I have also encountered the error "Error in diag(fit$robust.cvar) : nonvalid 'nrow' value (too large or NA)" in my data analysis.
Here is the link to several time series, which all experience either the above error or the "failed to invest hessian" error, which seems to have a similar error source (non-convergence). https://www.dropbox.com/s/ns6okxeib42syxo/NOT_WORKING.csv The reproducible code for this data is: spreads <- read.zoo("NOT_WORKING.csv", header=TRUE, sep=",", format="%d-%m-%y") spreads <- na.locf(spreads) returns1 <- diff(log(spreads), 1) remove(spreads) returns_crisis <- window(returns1, start="2007-07-23", end="2009-01-30") returns_post <- window(returns1, start="2009-02-01", end="2014-05-12") model<-ugarchspec(variance.model=list(model="sGARCH",garchOrder=c(1,1)), mean.model=list(armaOrder=c(1,0),include.mean=FALSE),distribution.model="ged") model_crisis <- lapply(returns_crisis, ugarchfit, spec=model, solver="hybrid") model_post <- lapply(returns_post, ugarchfit, spec=model, solver="hybrid") I've tried playing around with adding constrains but was unable to find a solution (even using the hybrid solver). My original data contains about 100 additional time series (all covering the same period, however with more/less NA's depending on the company). However, these all converge without failure so I have not shared them (I can upload them if necessary). -- View this message in context: http://r.789695.n4.nabble.com/understanding-an-error-from-ugarchfit-tp4684836p4694076.html Sent from the Rmetrics mailing list archive at Nabble.com. _______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.
