These results seems interesting. They match with other observations. But couldn't those results be interpreted as a probability of origin instead of a date of origin ? It seems by those data that China is the most probable source, but not the only possible one. You seems to estimate the apparition of covid-19 in China as November 17 while also allowing it to be possibly in Italy as soon as Nov 3 (2.5% chance). Is there any reason to exclude it ? The origin point seems very movable still.

On another note, doesn't the technique used rely heavily on the assumption that equal efforts were produced to detect covid-19 all over the world. Which seems erroneous to some extent. I remember that very little tests were made in America during January or February. Your results seems to show that it could have been done if it was tried. On the other side, you have a first detection in Wuhan but it was realistically one of very few labs that would have done a genetic sequencing of a coronavirus unprompted. Which most likely played a huge role in the fact that it was detected there first. It's not quite like observing a bear in your backyard.

All this to say that the 95% certainty claims seems a little presumptuous. By shuffling a bit the point of origin and factoring for the difficulty of detecting covid-19, specially in early stage, I'm sure we could have a good chance to push the origin date in september, maybe even august.

These dates seems good IF covid originated from Wuhan. But those results also show, it might not be the case.