You cannot know the specific odds, but you can know for certain that they are on the side of God's not existing.
I could be wrong, but most worldly probabilities are calculated on the basis of frequency of repetition of an event. For example, the probability of a car crash in a certain stretch of road is derived from how many car crashes have happened there in a certain time (or perhaps per number of total cars).
How can we even make estimations on the odds of a one-off event or entity? We don't have other universes to study and figure out how many of them have omnipotent creators.
That is what is called the statistical 'definition' of probability. Today, probability is simply defined in an axiomatic way (due to Kolmogorov) as the following:
If
τ is the field of random events associated with an experiment and
P:
τ →
R is a mapping satisfying the following properties:
1)
P(
A) ≥ 0; - positive definite
2)
A∩
B = Ø =>
P(
AU
B) =
P(
A) +
P(
B); - additivity
3)
P(Ω) = 1. - normalization (Ω is the certain event)
From 2 and 3 it follows that:
P(A) = P(A U Ø) = P(A) + P(Ø)
P(Ø) = 0
That is, the probability of the impossible event (Ø) is zero. However, the converse, namely: "if the probability of a certain event is zero, then that event is impossible" is not true in general. This is obvious for the so called continous random variables. The probability of that variable for having an exact value is always zero, but, whenever an experiment is performed, an exact value is always obtained, i.e. the random event of the random variable having a definite value is not impossible.
Also, starting only from these axioms, one can directly prove the so called Law of large numbers, which gives the basis for the practical application of probability theory to statistical problems.