It's pretty much a given that your question has no solution since it's not calculable whether a set of numbers would pass or fail a randomization test. If it were, there'd be no need for such testing, you could predict the outcomes. Perhaps a more useful vehicle would be to define a few, fairly simple sets of "this isn't right" that are determinable in some way. For instance, you could look at the number of sequences which have two or more consecutive numbers. Not sure if it's possible to work out what the probability is there, but it seems a far more tractable problem than genuine randomness. After that you might figure out the odds on the list taking up any particular set of rules generating a non-random pattern and remove that as well. I think that's as close as you could come.