Hello, I would like to modell arbitrage profits between cryprocurrency exchanges. I have have 9 pairs, and 3864 hours per pair. The dependent variable is censored by zero. I would like to use CLAD because of nonnormal and heteroskedastic errors. However I don't know how to get valid standard errros for the coefficient estimates. Since I have autocorrelated errors within exchange pair, they are not valid. I found that there are bootstrap methods to get autocorrelation robust standard errors. Are there any STATA procedures to implement these bootstrap methods?