i'm not going to argue with you directly; you made assumptions and came to reasonable conclusions from them. but, i'd argue it's more likely that a scheme such as this would create a groupthink shift of otherwise tory voters than a backlash of retreat to the conservatives. that is, if unhappy conservatives and red tory voters saw a unified force develop, i think they'd be likely to support it, and that could push the conservatives into the mid 20s. if you wanted to do this right as a probabilistic calculation, you'd have to integrate the possibility that it would take votes away from the tories into your calculations, as well. you're consequently working with an incomplete sample space, which is skewing your numbers - you're only considering a fraction of the possible outcomes.

www.huffingtonpost.ca/2015/09/30/vote-splitting-mulcair-trudeau-harper_n_8220820.html