That one-in-500 chance

Status
Not open for further replies.

Iminyourcloset

OBJECTION! What do you mean I have a weakness now?!
is a Battle Server Moderator Alumnus
#2
Why would you make this a thread...
This is how percentages work:
let's say you use a 99% accurate move. Technically, that move may never hit. because there is that 1% possibility, that continually diminishes, but never hits 0. It's terrifyingly small, but it's definitely possible, and always will be possible. You, my friend, have been hit with something called "hax"-- aka Baaaad Luck. And also, paralysis chance is 25%.
So, before you suspect our simulator to be biased, consider how the real world works.

EDIT: Make that your opponent had hax*
 

tehy

Banned deucer.
#3
So have you played 500 matches?

More to the point, i'm sure more than 500 matches are being played on PS! at any given time. So in one of them, something insane happens, a 1-in-500 chance.

It's very unlikely that the simulator is biased in such a manner. If you notice, in the credits, the actual names of the people who made PS! are there. Why would they attach their names to something purposefully flawed? And I don't think they could miss such a mistake, either.

If you really want, though, why not try it out? Get a paralysed harvest leppa mon versus a harvest leppa mon with permanent sun up, both with non-damaging attacks. Attack 1000 times. Record the results. If it's really heavily skewed, bring up that data. (I mean, it'd probably have to be tested more, etc. But that's a start.)
 
#5
The point is, games where things are meant to be 50/50 usually record past events, and then work to normalise the distribution - i.e. each time you don't get para, it increases the probability of para next time, so that you are consistently drawn towards 50:50 on a micro scale, not over a sample of 1000000 turns. The simulator is clearly not doing this.
 

tehy

Banned deucer.
#8
Uh to clarify

The actual pokemon game-and most games like that in general-don't have coding that makes stuff regress to the mean. Past events are NOT taken into account. Nor should they be.

Also, yeah, paralysis chance is 25%.
 

NixHex

Doing just fine, here at the top of the world
is a Site Staff Alumnusis a Forum Moderator Alumnusis a Researcher Alumnusis a Contributor Alumnusis a Battle Server Moderator Alumnus
#9
You're right, the simulator clearly doesn't do it. Mystery solved. Read tehy's post.

By the way, you know showdown is open source, right? Go here: https://github.com/Zarel/Pokemon-Showdown
You can peruse the code you want and determine the probabilities from there. Since you seem to know a lot about random processes and their implentation in videogames, you shouldn't have any trouble reading the code.
 

Zarel

Not a Yuyuko fan
is a member of the Site Staffis a Battle Server Administratoris a Programmeris a Pokemon Researcheris an Administrator
Creator of PS
#11
The point is, games where things are meant to be 50/50 usually record past events, and then work to normalise the distribution - i.e. each time you don't get para, it increases the probability of para next time, so that you are consistently drawn towards 50:50 on a micro scale, not over a sample of 1000000 turns. The simulator is clearly not doing this.
No, they don't.

In fact, the Pokemon games, including its random number generator, are well disassembled and understood, and they definitely don't do that. (Not to mention that trying to patch a random number generator not to be "too random" is bad game design for a number of reasons.)

So unless you're telling us to stop simulating pokemon and start simulating something that's not pokemon, this is entirely a non-issue.
 

Pikachuun

邪法使い_CC
is a Pre-Contributor
#13
am i the only one who's interested why he's whining about not getting paralyzed?
Nope. Why would you whine about not getting para'd when you can whine about your opponent not getting para'd? Or you could even whine about getting para'd. Otherwise I don't see enough reason to whine.
 
Status
Not open for further replies.

Users Who Are Viewing This Thread (Users: 1, Guests: 0)