Quote:
Originally Posted by Emus
I am putting some thought in trying to understand your std and variance numbers after a MC run.
I found the above post in this thread.
Firstly, related to the variance number (2nd number, measure for spread but in reality calc as stdev)
If I try to calculate the variance (meaning stdev of the spread in results) = sqrt(((1.075-13.5)²+(1.075-22)²)/2)
i get 17.2 as result.
What am I doing wrong?
Well, that's a 4 year old post right there, and I seem to be getting 16.9 now instead. Not sure why. Anyhow, there's two things that you definitely need to do differently:
1) 22 is a negative number. So it should be -22-1.075.
2) The weight of 13.5 is 65% and the weight of the -22 number is 35%
So it should be sqrt((1.075-13.5)²*65%+(-22-1.075)²*35%)=16.9.
Why is it suddenly 16.9?
Not sure.
Maybe it's a rounding error.
Maybe it's an estimation error due to it being monte carlo.
Maybe I was in a hurry at the time and made a mistake when calculating that number.
Let's chalk it down to a mystery.
Quote:
Originally Posted by Emus
Secondly, related to the std (first number, measure for error of MC run)
Is it correct to say that your MC calculated EV value is following a normal distribution?
Pretty sure it's not since the possible outcomes don't have a normal distribution.
Quote:
Originally Posted by Emus
Meaning, is it correct to do the following interpretation?
68.27% chance EV lies between 4.02 +- 0.15
95.45% chance EV lies between 4.02 +- 0.30
99.73% chance EV lies between 4.02 +- 0.45
It's undoubtedly close to that, but normal distributions are rare.
Distributions that look a lot like them are pretty common.
Treating a distribution as normal usually makes things a lot easier to work with.