Literally any moment in the next 10000 years. The probability distribution (afaict) is pretty flat, and either 1000, 10000, or 1000000 years long.

Part of the problem is that direct observations of supernovas is small and indirect. We have small numbers of recorded occurrences to judge from.

Also, we don’t have observations of the stars before supernova. Most are in other galaxies so don’t even see the star. Betelgeuse is first case where we can observe late stage star closely.