The chance for a particular mutation isn't more or less depending on whether it's neutral or not. They are all independent events. What it does mean is that the phenotypic expression moves in chunks. You can see how, at the simple level, a group of two mutations can move en block into a selectable feature. The demand for the selectable intermediate is removed. Even greater steps are taken at the protein level. Now, add the hierarchies and networks where a single mutation can effect changes for multiple features and you can quickly see how big morphological changes can take place without selectable intermediates.
Sigh. Again and again Creationists make this simplistic fallacy, again and again their mistake is pointed out to them and explained in detail.
Then they make it again.
All you've computed is the odds of N specific mutations occurring *in a row*, with *no other mutations occurring*.
Sure, that would be practically impossible, mathematically enormous odds.
However, your mistake is in assuming that the "N" mutations have to happen one after the other without fail, bam-bam-bam.
This is *not* what evolution postulates, AT ALL. Therefore, your result has absolutely nothing to do with the actual likelihood of successful evolution. It's just mathematical masturbation.
Instead, evolution postulates that somewhere, among thousands or millions of individuals in a population, amid countless other mutations occurring within the population during reproduction, beneficial mutation "N1" happens to occur.
Your calculation includes a term (P1) for N1 happening, FIRST TRY, AND NOTHING ELSE BEING ALLOWED TO OCCUR (which is of course highly unlikely). This is where you make your first major error.
Instead, you need to calculate the odds of mutation N1 happening, *at all*, over a huge number of individuals, across a large number of generations. This is, needless to say, much less unlikely. In fact, over sufficient time, it approaches certainty.
Then, your calculation presumes that mutation P2 must happen ON THE VERY NEXT MUTATION EVENT. This is, again, highly unlikely. But that's not what evolution postulates, so that's your error #2.
Instead, you simply need to calculate the odds of mutation P2 happening *sometime* in the future, to *any* of the countless generations of offspring which happen to inherit mutation N1 from the original individual in which it occurred. Note that it doesn't have to be the very next mutation event, as you presume, it could happen in the fortieth generation, among any one of thousands of descendants, after 100,000 other mutations had occurred within the population.
Again, note how much more likely this is than your presumption that the "N" mutations must occur *exclusive* to any other.
And so on through the rest of the N mutations.
If you're going to use math to analyze something, please be sure you first understand the process you're attempting to model.
Class dismissed.