(Apparently it is possible to
recognize a broken humerus as velociraptorine with
enough certainty to leave no place for a
question-mark. Intriguing indeed.
Not at all. You see, a cladogram is not an opinion, it's a result.
Cladistics programs don't output question marks. If you want probabilities
for clades, analyze the support (bootstrap, jackknife, Bremer... or Bayesian
analysis in the first place).