In his research paper on valuation of financial firms(http://people.stern.nyu.edu/adamodar/pdfiles/papers/finfirm.pdf), the NYU Stern Professor and renowned valuation authority Prof Ashwath Damodaran gives a few insights on why DCF is not practical here. Besides the heavy regulatory dependence(on revenues, costs, leverage), banks also suffer from the handicap of classifying cash flows. Simple looking metrics like interest, capex, debt all need to be abandoned or redefined in case of banks.
But DCF has been the holy grail of valuation, so even if we abandon it how can we take anything else without some base? Well, for a going concern, the book value is the very minimum that the share price can fall to. Not coincidently, one of the most common metrics used is price to book value, where the excess>1 is defended based on intangibles, earning power, P/E and other factors. But it is agreed that these all are relative, and you won't see a DCF for a bank anytime soon. So is that model fundamentally flawed? Given that valuation is inherently subjective, this is not too bad a bargain.