you don't need 10000 hits to get a decent number.
Basically, if you do 10 hits, you'll get a "order of magnitude" result, doing 100 hits should give an approximate amount, with pretty reasonable confidence, and doing any more than 100 is just increasing your precision/confidence.
Doing 10000 hits is WAY WAY WAY overkill. 1000 is overkill. Recording 1000 hits should give you a percentage in the decimal range with fair confidence. I'm sure the overall confidence and error margins introduced by the slide bar provide you with far greater error than the percentage of landed/missed hits were for those 100-1000 hits.
10 hits? Not nearly enough. 100 hits, nearly there. 500-1000 sounds about right.
It doesnt represent the reality of the situation you're trying to solve.
The whole reason people want to know the evade/AAD/AR interaction is because of the inconsistency. You wont get that inconsistency in 10 hits, you'll get a massively inaccurate portion of results. In 500 hits, that "omg evadez broke" and "wow cnt hit you j00" will balance out and give a far better representation.
how many hits you would guess gives somewhat accurate result how SA skill affects sneak attack multipliers?
lets say first 100 hits gives you only 3-9 and 15. its safe to asume 10-14 are skipped and raising SA effects only max multiplier?
slightly oftopic, but you really cant say anything about anything with 10-100 hits
Last edited by Otinsainpas; Aug 28th, 2010 at 12:27:11.
You hit Tarasque with nanobots for 18280 points of melee damage.
First shade with Blades of Boltar
---
How much is enough?
Member of Halinallet!
sneak attack multipliers and chance to hit can't really be compared with respect to how much data is needed. chance to hit has 2 bins for the data to be sorted into - hit or miss. sneak attack multipliers has a lot more bins and thus requires more data to get a decent amount for each bin. sneak attack multiplier requires even more data than a situation with roughly equal chance of occurence because some of the bins occur infrequently compared to the rest.
one of the great mind-screws is the laws of probability in mathematics:
example 1) a single 6 sided die when rolled has
1 way of rolling a 1
1 way of rolling a 2
1 way of rolling a 3
1 way of rolling a 4
1 way of rolling a 5
1 way of rolling a 6
hence, you are really flipping a six sided coin by rolling one die, and results are completely random.
example 2) two 6 sided dice, when rolled together have
1 way of rolling a 2
2 ways of rolling a 3
3 ways of rolling a 4
4 ways of rolling a 5
5 ways of rolling a 6
6 ways of rolling a 7 (why we say lucky number seven)
5 ways of rolling a 8
4 ways of rolling a 9
3 ways of rolling a 10
2 ways of rolling a 11
1 way of rolling a 12
how does this example relate to the hit and miss ratio in relation to stats? I haven't a clue
but consider that the more variables (dice) you throw into the mix, the more one outcome becomes favoured over the others when the dice are rolled.
something to consider when testing a lot of variables together and expecting to see definite patterns emerge from the data. not to mention, variables that we may not have control of. (ie - a players level/hidden stats messing with results)
*edit* forgot to put in here that my point being, if you have multiple things working together (evades, aad, full def etc...) you are rolling more dice (with pre-define parameters/stats) and a specific outcome becomes more likely than one die working on its own, contrary to the logic of one specific stat being more constant than many working together.
Last edited by SoulTarder; Aug 31st, 2010 at 10:38:35.
wtf happened to my avatars eyebrows?
I used to listen to Dubstep in the 90's... every time I connected to the internet.