My wild guesses are that MNTA has a 40% chance to prevail on the obviousness/double-patenting argument, a 30% chance to prevail on non-infringement, a 10% chance to prevail on indefiniteness, and a 5% chance to prevail on inequitable conduct. If you do the arithmetic, the above probabilities yield a 64% chance for MNTA to win the case by some method.
I am surprised you consider inequitable conduct that unlikely. From memory, I thought that TEVA had some significantly different descriptions in the EU patents, and this was not disclosed in the US application.
OTOH, I think you are too high on obviousness. The courts seam to defer to the USPTO on this. (Though you have this combined with double patents, which could be very strong).
EDIT: I assume from these you do not expect a summary judgement next spring.