Everything you say is true. What you seem to be ignoring in your analysis is the bulk of the evidence.
It's not a done deal until a successful P3 trial or the unlikely PA. Having said that there is, as is said in civil trials, a preponderance of evidence that at least one or more trials will be successful.
I suggest that the odds are much better than 50/50 at this point.
The safety record of 2-73 in all trials is much better than most, if not all, CNS drugs and that overcomes many of the hurdles CNS drugs need to overcome for approval.
That alone increases the odds of approval.