I want to do a daily, weekly and monthly regression between InvestmentGrade Credit Spreads (Dependent Variable) and Treasuries (Independent Variable). My starting point is daily spread data and daily prices for US treasuries. Should I convert the US Prices into log returns i.e. log(Pt/Pt-1) or simple daily returns (Pt/Pt-1 - 1) for this analysis. What about Credit Spreads - credit spreads is like a return - so should I take log(spread) or simply use the spread. Lastly , the aggregate from daily to weekly or monthly , what functions should I use - Do I aggregate the log returns or the simple returns - do I take a simple sum or a cumulative aggregation - and how would I do it in R I am inclined to use the log returns for everything as regressions, correlation calculations all assume normality. But then to aggregate I would need to use the original returns (or spreads) - aggregate cumulatively - and then take the log ? [[alternative HTML version deleted]]