I want to test the hedging performance of several term structure models. I have daily swaption and swap rate market data over a period of five years.I thought of the following very simple experiment:On the date of a swaption's inception construct a portfolio which is delta-neutral according to the model and has zero value. Re-balance this portfolio according to the model with a self-financing trading strategy until the swaption's expiry date. On this date record the portfolio's payoff (which would be zero in the case of perfect, continuous hedging). Do this for all individual swaptions and compute the standard deviation of these terminal portfolio values.In the literature I have also seen experiments where the daily changes in value of the "replicating" portfolios are analyzed rather than the terminal values. But with this method I see the problem that these intermediate values are not driven by real cash flows (contrary to the terminal payoff). Instead they depend on the (interpolated) market price of the swaption which might be wrong or irrational.What do you think about this from a methodological standpoint?