hints

I've been playing with the simulator on botsunauthorized.net. It says it's made from AquaRegia's script. Is this simulator pretty accurate? It seems flawed to me.

Are there any others out? That I can double check results with.

Download aquas script from userscripts.

Mine seems to not like dex bots. It's weird because I set it up exactly like his.

The results are simulated. It'll give you a rough guide but remember, this game is still random.

it sims 100k fights, every sim will be different. It's not an exact science

just increase the number of fights simmed and the accuracy will increase ;)

not necessarily, regardless of how many times you sim something it will always give a different result.

It's like shooting a gun. Shoot a bullet, you won't know the kick. Shoot an entire mag, you'll get a good idea. Fire 100 mags, well that's a hell of a lot more accurate feedback on the target than 1 mag.

exactly, It's not perfect but it will give you an idea

not necessarily, regardless of how many times you sim something it will always give a different result.

That is incorrect. As per the Law of large numbers, if you increase the number of trials in an experiment/simulation/trial, the results will tend to converge towards the expected value.

It doesn't matter how many times you sim it, the script generates random data, this data is what the fight is, you could sim 100 googol fights numerous times and the result would always be different.

No. You're not following because in the above post you seem to have changed from talking about the win/loss result to what I'll call 'fight statistics' i.e. blocking, missing, how much you hit, etc. Focus on win/lose.

Whilst fights may certainly appear random, due to the variables at play within the fighting equation, over time results are in fact biased towards an expected outcome, win or lose. This is probability theory. Put simply, these fights will exhibit a pattern (In this game due to differences in builds, although there is one exception to this rule which I'm not going to bother mentioning) and will converge towards the aforementioned expected outcome, the weighted average of all the variables/outcomes at play. This is a purely theoretical number and in a finite series may not even be achieved through millions of fights, which may seem counter intuitive to the idea of bias. Going back to you talking about the fight statistics, these too will converge (However much you seem to think they are completely random), but it is much more complicated than simply win/lose.

As I stated in my previous post, according to the Law of large numbers (Not something I'm simply pulling out of the air based on a misconception to increase my post count like you), if you increase the number of fights to be simmed, the results move closer and closer to this expected outcome, e.g. in 100,000 fights the simulator says I win 99% of the time against you and in 10^20 fights I win 99.95% of the time. Obviously we'll need to increase the number of fights to move closer to the expected outcome of me winning 99.99% of the time.

tl;dr increase number of fights in the simulator to get more accurate results.

Thanks for the help guys. But I'm sure that specific simulator was giving bogus results. I got the original script now and its working much better I'm seeing the results I would expect from what I know about certain builds. Thanks again.

your welcome :)

you're*

nos o/

\o

Guess who's back....Back again....Oh wait, I never left. Just took a "break".

fair point ^

Can you take another one please? There are two kit kats in a package.... ;)

kit kats?

Wafers topped with caramel completely coated in chocolate. Watch a jingle online and you'll get it then.

I know what they are, I jsut don't get it :|

just*

The kit kat jingle, "Give me a break, give me a break, break me off a piece of that kit kat bar."

fLight simulator? I really have no idea how it works with bots4...

http://bots4.net/forum/6/3582

Head over to that thread and you will discover how.

nos you spammer