They are still mentioning this 15-20% chance of Darapladib being approved. No idea what math is being used to calculate this number... seems arbitrary.
$1.5B spent on the trial; $3B spent on HGS; new test in combination with Midazolam...
GSK has had plenty of opportunities to end this trial if it was going badly. They didn't need to concentrate ownership by buying HGS. Sounds to me as if GSK has more than 20% confidence that this will pass. And failure would be such a gross error that I suspect that people will be fired for wasting that $4B.
Ignoring the science, and judging GSK's behavior, suggests to me that GSK is more than 50% confident in this.