Jmeizis, you can add one more factor into your little hypothetical equation that will radically improve its usefulness, without requiring a supercomputer to calculate the probabilities:

Equalization

Let's say you are capable of creating an anchor that equalizes somewhat (distributes would be a better term than equalizes).

The point of the anchor is to be capable of holding the worst possible fall - a FF2, which can easily generate forces up to 18kN on the anchor.

Now let's look at your pieces again, and say that you estimate that each placement is 90% likely to fail at 18kN, 20% likely to fail at ~10kN, 10% likely to fail at ~7kN, and 2% likely to fail at ~5kN.

A single piece anchor gives you around:
90% chance of failure.

Two pieces, equalizing at 40/60 give you around:
20% x 10% = 2% chance of failure.

Three pieces, equalizing at 30/30/40 gives you around:
2% x 2% x 10% = 0.004% (four in 100,000)

Add a fourth piece, equalizing at 25/25/30/20, and you might see something like:
2% x 2% x 2% x 1.5% = .00001%

So three looks like the magic number to me.

GO