November 18th, 2007, 6:41 am
to my knowledge, discounted F and H processes are not Q-Martingales but the discounted total gain process is a Q-martingale, i.e. dG = dF + Hdt has a drift rate r under Q measure. (Think of a dividend paying stock price process, under Q it has a drift r-y, where y is the dividend yield and the actual payout is y*S. Substituting back we have dS + y*S dt has a drift rate r.)