The problem with a probability analysis of Spoils is that it fails to take into consideration that Spoils will increase the win % based on being such an efficient tutor. In the case of Belcher, if you don't win on the first turn, you lose regardless, so you don't care if Spoils kills you or not.