye can get the /check reports for def/evasion of those goldfishes, you can report them here, so we can use those calculations to get the missing AGI.
BUT those calculations needs also RaceRanks for VIT/def and AGI/eva, which are only obtainable only through checks for AT LEAST 2 levels of same mob. After that, it's just a play of matching with formulas i have already Excel'd.

EDIT: question Deadgye: even crits were doing 0dmg o.O?

[Massimo]

Hey Deadgye. I never looked at your actual Corazza testing. I agree that that is a great concept/method. But how did you produce 0 dmg btw? I'm drawing a blank on this. Level1 dagger and -attack gear is enough? As for Agi; as you probably well know, the values are unknown for most mobs in the game. I'd bet goldfish are no exception. I can't remember offhand the method in which one could figure out the job/family factors which determine attribute progression by mob level. At least, if we were sure that crit% cap was based on a certain amount of dDex (50-51?), we could work backwards and find mobAGI, albeit painfully.
Been doing some sampling but it's slow. Work, lethargy, people stealing my EM crabs.. etc. Don't wanna see this effort go into the dark again, though.

I'm pretty sure he used that event Great Katana with 1 base damage, all jobs. Dagger hit too fast, as the parser took some of his normal hits as Triple Attacks, or some shit.

So I guess he had 0 skill in GK, which means he should be doing 0 damage.

Edit: The Pitchfork/Trick Staff you can get from the current event should work, I guess.

[Masamune]

Back from Sea and testing on Goldfishes...

Pretty disappointed at the idea for several reasons:
- Never managed to hit for 0 consistently on any job combinations, always got at least 1hit for 1dmg, and that's enough to make the fish aggro and rape yur ass.
- Severely lacking opportunities for gears swapping regarding different dDEX values testings, since gotta load up on accuracy like mad when using a weapon with 0skill.

Best case i managed to build was BLU75/SCH1 with 54attack, loaded all spells with DEX+stat + accubonus trait, Ohat/MirageBody/Swift/sniperX2/Peacock/Homamhandslegsfeet along the LotusGK and PoleGrip/BrutalEar. i had like undercapped accuracy (20%?) till my 6th hit got 1dmg, then the fish opend its mouth and raped my ass before i could zone lawl < <
Edit: mob was a Ul'Hpemde, just around the GrdPalace entrance.

Mind develop how you did it Deadgye? Maybe you did that on Om'Hpemde ?

EDIT2: nvm just did Round#2 on Om'Hpemde: critical for 2dmg lawl...Pretty sure now Deadgye didnot use LotusGK or any 2handed weapon, pDIFs are too high still. Maybe TraineeBurin(Katana) ?

[Massimo]

Back from Sea and testing on Goldfishes...

Pretty disappointed at the idea for several reasons:
- Never managed to hit for 0 consistently on any job combinations, always got at least 1hit for 1dmg, and that's enough to make the fish aggro and rape yur ass.
- Severely lacking opportunities for gears swapping regarding different dDEX values testings, since gotta load up on accuracy like mad when using a weapon with 0skill.

Best case i managed to build was BLU75/SCH1 with 54attack, loaded all spells with DEX+stat + accubonus trait, Ohat/MirageBody/Swift/sniperX2/Peacock/Homamhandslegsfeet along the LotusGK and PoleGrip/BrutalEar. i had like undercapped accuracy (20%?) till my 6th hit got 1dmg, then the fish opend its mouth and raped my ass before i could zone lawl < <
Edit: mob was a Ul'Hpemde, just around the GrdPalace entrance.

Mind develop how you did it Deadgye? Maybe you did that on Om'Hpemde ?

EDIT2: nvm just did Round#2 on Om'Hpemde: critical for 2dmg lawl...Pretty sure now Deadgye didnot use LotusGK or any 2handed weapon, pDIFs are too high still. Maybe TraineeBurin(Katana) ?

Yeah.. I just revisited the thread, and it apparently wasn't the same guy.
What I was thinking of, is Kroot using Katana/Pitchfork as WHM/BLM, to test Brutal.

My mistake, same concept though.

[Shamaya]

Indulge us DG!

[Byrthnoth]

Didja remember Aiming bracelets?

[Deadgye]

It has nothing to do with attack, because then crits would do damage. You NEED a DMG:1 weapon, and then you need your str to be below the mobs vit so that you an fstr of -1 causing your damage for every hit to be 0. That way no matter what you will hit for 0.

I believe a str somewhere around 60-70 is the highest you can be before you start to hit >0 on them. The test was a while ago though so I can't be sure. I'd recommend using a job with low str and high dex or a job that can equip a lot of dex gear without str gear. Getting rid of your sub job helps a little bit imo. I'm pretty sure my thf/nothing should be able to hit a really high dex value (120-130) without going over a fstr of -1, but my account is unactive right now until I get myself a little bit more money to reactivate it.

I originally started my test with my charm wand, and it was going fine until i realized that the attack rounds were to fast for the parser to correctly register the triple attacks. I switched over to my Treat Staff II and found that I attacked too slow to hold the goldfishes attention. By adding on a little bit of haste gear I was able to hit a nice spot where the attack round was slow enough to register multi-attacks yet fast enough to hold the goldfishes attention. As we are just measuring critical hits vs normal hits we can use the club or a faster DMG:1 weapon like a ceremonial dagger if we want.

If I reactivate my account I'll be sure to help test. With a ceremonial dagger on thf/nothing I will have DMG:0, plenty of acc, 53 str, and a dex of 70+46. Since my str is that low.. I can add on a rajas ring and warwolf belt to boost dex another 10 and my fstr should still be -1 I think. (Ceremonial Dagger, Lightning Bow+1, Ass Bonnet, Love Torque, Hollow Earring, Skadi's Cuirrie, Rogue's Armlet's+1, Thunder Ring, Ass cape, Heca subligar, Rogue's Poulaine's.) I can also just go /nin and have 62str and 80 base dex.. there are lot of mix-and-match options available.

[Shamaya]

Ahh, I see. I have to admit, that is fairly clever. Right, so if you have a net base damage higher than 0, and even if your pDif avg is 0, your crit will hit for more than 0. So a basedmg1 weapon. I don't know how you got your parser to register triple attacks, but gj. I suppose if Masa's proposed evasion check on diff level mobs does reveal their agi, this would be worth it. Btw I haven't seen anyone remove their sub job for a long time. That even still possible? Guess you've gotta be really fast.

[Masamune]

Thank you Deadgye, makes lot more sense now (i'm elvaan 5str merited...), also i gotta thank you very much for something else:
Maybe you didnot know while posting this, but you also provided an interesting alternative way to check a mob's VIT other than /checks !! seriously that's ahnds down more valuable than this whole crit% test XD.

On mob's agi topic, i'm on a series of /checks atm on attempting to fill the "Family table" JPs left unfinished (with the letters A to E). So far i did Flamingos, Colibris, Tragopans and Orderly Imps (wiki discussion pages updated), and continuing with Marsh Murres, Jnuns, Skimmers and most mobs of theBoyadaTree. I'm also experiencing a non matching problem inside bird family between Murres & Flamingos ;;
I can do a quick detour @ sea to update Hpemde stats also ^^

EDIT: just went check Hpemdes, check results & stats calculations @ http://wiki.ffxiclopedia.org/wiki/Talk:Ul%27hpemde, and results are:

Perfect match for job=DRG/drg and:
RaceVITrank= D
RaceDEFrank= C
RaceAGIrank= D
RaceEVArank= C

- Ul'Hpemde lvl72:
VIT= 75
AGI= 67
DEF= 292
EVA= 280

- Ul'Hpemde lvl71:
VIT=75
AGI=67
DEF=287
EVA=275

- Ul'Hpemde lvl70:
VIT=73
AGI=67
DEF=282
EVA=271

- Ul'Hpemde Lvl69:
VIT= 72
AGI= 65
DEF= 277
EVA= 265

- Ul'Hpemde Lvl68:
VIT= 71
AGI= 65
DEF= 271
EVA= 260

Note: i don't know if their stats change when they go under that berserk mode.
EDIT: checked LowDef@367 => lvl72 has indeed floor((367-1)/1.25)=292DEF

[Enedin]

Note: i don't know if their stats change when they go under that berserk mode.

I doubt that, all damage is just doubled, including magic.

[Masamune]

Sorry, doubts arenot allowed when testing anything.

btw i'm gonna test Deadgye fstr= -1 method on Fortifications one of these days, and on those Hpemede whose vit just been determined: if that method work, might be able to calculate HNMs and other NMs 's vit. /em cross fingers.

EDIT: meh that's offtopic, my bad, forget this post < <

[Siniroth]

I doubt that, all damage is just doubled, including magic.

It's not necessarily that simple, they do more damage, and attack much slower, when they're in that mode, but I don't see relevant stats changing.

[Masamune]

It's not necessarily that simple, they do more damage, and attack much slower, when they're in that mode, but I don't see relevant stats changing.

semi-/nod to Siniroth, i was thinking same as you, even further with taking euvhis flowers as example: they change litterally jobs between closed/opened "modes", resp. thf/drk.
So only check tests when Hpemdes have mouth opened can prove if yes or not they change job from drg/drg to some other combination under that "mode", which means obviously stats changed.

[Mojo]

I'm pretty sure that they don't open their mouths unless you damage them, so if you're talking about 0 damage hits here then you shouldn't have a problem.

[Masamune]

good point mojo, didnot pay attention if they open mouth at exactly 30 seconds only, no matter dmg yu did; or at a set amount of dmg done.
But indeed im also pretty sure they close their mouth at a fixed amount of dmg.
(sound so like Starr dont yu think lol?)

[Siniroth]

Open at 30 seconds after taking initial damage, I figured it out once but forget what damage is needed to force it to close.