+- Southperry.net (https://www.southperry.net)
+-- Forum: Maplestory (https://www.southperry.net/forumdisplay.php?fid=15)
+--- Forum: Game Mechanics (https://www.southperry.net/forumdisplay.php?fid=33)
+--- Thread: Scroll probability formula (/showthread.php?tid=23874)
Pages:
1
2
Scroll probability formula - Kaasoljoyyx - 2011-02-06
My assumption ever since i joined is scrolls are independent of eachother. I see burn scrolling the same as gambler's fallacy.
Scroll probability formula - vx-2 - 2011-02-06
Shidoshi Wrote:Occam's Razor:
-Scrolls take information from past scrollings made by the user in X ammount of time in order to average out to about Y% probability to work, demanding the server to remember past events.
-Scrolls work Y% of the time.
:f6:
If there's a "rememberance of past events" means there's a collective set of data; which would allow statistical calcuation to work in:
a) studying the pattern
b) probability prediction of outcome by "averaging" out used scrolls in the past.
Both of which will allow 'an expected future value of success', no?
I've used a simulation program, called 'Crystal Ball" (by ORACLE), which would allow one to predict future values. (from either independent or dependent collective set of data - as long as there's data and formulas, u're set to get it). Used it in my CAD-decision making class in order to predict future pattern from given statistical set data and formula. Of course, it's not 100% accurate.
Scroll probability formula - hadriel - 2011-02-07
Either ways, they return to the scroll's probability in the long run, but there's no way to predict the short term outcome accurately because you can't exactly control statistical "noise".
Statistical assumptions made are still valid, and for very good reasons. So I agree that scroll burning is just like gambler's fallacy, while mass scrolling for that e.g. 15dex earring is pretty reasonable but not always reliable because expectation =/= observed.
Hadriel
Scroll probability formula - Shidoshi - 2011-02-07
Anyways, definitions:
Statistics: the concept that you will, on average, get at least 1 item with 10 10% scrolls working on it after each 10^10 tries (expected value). That's a prediction made on statistics, it doesn't predict whether the next scroll will work or not, just that after a sufficient ammount of tries things tend to the expected value.
http://en.wikipedia.org/wiki/Expected_value
Probability: the fact that there is a 10^-10 chance of getting 10 10% scrolls to work on an item.
Scroll probability formula - vx-2 - 2011-02-07
Shidoshi Wrote:Anyways, definitions:
Statistics: the concept that you will, on average, get at least 1 item with 10 10% scrolls working on it after each 10^10 tries (expected value). That's a prediction made on statistics, it doesn't predict whether the next scroll will work or not, just that after a sufficient ammount of tries things tend to the expected value.
http://en.wikipedia.org/wiki/Expected_value
Quite right; and you will use statistics along with the collected data (i.e. ur scrolled result: success/ fail; on which number of slot/ slot remaining/ type of scoll% and etc.) to run a "simulation" to get it (future expected value). This of course requires a simulation program. And like I've said; with a given dataset and formulas; you're all set to go. And here's the example that I've given from my previous post. (URL: http://www.oracle.com/us/products/middleware/bus-int/crystalball/index.html). Simulation is the computerize-end of using statistic, and that is why statistics is useful, otherwise, it'll just be a bunch of jargon numbers.
Scroll probability formula - Shidoshi - 2011-02-07
That program has nothing that could predict the next result of a coin flip (or scroll) based on past results. It just doesn't work like that.
There is no way to determine the future of an independent random event, for that is its definition (http://en.wikipedia.org/wiki/Independence_%28probability_theory%29).
Considering the seed generator at nexon is sufficiently random (which is generally true), there is no way to tell the next outcome based on previous outcomes.
Considering your response to my "occam's razor" post, I guess you didn't really understand what I was saying. LURK MOAR >.<
Scroll probability formula - StringStrider - 2011-02-08
Back on topic, the bunch of tables done up do not properly reflect dark scrolls, since they don't account for blowups...
Scroll probability formula - Fumni - 2011-02-08
StringStrider Wrote:Back on topic, the bunch of tables done up do not properly reflect dark scrolls, since they don't account for blowups...
Yeah, Stereo pointed that out and made a table to add to the discussion. When I get more free time, I'll add in standard deviation for all of them and fix up the 30%/70%, but that won't be until the weekend. In the meantime, if anyone wants to add their input, go right ahead.
Scroll probability formula - vx-2 - 2011-02-09
Shidoshi Wrote:That program has nothing that could predict the next result of a coin flip (or scroll) based on past results. It just doesn't work like that.
There is no way to determine the future of an independent random event, for that is its definition (http://en.wikipedia.org/wiki/Independence_%28probability_theory%29).
Considering the seed generator at nexon is sufficiently random (which is generally true), there is no way to tell the next outcome based on previous outcomes.
Considering your response to my "occam's razor" post, I guess you didn't really understand what I was saying. LURK MOAR >.<
If only you have used that program b4, then you will realize what it's capabable of doing. And I'm not saying it's 100% accuracy, but it's at least some confidence interval/ level.
EDIT: Similarly, you might want to broaden your spectrum of knowledge in regards of your beliefs in "Randomness of Coin Flipping"
Article summary: (http://www.codingthewheel.com/archives/the-coin-flip-a-fundamentally-unfair-proposition)
Research Thesis/ Paper: (http://comptop.stanford.edu/u/preprints/heads.pdf)
Scroll probability formula - Kaasoljoyyx - 2011-02-09
Fumni Wrote:Yeah, Stereo pointed that out and made a table to add to the discussion. When I get more free time, I'll add in standard deviation for all of them and fix up the 30%/70%, but that won't be until the weekend. In the meantime, if anyone wants to add their input, go right ahead.
Um, why would you even do standard deviation for this?
Scroll probability formula - Fumni - 2011-02-10
Kaasoljoyyx Wrote:Um, why would you even do standard deviation for this?
For fun. Why else would I even bother with this at all, since each scroll is independent of each other?
Scroll probability formula - Kaasoljoyyx - 2011-02-10
Fumni Wrote:For fun. Why else would I even bother with this at all, since each scroll is independent of each other?
But I don't see any use for it. It seems just more numbers but doesn't really add any important information at hand, thus it's clutter and makes it more confusing.
Unless i'm mistaken, for a binomial it's just SQRT(npq)