Critical, Hardened Skin, Orb of Lightning/Slow high proc rate bugged?

[Joccaren]

Sure, if you can tell me how to add an attachment without it using any of your attachment space. I've filled mine up. (Or how to add pictures without using attachment space, either would work)

 

[Chocobo]

post triggers you use then else :

files : http://www.megaupload.com/
imgs : http://imageshack.us/

 

[Joccaren]

No triggers and how do upload to megaupload?

 

[KileRatZ]

My critical strike works at any % level 5,10,15,20,25,30,35,40,45,50,55,60,65,70,75,80,85,90,95 and 100% all work for me.

Yes, but do they work correctly? That's the issue we're discussing here. Set the chance to 95% and I'm sure it can be seen even by the naked eye that your unit won't crit 95% of the time.

Chocobo, I tried that link on another browser and it returned the same error page :\. I guess you'll have to register + log in at dota allstars forums first then visit that link, sorry about that.

 

[Joccaren]

It does crit around 95% of the time. Bsically its Crit, Crit, Noncrit,Crit,Crit,Noncrit, Noncrit, Crit. ect.

 

[Bird]

^ Heh I like how he gave an example of 62.5% as if that confirms it works 95% of the time...

Interesting research by the thradstarter!!

 

[Joccaren]

I now think I get what your talking about. You are thinking that experimental and theoretical probability are the same. THEY ARE NOT. Just because something will theoritically activate 95 out of 100 times does not mean it will. It has a 95% chance of activating EACH USE. The section in the 'chance to activate (WE is not responding so can't give real name)' Section is not how many times out of 100 it will happen, it is how much chance it has to activate each attack. It is possible for it to have a 99% chance of activating yet not activating once because it keeps hitting that 1% chance not to activate and therefore, not activate.

Its like making a custom spell with a 17% chance of working with triggers. If the Integer which your set to a random number from 1-100 is equal to or less then 17, it activates. It doesn't mean that it will activate 17 out of 100 events.

That is probably where your going wrong. It is not bugged, it just keeps hitting the 'above 17' (Example) numbers, which tells it not to activate.

My crit Strike activates most of the time. The more i increase the level, the more it activates. If you want it to activate X out of 100 times, you'll probably have to use integers, dummies and abilities that have 100% chance of occuring.

Hope that helped

 

[KileRatZ]

Experimental and theoretical probability are supposed to converge to the same number when you put the number of experimented hits/crits to infinity. And research shows that even after many many hits, that doesn't happen as you can see from the tables...

Now I'd say it's pretty much impossible that a 95% theoretical chance gives me 81.380% with experimentation even after a million hits. In fact statistics say this can happen, but only in about one in every (20^140000) tries, I'm sure you can imagine what a low chance that is.

 

[black.sheep]

Could this be because blizzard uses a system where each time you don't crit, your %chance to crit increases, and each time you do crit it decreases, in order to avoid long runs of crit or no crit?

I belive I read on some dota forum that the crit chance is "balanced out" or something, that if you get a string of non critical hits, then when you go into combat again, you will have a higher chance of criting.
Sorry if i just brang up something you allready knew
Oh and if you wanna do more, check stuff like, strings of crits (IE how many crits you can get before not criting), largest clusters of crits, triggered crits, etc.

 

[Joccaren]

If it happens consistantly 8/10 times every ten attacks, then I will agree that it is glitched. If that is just the average, tell me the numbers that went into the average and then I will decide if I think its glitched.

 

[Chocobo]

>Oh and if you wanna do more, check stuff like, strings of crits (IE how many crits you can get before not criting), largest clusters of crits, triggered crits, etc.

I was already working on it and it seemed a 25% chance would cluster around exactly after 4 hits, seems to be pseudo random distribution where the "exact" amount of hits needed for a crit would be (rnd(current proc rate/0.05)/25)^-1

ex: 25% -> 0.25^-1 = 4
26% -> 0.25^-1 = 4
20% -> 0.20^-1 = 5
30% -> 0.30^-1 = 10/3


>You are thinking that experimental and theoretical probability are the same.

experimental probability - theorical probability are always supposed to converge to 0.


>Section is not how many times out of 100 it will happen, it is how much chance it has to activate each attack.

ex: 95% proc rate : 1.00*0.95 = 0.95 = 95% rate
(1st screenshot) : 95% proc rate does 80.912% rate, which is 852 crits out of 1053 would require 201 fails of 0.05 while the theoric value is 1000. odds of happening : either 5.04.10^-4 or 2.80.10^-193 (idk if it first or second one) (would counter what is said at 2>)