At my previous work place, we used to calculate CVA by considering the potential positive exposures as an option to enter into a swap and the potential negative exposures as an option to cancel a swap at different points in the future. The time periods were allotted in a way corresponding to the payment schedule of the underlying swaps. And then accordingly, one would use the counterparty's credit spread to discount positive exposure while using using entity's spreads to discount negative exposures. I want to understand the theory behind this in a detailed way. However, if I search for CVA methodology, all I find is the MC method. Is anyone aware of the swaptions way of calculating and could you direct me to a paper/book that explains the method in details? Will Jon Gregory's book have this method?

Just write down the equations:[$] \begin{align*} \mbox{CVA}(t) &= (1-R) \int_t^{t_m}{\mbox{EE}(s) d\mbox{PD}(t,s)}\\ &= (1-R) \int_t^{t_0}{\mbox{EE}(s) d\mbox{PD}(t,s)} + \int_{t_0}^{t_1}{\mbox{EE}(s) d\mbox{PD}(t_0,s)} + ...\\ &= (1-R)\sum_{l=1}^{m} \mbox{EE}_t(t_l) \mathbb{P}_{\tau \in \left[ t_{l-1}, t_{l} \right]}\\\end{align*}[$]If you consider a plain vanilla interest rates swap:[$]\begin{align*}\mbox{CVA}(t) &= (1-R)\sum_{l=1}^{m} \mathbb{E}_t^{\mathbb{Q}} \left[ e^{-\int_t^{t_l}{r(s)ds}} \mbox{IRS}(K)^+ \right] \mathbb{P}_{\tau \in \left[ t_{l-1}, t_{l} \right]}\\ &= (1-R)\sum_{l=1}^{m} P(t,t_l)\mathbb{E}_t^{\mathbb{Q}^{t_l}} \left[ \mbox{IRS}(K)^+ \right] \mathbb{P}_{\tau \in \left[ t_{l-1}, t_{l} \right]}\\ &= (1-R)\sum_{l=1}^{m} P(t,t_l) \mbox{SWAPTION}(t,t_l,K) \mathbb{P}_{\tau \in \left[ t_{l-1}, t_{l} \right]}\end{align*}[$] However, with this method, you cannot capture netting effects occurring at the portfolio level. That is why you want to use a Monte Carlo method.

yeah it can't be done at the portfolio level. thanks.