You need about 10000 coin tosses to establish with 95% certainty that their probabilities are 50% each (this is because you have infinite random outcomes instead of a finite population which would makes things easier)
So you can imagine that proving that the RNG for burgle is wrong would take a lot of time and effort.
But ok, assume I'm bored at work, let's say you're testing burgle against the same deck (say druid) all the time. Let's assume there are 25 different possible class cards. That means you would expect a 1/25 chance (4%) of drawing that card. In order to test this in Excel I use Int(randbetween(1,25)) to generate the random numbers (I can drag this formula down as much as I want) and then I count the number of times, say, a 1 appears (1 would represent a druid card. I then divide this count by the number of simulations to obtain the % appearances in my data set.
In 10000 simulations I always get 4% chance which seems to prove that my results are OK for RNG applied. If I used 1000 simulations however, I have between 3-5% chance of getting a 1 (which would represent a druid card) which is not a very good result. If I did only 100 simulations, I seem to be getting only between 0% and 8% chance which means big margin of error.
So yeah, basically you need a big data set. Good luck lol.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
⚙
Learn More
Cosmetics
Related Cards
Card Pools
✕
×
PopCard Settings
Click on the buttons to change the PopCard background.
Elements settings
Click on the button to hide or unhide popcard elements.
You need about 10000 coin tosses to establish with 95% certainty that their probabilities are 50% each (this is because you have infinite random outcomes instead of a finite population which would makes things easier)
So you can imagine that proving that the RNG for burgle is wrong would take a lot of time and effort.
But ok, assume I'm bored at work, let's say you're testing burgle against the same deck (say druid) all the time. Let's assume there are 25 different possible class cards. That means you would expect a 1/25 chance (4%) of drawing that card. In order to test this in Excel I use Int(randbetween(1,25)) to generate the random numbers (I can drag this formula down as much as I want) and then I count the number of times, say, a 1 appears (1 would represent a druid card. I then divide this count by the number of simulations to obtain the % appearances in my data set.
In 10000 simulations I always get 4% chance which seems to prove that my results are OK for RNG applied. If I used 1000 simulations however, I have between 3-5% chance of getting a 1 (which would represent a druid card) which is not a very good result. If I did only 100 simulations, I seem to be getting only between 0% and 8% chance which means big margin of error.
So yeah, basically you need a big data set. Good luck lol.