SCH Theory - Dropping MND for Crit?

[FinagleABagel]

This is actually the first time I'm hearing of this and now you've gotten my attention, whats the % rate cap at? Was there some testing posted somewhere to back this up ?

Also, while this is slightly off topic, anyone else notice there embraces criting far more often then physick/aldo? Could just be coincidence but wasn't sure if the fairy's crit had a slight increase in relation to your base crit when summoned.
Anyways carry on, I'd love to see it if there really is some testing on the aldo crit % rate cap.

I cast Embrace a TON more than anything else. Thus, I see more crits but I'm thinking I'm more of a victim of selection bias than finding some sort of special different in critical rates.

[Sygmaelle]

Yes but even less then it helps you

1 MND = 1.175 HP for Physick (average)
1 MND = 0. somethig for Embrace iirc

5 DET = 1 mnd

Make the mats :P

Hah fair enough, thank you :P

Otherwise yup, Adloquium is crit capped at 470 precisely, its like 17%. I don't have more than that rate of Crit adlos done since I went up in crit.


http://www.reddit.com/r/ffxiv/commen...quium_testing/

Didn't read again but think it's there I found the info


As for the crit thing : its w/e really, purely RNG based guys'n'gals. Sometimes your fairy will crit 8 times in a row, sometimes zero at all, sometimes its you.

[FinagleABagel]

I'm sorry but that link is a load of bunk. No testing method on the crit cap other than the dubious claim of "using a parser".

I pretty much trust nothing that comes from Reddit. A lot of "theory crafters" that have a poor understanding of statistics or testing methodology.

There's RNG involved with just regular healing as well, there's a defined range. Sometimes you will hit on the high-end, other times the low.

The point is that we are talking about AVERAGE heal over a given amount of GCDs. I'm not convinced that a critical build will lead to a lower AVERAGE heal over a certain number of GCDs. I do believe that there will likely be an inflection point based on battle duration that will either give crits or potency the edge.

[Sygmaelle]

I dont think you get what I tried to say.

I for one just never trust forums because its a lot of "basic infos" that are not explained in details on purpose, and in general, the "bigger" picture is all the time pretty much not included in profit of pure non sense personnal theory crafting. While its cool to know how your class behave you simply can't rely on things said and experiment what work best for you in your usual gaming environment.


There's no crit cap on healing output of a Scholar so to speak. There's however one on adloquium and thats a granted, if you can't theory craft things out of a parser by doin yourself % , how can you claim that data is wrong ?

From my tests, Adlo is crit capped and I found an info somewhere sayin that the limit was at 470 : from my personnal sch experience it sound about right.

For everything that is not Adloquium, there's no cap at all. The more crit the merrier.

/cheer


To test that out, just do Adloquiums with X amount of crit and raise that amount, see if you do more of them, but from 470 I can guarantee that you ll hardly see a difference. It also apply for the adloquium goin after a succor. While succor isn't crit capped, the adloquium after is.

[FinagleABagel]

stuff

Sorry, math doesn't work that way. You've not accounted for a ton of variables, which is why I trust nothing you are posting.

Implications that you need to reconcile before anyone takes you seriously:

1) Is 470 critical rate equal to 17% on crit on all skills? Did you test all skills? Do they have a central tendency normally distributed around 17%? If not, your methodology or understanding of underlying systems is incorrect.

2) What is the growth rate on critical rate? Please derive the logarithmic function with r-square. If you have so much data, that regression should be easy.

3) If other skills are uncapped on critical rate, what is their growth rate? Are they exponential? Polynomial? Linear? Non-standard "stair stepping"?

And so on...

Basically, if you don't have a degree in econometrics you don't stand any chance of getting this stuff right.

[cainejw]

Bagel, seriously, confounds do not undo research. Confounds just add to error rate. Additionally, standard deviations and variances do not undo research. Chance does not undo research. Errors are built in. In fact, stats always identifies potential error.

So, yeah.

[FinagleABagel]

Bagel, seriously, confounds do not undo research. Confounds just add to error rate. Additionally, standard deviations and variances do not undo research. Chance does not undo research. Errors are built in. In fact, stats always identifies potential error.

So, yeah.

I don't doubt that you did work to come up with your results. I'm saying you can't apply them to anything because you didn't account for confounding variables. If there is a huge error rate with your results, would you feel comfortable with them?

Good research knows the sources of error and how they may change the results. If I asked for a 95% CI, could you give it me? I have a feeling that interval is going to be something like 300-600 being the "cap" because you have that much error built in.

Quantify things.

[Sygmaelle]

1) Is 470 critical rate equal to 17% on crit on all skills?

Yes sir, you can do some research around and see by yourself that 500 crit is ~20% crit rate. I can't really be totally precise here if its really what you want, but I can dig the data I found on Bluegarter forums, between others. Anyway, my point was : Adloquium is crit capped. All the other heals aren't.

And somehow I find it to be normal, otherwise Adlo would simply be too powerfull

[FinagleABagel]

I'm not theory crafting for a reason, I know how many variables need to be accounted for and I know the data simply isn't there. Yes, you can't trust other people.

I don't trust Blue Gartr. I don't argue that they are very skilled players, but to think that anyone is going to get this correct is pretty silly. Backing into a function solely on the basis of point estimates, regardless of the quantity, is tough to do.

You don't know what's at play for round of testing. Your extrapolation may correctly predict in a vacuum, but making those numbers universally applicable is a total joke.

[jumpypuppet]

Yes but even less then it helps you

1 MND = 1.175 HP for Physick (average)
1 MND = 0. somethig for Embrace iirc

this is wrong. DET impacts fairy more than it does yourself. whereas the DET:MND ratio for you is somewhere around 5:1 or more, the DET:MND ratio for fairy heals is closer to 1:1. test it

re: crit, the prevailing logic is and has been that it's beneficial for scholars to stack crit because of the inherent bonuses that come with it. WHMs don't stack crit, for example, because they have no built-in incentive other than the chance to (over)heal for more. however recently i've been seeing people challenge this idea, which is interesting to watch. i'm curious to see how it plays out, and whether the prevailing logic ultimately favors another stat (perhaps SS, which Selene magnifies strongly, or DET, which as mentioned highly benefits the fairies). in the meantime, though, i'll continue to stack crit, and recommend that TC ditch the hat