• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Monte Carlo Simulation Test

Joined
3/3/11
Messages
22
Points
11
So I've got a Monte Carlo simulation runnng in Java. What's the bext way to test it? (currently I have to run it manually many times to get a range of results)

Thanks.
 
So I've got a Monte Carlo simulation runnng in Java. What's the bext way to test it? (currently I have to run it manually many times to get a range of results)

Thanks.


Are you requesting for example or...? "Best way to test it"?? Please be more specific.
 
Sorry, what I meant to say was how do I know if the result simulated from my Monte Carlo is in fact in the correct range?
 
Well what exactly are you simulating ? Let's say you are simulating a geometric brownian motion, you can look at the distribution's property (so you will need to store results) at a given time of your simulation and compare with the mean and std deviation you should be obtaining.
 
Sorry, what I meant to say was how do I know if the result simulated from my Monte Carlo is in fact in the correct range?

You should check for actual values. Take a computable example and compare the simulated outcome with the actual one. That's what I asked first, if you need an example then consider the above post suggesting to simulate a brownian motion and compare the simulated and actual characteristics.
 
It all depends on what you are simulating. If an analytic solution exists, you can compare it to that. If not, compare it to some other numerical solution. If no known solution exists, then for example you can fit it to a normal distribution and find out the confidence intervals. You can then increase the number of samples and see how impacts the confidence intervals.
 
You might want to employ the outcomes of other simulation techniques to compare to that one. Knowing the particular problem would let us give more specific suggestions. I like the above post, you can employ central limit theorem to the distribution and compare the result to simulated impact.
 
Back
Top