this is my first post! I'm not a mathematician, so please reply in layperson's terms.
I read Douglas Hubbard's excellent book, "How To Measure Anything," in which he describes how to use Monte Carlo simulations to model the likelihood of different event outcomes in Excel.
The formula he uses is $=\text{NORMINV} (\text{RAND()}, \text{mean, (max value - min value)}/3.29)$
However, this formula assumes that outcomes occur in a standard distribution. What if my distribution is not standard?
For example, let's say I'm doing a Monte Carlo to see how much my house might sell for. The max expected value is $\$500,000$ and the min value is $\$400,000$. However, the likeliest outcome is not in the middle, but slightly lower -- $\$430,000$ rather than the mean of $\$450,000$, depending upon how buyers respond to repairs that are needed.
Any idea how I would model this calculation instead? Thanks!
RAND() creates a random number between $0$ and $1$. It represents the value of the cumulative distribution function of your prices. If you have a desired distribution in mind you need to come up with an expression that takes a cdf value and returns the price, so if you are given $0.3$ your respond with the price that will be too high $30\%$ of the time and too low $70\%$ of the time. This is what NORMINV is doing for the normal distribution. There is a discussion of the transformation method for getting your desired distribution in Numerical Recipes, starting at page 288.