Many issues with these opening statements, but first maybe, is that since a common σ calculation is but a special case of a GARCH set-up (for example, the limit as measured heteroskedasticity tends to zero), and any GARCH-calculated σ is going to carry biases by construction. Why would you...