It's just acumulating the independent probabilities of the unrelated events
Are you sure they are independent? These aren't dice, binomial distribution may not apply. Considering this is a network i would doubt the trials are independent.
It would take some rather deep analysis of the process to see what the dependence is, but I expect the
lowest probability of getting de-anon'd should be the independent probability case since any relation between the transactions should make them easier to de-anon, given proper analysis.