[X] My Assistant

From some IW records(Almost 50) I gathered from other IW shops, the average shards spent on a J5 is approx. 11 shards.

It will depend on the # of possible potencies. After updating my simulator with the algorithm from https://forums.e-hentai.org/index.php?showtopic=221264, it seems to match its results (~15.24 for 7 possible and ~13.68 for 8 possible) after 10 million simulations. The math used in that post looks to be correct as well.

Of course, this is assuming that the assumptions used are correct (pretty much follows wiki's steps); only part where I'm uncertain about is if the order you got the potencies matters at all; i.e. currently assuming if you unlocked 2 potencies and both aren't at level 5, the chance for getting each is 37.5% and 37.5% rather than 50% and 25%.

Shard costs fluctuate quite a ton since it is not normal distribution; with 50 simulations / sample size I'm getting a huge amount of variance (tried with 50 sims *10 times and I got averages of between 8.74 and 18.20), but it stabilizes within 0.01 shards of error at around 10 million simulations.

If anyone has a larger amount of sample actual data I would appreciate it, as perhaps the system is different from what was assumed in the code after all.

Code: [pastebin.com] https://pastebin.com/HKMcKQqy

Data:
(IMG:[i.imgur.com] https://i.imgur.com/xH9KB43.png)
(IMG:[i.imgur.com] https://i.imgur.com/ePZB1es.png)

This post has been edited by t_t_z: Mar 11 2019, 12:02

-Snipe-

I see. That's a much reasonable approach. And how many IWBTH and PFD runs respectively?
Also would you mind telling me the shards and runs needed for p5s4, butcher fatality only and J5 C5 if you have simulated them? Thank you regardless.

I see. That's a much reasonable approach. And how many IWBTH and PFD runs respectively?
Also would you mind telling me the shards and runs needed for p5s4, butcher fatality only and J5 C5 if you have simulated them? Thank you regardless.

#runs would slightly depend on pxp0; it appears for J5 7P at least as pxp0 increases the # of PFD runs decreases (since more rounds per run, scaling more than total pxp required; do take into consideration that turn and stamina usage is per round and not per run however), until it hits high enough pxp0 to cap out at 100 rounds/run; then it starts increasing.

The following 90/95/99 percentages are confidence intervals.

Assuming a PXP0 of 362:

CODE

J5, 7 possible (elemental), 10^7 sims
IWBTH runs:    avg=10.89    90%:24    95%:32    99%:48
PFUDOR runs:    avg=20.80    90%:37    95%:46    99%:67
Shards cost:    avg=13.68    90%:33    95%:43    99%:67

CODE

J5, 8 possible (non-elemental), 10^7 sims
IWBTH runs:    avg=12.44    90%:28    95%:36    99%:55
PFUDOR runs:    avg=20.80    90%:37    95%:46    99%:67
Shards cost:    avg=15.24    90%:36    95%:48    99%:74

CODE

J5C5, 7 possible (elemental), 10^7 sims
IWBTH runs:    avg=149.68    90%:344    95%:447    99%:687
PFUDOR runs:    avg=58.05    90%:120    95%:154    99%:231
Shards cost:    avg=184.14    90%:425    95%:553    99%:850

CODE

J5C5, 8 possible (non-elemental), 10^7 sims
IWBTH runs:    avg=199.65    90%:459    95%:597    99%:917
PFUDOR runs:    avg=62.97    90%:132    95%:168    99%:253
Shards cost:    avg=238.58    90%:550    95%:716    99%:1101

(note that weapons in general have lower pxp0 than armor but the following is still assuming pxp0=362)

CODE

B5F4(Hollow1), 4 possible, 10^7 sims
IWBTH runs:    avg=64.18    90%:147    95%:191    99%:294
PFUDOR runs:    avg=55.41    90%:114    95%:146    99%:218
Shards cost:    avg=96.60    90%:224    95%:291    99%:448

CODE

S5P4(Hollow1), 5 possible (non-focus), 10^7 sims
IWBTH runs:    avg=106.95    90%:246    95%:320    99%:491
PFUDOR runs:    avg=62.42    90%:130    95%:166    99%:250
Shards cost:    avg=146.04    90%:338    95%:440    99%:676

Note that I am still unsure of the generalized optimal IW strategy for double potencies, currently reforging only if impossible or if any undesired potency is unlocked before both of the desired potencies are unlocked.

This post has been edited by t_t_z: Mar 11 2019, 14:05

Snipe

Thank you very much ヾ(*ΦωΦ)ツ

It seems the results I estimate according to these simulators are significantly more expensive than what I got from various shops.

This post has been edited by -vincento-: Mar 11 2019, 14:22

Thank you very much ヾ(*ΦωΦ)ツ

It seems the results I estimate according to these simulators are significantly more expensive than what I got from various shops.

Yeah I noticed that too. Trying to figure it out, it seems that in the calculation of the average, there are some super huge # of shards (sometimes over 1000, albeit very rarely) that are still taken into consideration. I believe it is either that or perhaps the shops are using a better strategy (I can't find an optimal one posted for dual potency).

Do you happen to have found/collected more data on the double potency requirements from shops? They seem very rarely requested without allowing substitutions, and even if they are, sometimes they run out of budget and end up with something else (thus may differ from my results). Looking through Kadokura's shop's logs, I found only:

IWBTH/PFF/Shards:

For J5+specific potency5, 2 usable samples and 1 skewed:
74+/20+/81+ (changed to J5C4C1)
175/51/211
32/23/39

For S5P4, 9 usable samples and 4 skewed:
60/37/82
31/18/39
102+/46+/131+ (changed to P5S3A1)
44/23/55
152+/93+/217+ (changed to P5A4)
0/13/0 (started with S1)
18/20/25
6/9/5
47/42/67 (Focus)
2/9/1
57/35/78
37/12/39
112/76/36

Allowing substitutions after running out of budget or free runs may skew it as those are not used in my simulator. Not sure how much of an impact it has, but the ones that have run out of budget will likely cost a ton more if used my simulator, maybe that leads to the discrepancy? (budget limit kind of completely removes the samples with large shard usages/such). Sample size is also very low.

It may also just be an error in the code or such (just spent quite a while looking through it / going through unit test/cases), if anyone else is interested and found a bug, please let me know. (although I know, it's very ugly and crappy code and probably very hard to understand)

In the end, I'd only recommend using the simulator as a general guideline; the variance for shard usage, especially for dual requirements seems to be very high and thus quite unpredictable (also why high confidence intervals are so much higher than the average)

This post has been edited by t_t_z: Mar 11 2019, 16:01

From some IW records(Almost 50) I gathered from other IW shops, the average shards spent on a J5 is approx. 11 shards.

so you're saying i got cucked massively by snowflake. this is sexy

Do you happen to have found/collected more data on the double potency requirements from shops? They seem very rarely requested without allowing substitutions, and even if they are, sometimes they run out of budget and end up with something else (thus may differ from my results). Looking through Kadokura's shop's logs, I found only:

Sorry I only recorded a few of them to have a rough estimation, mostly from Kadokura and Lostwizard. But you can visit these two shops to collect some more data. Also detailed Strategy explained by Lostwizard. You can go take a look.
Vivi's Shop
Lostwizard's IW history

This post has been edited by -vincento-: Mar 11 2019, 19:14

Sorry I only recorded a few of them to have a rough estimation, mostly from Kadokura and Lostwizard. But you can visit these two shops to collect some more data. Also detailed Strategy explained by Lostwizard. You can go take a look.
Vivi's Shop
Lostwizard's IW history

Oh, didn't see these posts. They seem really helpful, thanks!

Yeah it actually appears somewhat close after you take into consideration of the fact that people may change / accept substitute potencies after reaching their budget.

Here's the data I got with LostWizard's data using all attempts where the initial requested potency is P5S4 (reverted all attempts to change the requirements after a "close, acceptable, but not perfect attempt" to a reforge (valuating cost of reforge = average)): [drive.google.com] https://drive.google.com/file/d/19EzPoUnrOG...iew?usp=sharing

Then again, it's still too low of a sample size. In the process of adding more, a bit confused about how Vivi keeps reaching tier 2 and not tier 3 after initial PFUDOR runs, I believe that shouldn't be possible if using current wiki formulas? Or perhaps I missed something.

This post has been edited by t_t_z: Mar 11 2019, 21:12

In the process of adding more, a bit confused about how Vivi keeps reaching tier 2 and not tier 3 after initial PFUDOR runs, I believe that shouldn't be possible if using current wiki formulas? Or perhaps I missed something.

This was before the patch that changed the experience multiplier on the IW. Before, both IWBTH and PFUDOR had the same multiplier: 16x. Later it became 15x to IWBTH and 20x to PFUDOR.

This post has been edited by Basara Nekki: Mar 11 2019, 21:32

so you're saying i got cucked massively by snowflake. this is sexy

Talk about cucked by snowflake (IMG:[invalid] style_emoticons/default/cry.gif)

Talk about cucked by snowflake (IMG:[invalid] style_emoticons/default/cry.gif)

You dont even know what that means. (IMG:[invalid] style_emoticons/default/wink.gif)
179th reforge. Not long now. (IMG:[invalid] style_emoticons/default/happy.gif)

You dont even know what that means. (IMG:[invalid] style_emoticons/default/wink.gif)
179th reforge. Not long now. (IMG:[invalid] style_emoticons/default/happy.gif)

(IMG:[invalid] style_emoticons/default/ohmy.gif) (IMG:[invalid] style_emoticons/default/ohmy.gif) (IMG:[invalid] style_emoticons/default/ohmy.gif)

I hope you get it soon. (IMG:[invalid] style_emoticons/default/wink.gif)

I have two examples of how this type of IW can be quite variable.

No luck (IMG:[invalid] style_emoticons/default/sad.gif) : https://forums.e-hentai.org/index.php?s=&am...t&p=5201798

With luck (IMG:[invalid] style_emoticons/default/smile.gif) : https://forums.e-hentai.org/index.php?s=&am...t&p=5370711

Yeep
With luck:

https://forums.e-hentai.org/index.php?showt...p;#entry5366730

3 Amnesias for but5fat4
(IMG:[invalid] style_emoticons/default/laugh.gif)

I hope you get it soon. (IMG:[invalid] style_emoticons/default/wink.gif)

Me too. I am slowly running out of shards. And tbh, i dont have the credits to buy another big pack. So when i dont get it with those shards i have left, this will take a while.

No luck (IMG:[invalid] style_emoticons/default/sad.gif) : https://forums.e-hentai.org/index.php?s=&am...t&p=5201798

131 reforge is not "No Luck" this is only bad luck. No luck is atm at 182 reforge. (IMG:[invalid] style_emoticons/default/happy.gif)

With luck (IMG:[invalid] style_emoticons/default/smile.gif) : https://forums.e-hentai.org/index.php?s=&am...t&p=5370711

And again i have to disagree. This is with luck.
https://forums.e-hentai.org/index.php?s=&am...t&p=5366494

And again i have to disagree. This is with luck.
https://forums.e-hentai.org/index.php?s=&am...t&p=5366494

Uncle, luck and nigh perfection are not the same thing (IMG:[invalid] style_emoticons/default/tongue.gif)

Uncle, luck and nigh perfection are not the same thing (IMG:[invalid] style_emoticons/default/tongue.gif)

Perfection is a goal. A never ever reachable goal.
Luck = preparation + occasion + skill
Why would i think those are the same? (IMG:[invalid] style_emoticons/default/huh.gif)

Why would i think those are the same? (IMG:[invalid] style_emoticons/default/huh.gif)

Because Basara Nekki's luck example is still luck (IMG:[invalid] style_emoticons/default/rolleyes.gif)

1/3 and still lose
sigh

Another monsterslot unlocked. (IMG:[invalid] style_emoticons/default/happy.gif)

And 186th reforge. I have to be close now. Right? ...... right? (IMG:[invalid] style_emoticons/default/unsure.gif)

Another monsterslot unlocked. (IMG:[invalid] style_emoticons/default/happy.gif)

Grats (IMG:[invalid] style_emoticons/default/cool.gif)

And 186th reforge. I have to be close now. Right? ...... right? (IMG:[invalid] style_emoticons/default/unsure.gif)

Not so grats (IMG:[invalid] style_emoticons/default/sad.gif)

But you'll do it, if not today or tomorrow, definitely before the end of the week (IMG:[invalid] style_emoticons/default/happy.gif)

It is a peerless, right ? Otherwise you're a bit crazy, I'd say. (IMG:[invalid] style_emoticons/default/heh.gif)