I did some Monte Carlo simulation based on empirical data and would like to know if there are any quantitative approach to evaluate results' convergence. So far, I am visually check the cumulative mean value for my outputs. Below is a simplified my MC approach and thanks for any suggestions:

  1. Variable A is a 100k*1 vector, and variable B is a vector of 500 rows
  2. Each step, randomly sample n rows from vector A as A_sample and m rows from vector B as B_sample
  3. Multiple A_sample and B_sample as the output variable
  • $\begingroup$ If you know the true value, sum of distance from it should work. $\endgroup$ – SmallChess Jul 6 '17 at 0:02
  • $\begingroup$ In this case, I do not ... $\endgroup$ – TH339 Jul 6 '17 at 0:10
  • $\begingroup$ Have you looked quant.stackexchange.com/questions/17204/…? Mark Joshi's answer is nice. $\endgroup$ – SmallChess Jul 6 '17 at 0:19
  • $\begingroup$ Inverse square root of sample size? Does that work for you? $\endgroup$ – SmallChess Jul 6 '17 at 0:20
  • $\begingroup$ @SmallChess, thanks for pointing to the thread. Will check if there is a citation for this. $\endgroup$ – TH339 Jul 6 '17 at 2:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.