Thread:Cookie-san/@comment-26247760-20150917195427/@comment-26348726-20150919025812

I haven't personally seen structured empirical data for searchlight draw yet, but I think the last time I read about it was when I was browsing through wikiwiki comment pages during the event. I also did post my overall findings on the chat around that time, though I don't remember what they were anymore... :( I'll see if I can dig it up again. 10% sounds about right though. It might be slightly more, but definitely not by much.

AbsoluteLuck wrote: However according to our current equations, we should be able to reach 100% CI rate with say a chuuha'd fs beaver with both searchlight and starshell.

Lol yeah nope. :P If shelling accuracy has a hardcap, there's no way that CI wouldn't have one.

But ughh, questions about yasen CIs... ._. so much RNG everywhere that it's so hard to isolate its different components after repeated testing. And it's just so hard to collect enough data that makes sense in the first placve...on my own SS yasen log that's been running since about 3 months ago, my Ro is still significantly outperforming my Go-ya (in terms of CI rate), and my Imuya is still cutting in more often than my Shioi and Hacchan. Which obviously doesn't make any sense, even though I've already surpassed the 300 sample size mark...

Anyway, there's tons of possibilities when it comes to different potential capping mechanisms, but at least everyone agrees that there is a softcap on at least luck's contribution to the CI rate. You'd be hard-pressed to find any sort of consensus among any community (including JP ofc) with regards to the full CI rate formula, including what the luck->CI% softcaps are for the different types of CIs. There are formulas that use luck multipliers other than 1.0, there are some formulas that modify luck in a way that is more complex than the simple linear multiplier, there are formulas that use a square root-based softcap on luck (similar to FP softcap), and some formulas even incorporate different base CI rates for different CI types. Based on some tests with 'unusually' high luck ships actually (I think I've seen 99 luck yukikaze and 100 luck hatsushimo, dunno about any others...and I didn't keep links to the data :( There's bound to be some of that data on shitaraba if you can get someone to look for you), CI rate contributions provided by non-luck factors such as FS bonus, chuuha bonus, and yasen equipment (search/star/lookouts) definitely do not apply on a +fixed% basis when you already have a very high luck or CI rate, but... Whether these contributions share the same cap with luck, or whether there is a separate soft/hardcap for all of them, or whether they have individual caps, or whether just one or a few of them share the same cap with luck etc...that is currently impossible to determine.

It is impossible to ever 'prove' the existence of any hardcaps obviously, but that's not the issue here; the main difference between discovering the well-known accuracy hardcap and this possible CI rate hardcap is the potential existence of all these other CI rate component soft/hardcaps. However, I have yet to see any reliable, *high* sample size CI rate experiment claim to exceed a raw value of 90% with any setup, no matter how much luck and ALL DA BONUSES stacking were applied to the setup. Whereas I have seen high sample size experiments claim ~85% +/-2% CI rates using just the luck and FS bonus contributions. So perhaps that's saying something.

If you're interested in some of the more recent compilations of our knowledge on CI rates, then these might interest you, although they don't really say or prove anything about the existence of a hardcap. http://bs-arekore.at.webry.info/201504/article_4.html http://userdisk.webry.biglobe.ne.jp/024/587/63/N000/000/002/143632838657526788179.png http://mmoloda-kancolle.x0.com/image/177103.png

In the end/TL;DR, until we are able to somehow obtain more reliable information on how the CI rate is actually calculated, I really wouldn't worry too much about optimizing CI rate besides doing the obvious (providing maruyu diets without focusing too much on one ship, using searchlights and star shells when you have the spare slots and it makes sense to use them).

The chinese forum link that you gave me confirms that both 20.3(2) and Prinz guns do not activate the +10% CA yasen acc bonus, although the sample size of 200 each is a bit on the low end. Nevertheless, it's a pretty large difference and it would make sense, so I wouldn't really doubt the results. Of course, you could still obtain the bonus by using at least one 20.3(3) like I mentioned in the previous talk thread.

On another note, I bore witness to the true power of bomber xp when my 2FBB 4CV comp totally annihilated the 5-5 preboss node (except for elite Re, which was left at chuuha) in the opening stages (airstrike + support)...... ...only to get trolled by the compass and go to L >_>" dammit.