I am not an Alumnus of Math. Fin. at Imperial College, but I know people who have attended it recently.
Definitely there is a lot of abstract work to be done, but the MSc is open only to people from Maths and Engineering, so I don't think it is the main problem.
The main problem is that, as a course, it is orribly organised. You will need a strong grounding in probability theory and stochastic calculus, so what they tell you, i.e. that a general maths background is fine, is not true. It is fine to get a low mark with a lot of sweat, but not to excel, as probability theory and stochastic calculus will be given for granted, completely. I think they should have been more honest, and organise pre-courses on these topics, or at least make this requirement clear, rather than just expand their pool of students by fishing in Maths faculties generically.
Also, beware: the fact that there are famous teachers does not necessarily mean they will deliver a good course (which is what you pay for). I know people complaining that they have done say American or spread options BEFORE doing anything on vanilla options, or introducing the Black-Scholes formula. And quite often they didn't have notes (or correct notes)/exercises with solutions etc. Also, together with the famous teachers, there are a number of lecturers which go there just to get "Imperial College" on their CV, meaning that the pool of teachers is not stable. This messes up notes, exams, etc. as there is no continuity. And even information on deadlines, events or class cancellations (so, pure administrative work here) is poor. In general the course is quite messy on this front, which is disappointing, given how much it costs.
That said, placement is excellent. If, in exchange for placement, you are ready to take very interesting topics with very (and I mean "very") poor service, that is the place to go. Be ready to sweat a lot, and not necessarily because of what you study, but because of poor organisation.
