Hi,I currently have data for 2 things: 1) CDS spreads (Weekly) 2) Major indices (Weekly)I would like to model volatility for these 2 series however, literature tends to use the sum of ln(5-min intraday returns differences) as the correct Y variables.My question is: Since it's impossible to get intraday ticks for my 2 sets of data how can I define a proper Y for each of those to try to model volatility using a GARCH(1,1) model?Thank you,Alan

What is the variable 'Y' that you are referring to?

I think Alan696 wants to compare the dynamic of the volatility of equity major indices to volatility of CDS indices.He defines Y1(t) as the CDS time serie and Y2(t) as the equity time serie.He wants to estimate the garch(1,1) model i.e.:h(t) = a Epsilon(t-1)^2 + b h(t-1)whereEpsilon(t)=z(t).sqrt(h(t))z(t) is i.i.d. with zero mean and unit varianceh(t) = ln ( Y(t) / (Y(t-1) )@Alan696 : I do not know the purpose of this study but in fact you will not find anything interesting with weekly time series.

Hi back,Initially my question was about the optimal measure of volatility. Since then I've reviewed different academic papers and the literature doesn't seem to propose any concluding statement. However, since my last post, I've had the opportunity to see my data set and concluded that using the close remains the best way (still debatable) in terms of efficiency, practically and because of hedging purposes. Note that even though most papers use the close as input variable, some papers propose different frameworks including intraday ticks, median of the day, average, realized vol, special metrics derived from implied vol etc. RentMe was right as my end goal is to derive a multivariate GARCH (DCC, Engle (2002)) including those 2 variables (the final model should include 5-6 variables)@RentMe: Does your last statement imply that I should use daily data? @RentMe: Can you please contact me at alan.harper696@gmail.comBest,Alan