Final Fantasy General DXCIX - /ffg/

>tfw the Marvel/Squenix crossover has ended.
It's something I never knew I wanted. Anyways how's the TMR farming going? Nearly done with TT, then time to throw another Sozhe into the grinder.

...

12th for Rogue

BEST MOUSE IN THE HOUSE!

Just reset the app you sperg

furfags must die

?
You don't mean reroll, right?

>how do I interpret data

You can s/l large tumult amounts.

Anyone get any Papalymo TM's yet? I'm barely finishing one. I really hope I get more lucky with pulls, I want at least 3

I mean reset the app so you don't get 5 tumults.

Shit, really? I thought RNG was rolled at the start of a turn, so resetting doesn't do anything.

I don't know you tell me.

Be careful, though, as S/Ling can fuck with the game a bit, causing buffs or debuffs to fall off, causing you to gain or lose mana/SB gauge, and such.

来た
youtube.com/watch?v=bUkhDTVAOko

Rat in the house?

The damage seed is stored per device, but enemy AI refreshes each time it's run.

Incorrect, if there's anything it's just visual bugs.

So:
>Locke increases chances of all drops but Giancrysts
>Xon increases chances of only Giancrysts
Is n large enough to draw those conclusions, especially for Giancrysts?

>incorrect
Oh, well. I do know at the least it resets Olive's Mortar ability so it can be used multiple times a fight.

Sample size.
That's not a significant enough difference for such a small sample size.

Learn how to stats.

Ah, so how much damage each attack would do is saved, but which attacks they do isn't?
Odd. I guess it's so they can react.
Thanks, though.

Well no shit, it resets the turn counter to 0.
So timed attacks (both yours and the enemy's) are irrelevant.
You can use this to your advantage to avoid Snort on titan right now.

It's the only way we beat maxwell

No, enemy AI uses a different seed or something.

When it's so clear that having Locke will increase the low rank drops and Xon the rare drops, just like their abilities tell us, it's fair to say that almost 2 000 runs is enough to say that they work, even if we don't know the exact value.

But hey if you still want to doubt and act smug, be my guest.

>almost 2000 runs
Only roughly 1k runs. None has ~500 runs, which is pretty low especially for the rarer Giancrysts, and Locke has around 450 - again, rather low.
Xon's got 750, though, so his data's getting pretty accurate by now.
We can probably trust the drop rate for T1 and maybe T2 mats, T3 if you're feeling risky, but Giants almost certainly have too low a sample size.

>it's fair to say that almost 2 000 runs is enough to say that they work
It's not even 500 runs for Locke, dumb-dumb. Even the guy making the spreadsheet says the sample size is still too small.

I'm really not expecting the results to change, are you?

why don't you take a gander in the cell directly to the right of the image you just posted, champ.
The drop rate is insanely low, some single digit number.

For the T4 mats? It's definitely a possibility.

Wow thank you I missed that. Maybe zoom a bit more.

For T1 and T2? Not really. T4? Yes, especially since Locke's number's lower than baseline. That seems odd, and I'd rather not specifically avoid using him when farming Giants if it turns out I'm dealing with wrong numbers.

I'm not really interested in the drop rate, just knowing that having Locke or Xon will increase it is enough. Time to pump those cactuars in

>post yfw Noctis was confirmed for Dissidia
I hope Hot Noctis or El Nocto are alts.

T4 is so rare that it's to be expected that the number isn't right. As long as it's around the same as the base number it should be the same.
Now Xon is slightly higher which could also be within the margin of error, but if Locke works for the lower tiers I don't know why Xon wouldn't for the higher ones.

Yes, but is it a significant enough change to justify throwing cactuars at Locke or Xon? Especially Xon - I'm not leveling a 4* to an 88 6* just for an infinitesimal increase.

Any increase is good when they're so rare. If you're upgrading abilities for multiple characters it could potentially save you hundreds of energy.

And it's not like Xon is a bad character on his own, he's got interesting skills so you could use him in niche situations.
On the other hand Locke is pretty fucking useless and a waste of mats.

but Xon's buff steal is pretty amazing
Also how are you not overrun by cactaurs from the raid event

Assuming Locke does boost the rates of all the lower mats, I'd say absolutely.
We also don't know what kind of effect bringing both or multiples along has, so if enough testing can be done to possibly confirm that it could be even more worth it. But as it is (assuming these stats are the real ones but Locke does not lower T4 rates), Locke is definitely worth it and it's up to you with Xon.

There is no reason to assume Xon's droprate is significantly larger than 0.02 given the sample size.
The probability of that happening with this many trials is P=.2326
That isn't even remotely significant at alpha=0.05. There's about a 1 in 4 chance of these values just happening due to chance, which is extremely high.

Apparently you did miss it.

Except that's not how statistics works you raging homo. We have no reason to believe that, it's not statistically significant.

>tfw already have a L100 Xon
feels good man

Oh shit, I haven't been pulling at all. I've probably got 200k at this point.
But yeah, his AoE 100% steal is pretty nifty, and I hope future events feature stealing currency. Now if only Chamber of Gems could have crysts *stolen...*

On that note, is there a list of things that can be stolen and from which monsters?

what about more than one Xon? I actually have 4 of them

>t1 and t2 jump from 1.1 and 1.6 to more than 1.3 and 1.7 with Locke
>it's not statistically significant!

Yeah it's probably just a fluke that it works as intended when you use that one exact character.

the more Xons, the merrier

>getting into the final stretch of classic realm dungeons
>every dungeon consists of a single battle
>half stamina cost and 120+ total stamina

Well, I guess I better start juggling elite dungeons if I want to optimize my stamina.

I don't think we have any data that the effect stacks additively, multiplicatively, or at all.
Plus, I only have one Xon.

Anyone else getting constant crashes lately on RK? It's becoming fucking annoying that I either get a random crash, or something like pic related before actually crashing.

Are you mentally handicapped?
This isn't up for debate, it's mathematically insignificant. We can't reject the hypothesis that p(xon) = p(none) given this data. P is bigger than 0.05.

Stats aren't about feelings. You can't poll six times and reject the null just based on feelings. Show me the math to back it up if you disagree, I've shown you mine.

So now you're saying only Xon maybe probably possibly doesn't work.
How about you add hundreds of runs into the doc if you want to prove me wrong that much? I'll welcome the initiative.

Similiar happens to me when I tab out just before the battle loads. Otherwise no crashes.

He mentioned Locke, you mentioned Xon. Not the best strategy.

>maybe probably possibly
That's literally how statistics works. Unless someone actually somehow datamines what code is running for droprates, we can only get "we believe it's likely that there's a connection" or "we cannot confidently say it's likely that there's a connection."
Also,
>the data can't support either of our claims yet, so until it can, I'm obviously right!
fuck right off.

I am saying
listen
that we cannot reject the hypothesis that xon has the same droprate as no xon given the data. It isn't statistically possible.

That's all I've said this entire time you mouthbreathing sack of shit.
There's nothing to prove here, the data can be analyzed objectively. I know you never took stats, but please for the love of god read a book nigger.

I've been talking about Xon the whole time, he's the one who replied to me about Locke.

>English announcement trailer still has japanese VO

Should we be worried?

the data is enough proof for me to believe it does affect drop rates.

haters gonna hate. I'm gonna use my xon when farming shit.

I figured that more or less confirmed nip voices only. I think you should go beyond worrying and into accepting.

Like I said, because Locke works, Xon should work.
Unless they intentionally bugged this specific skill from Xon for no reason at all.

You want to say it doesn't? Fine, bring more runs and I'll listen. The burden of proof is on you, you should say why something doesn't work. It's not up to me to prove that the game isn't broken.

That isn't how fucking proof works nigger. You can't just reject the null hypothesis until proven otherwise, you have to do the opposite. The burden of proof is actually on YOU to show that it is significantly different from h0.

That's literally the scientific method. Anything else isn't science, it's feelings.

>The burden of proof is on you
Not either of those guys, but I'm not seeing why that is. You're just taking something at face value while they're saying that it's not something that should be taken as fact.

it's pretty much confirmed

He thinks it's fact because big brother gumi told him so.

>he's the one who replied to me about Locke
Shit, you're right. Sorry, there's like 3 different arguments going on right now.

Same here, but only because I've already got one and can stand losing one spot on my Trust team.

>because Locke works, Xon should work.
That's terrible reasoning.
>Unless they intentionally bugged this specific skill
Okay, so in addition to not knowing how statistics works, you don't know how programming works.
>The burden of proof is on you
That is literally factually incorrect. The default assumption is the null hypothesis, that P(Xon) = P(Locke) = P(Neither). Especially since it's a mechanic that doesn't exist outside these units and, if we prematurely assume the data is representative, is very small. If the difference between "Xon gives more T4" and "Xon ain't doing shit" is 1 Giancryst every 500 runs, it's extremely, extremely hard to show that that's actually happening rather than just being a statistical fluke.
You absolute clod.

fucking absurd
gumi's done bug patches before, quite a while after they were noticed
if we have a bug that hasn't even been noticed yet, it might take gumi another 3 months before even acknowledging it's a thing

>Boosts the rate

sounds like multiplicatively. sorta like how event pulls " Boost" the currency drops

>Trance Terra
absolute madman

I'm probably gonna dump my stored moogles into Setzer to try out the fixed dice shenanigans, should be done pretty soon honestly. Excited to see those big dick numbers

...Event pulls stack additively. If you take a 50% unit and a 50% unit, you get a total of 200% base currency, not 225%.

they did the runs a total of 1779 times. I'd say that's good enough to confirm they do affect drop rates.

Are you actually expecting me to run the Vortex thousands of times just to prove you wrong? Any indication that it works should be enough, don't come and say it's not scientific, I know.

But if you care that much about needing the exact data before saying anything, go do it yourself, I'll be happy with my t4 drops.

remember, we need to be more like lady Ovelia

and how did that work out for her in the end?

Firstly, they only did 740 xon runs. That number also contains the locke and none data.

Secondly, statistics isn't about what you say. It's rigorous and objective, it isn't "hmmm well I feel like that's good enough." You can see the exact odds of these numbers happening due to chance, and they're crazy high. About 1 in 4.
The typical cutoff of scientific rigor is about 1 in 20, for comparison's sake.

So this data says absolutely nothing (yet). You can't say with any certainty that xon affects droprates.

There is no indication at all that it works is what I am trying to say. You just don't god damn get it.
These numbers could have easily happened just due to chance assuming the droprate is the same as "none."

You can't draw conclusions yet you spergmongling, turdfrothing, shitspackling retard.

If I flip a penny 197 times and a quarter 3 times, I can't claim a quarter has twice the chance of landing on heads as a penny just because I did a total of 200 trials.
I believe Locke affects low-tier drop rates, because of the higher ratio for them meaning each trial impacts the ratio less. I don't think we can say firmly that Xon's thing affects Giancrysts yet, because they're so rare that a single drop changes the rate rather strongly.

No, we're expecting you to admit that it's not proven that Xon functions as you think. You're free to believe whatever nonsense you like, just don't claim your nonsense is fact because you don't know how statistics works.
I'll be running with Xon because he's already max level for me and I can stand losing a slot on my Trust team, but it's still SUPERSTITION.

>it isn't "hmmm well I feel like that's good enough"
Isn't that technically confidence intervals? Yes, it's standard to use 0.05, but someone might, individually, decide that they'd personally be swayed by a 0.2 interval.

Ever since the latest update to Samsung S6 my FFRK experience has been ruined. For some stupid reason the screen scales weirdly and the taps are very inprecise

I am a faggot for asking this in a general but does anyone know how to fix this??

Well yeah, but I mean the actual P value itself isn't feelings. Sure there's lots of different models you could use, but they're all objective.
There are physics papers where they need to go out to like, 8 decimal places for it to be considered significant. Psychology only needs 0.1 sometimes (because it's bullshit). Never will you find a scientific paper where 0.25 is good enough.

This guy is straight up ignoring the stats in favor of his feelings at this point.

...

>does anyone know how to fix this??
uninstall RK and play BE instead

>tfw you unintentionally got gud relics for ovelia but you don't want to replace yuna for her as healer as yuna is superior waifu

Guys any good Xiao builds? I'm leveling up mine and Sakura for the trial. I've also noticed that she gets high attack even without weapons.

i switch to the pixel and now ffrk is literally unplayable on my phone. shit freezes constantly, multiple second long hangups, some crazy shit. I pretty much only play on nox now

I actually like Spiritus's design even if he's a less cool Jecht ripoff of Jecht and Auron. Just let Naora do the designs while Nomura draws them from now on.

delita this

If I recall it correctly even a NPC knuckle will grant you more ATK than unarmed so it's a wasted ability on every unit who has it. Also you can't dual wield with it.

>rotating "pixels"
Ew.

>mathfags arguing that their mobage data isn't like their physics paper

I'm arguing that it isn't even like a bullshit psychology paper.
And that says nothing about the data itself. We could do 10000 more trials and have the same P value, it wouldn't change anything.

That's the thing about stats, you can't ever prove a null hypothesis. You can only ever reject one. So unless there's data that rejects it with some degree of certainty, there's no reason to believe there's an effect at all.
It's all about falsification. Until you can confidently falsify the claim that there is no effect, you can't draw any conclusions at all.

This post doesn't make any sense. Did you have a stroke halfway through your shitpost?

I think he's trying to say "haha look at the NERDS using their math for a VIDEO GAME"

have you
tried
clearing
your
c a c h e

Yeah sure you can't draw conclusions, but for most people when you say it should work, they'll try it

Hell I'm using all the placebo I can in Granblue and I'm sure it's bullshit but I'm not taking the risk

That's what a couple people here are saying.
Believe it if you want, don't claim it as fact.

>saying it should work means it's a fact that it does
What did you mean by this

this post is gold

Statsfag here, I always run xon in my chamber of gems team just in case, but I wouldn't level a whole team of them or anything dumb like that.
You're fine to run him all you like, but just don't hold too tightly to the idea that he actually does anything.