Actually no, it doesn't. Rob's said it (win goal / loss limit) also depends on how he's doing for the year and other stuff.
Of course he'd try to squirm out of his system being run by a simulation with some bogus excuse that the "machines aren't random". How much evidence does he have that supports this claim? Even with the assumption the games aren't random, it'd still definitely be possible to code a simulation, with the non-randomness part included in the code (i.e.: when holding 4 cards, make the re-draw card be the same as the original 5'th card more frequent than expected on a random game).