I am curious for products with varying variance with time to expiry (Samuelson for example purposes) and regular contract rolling, is there industry baseline for model of covariance structure?
I know many are recommending in past to use constant maturity schemes for continuous contracts to handle “rolling” issues of jump in data. However I am thinking this will make correlation seem very high since it is interpolation from real contracts.
However using actual contracts seems to be problematic if contract is growing in volatility or similar with passage of time. Shall I do this anyway (no constant maturity) and simply downweight older data?
Best regards!
