Dear Johannes,
After Bayesian estimation, I simulate observable variables and generate autocorrelations of simulated variables, I found some autocorrelation of simulated variables are negative, I am wonder that should the sign of autocorrelations of simulated variables be positive or negative?
Thank you very much and look forward to hearing from you.
Best regards,
Jesse
PhD Candidate
There is no general answer for this. The autocorrelations should mirror the ones in the data, regardless of their sign.