I have no idea what kind of maths you're using, cub. My assumptions are based on the premise that each player's likelihood of leaving early is independent. There is no slotted number in each outlook category, they change every year.
I dont know the exact ranges, but I am confident that you are wrong on two of your estimations; first, that players listed as "staying" have a standard 1% chance of leaving early. If there are about 5 players listed as "staying" each year, and each of them only has a 1% chance of leaving, we would only expect to see one of those types of players leave early in a world once every 20 seasons. Basically, it would be the same as getting 5 chances to get one specific number from 0-99. I don't follow it that closely, but anecdotally from what I've heard on forums and conference chats, players listed as "staying" leave more often than once every 20 seasons.
For similar reasons, am quite sure that the "likely going" category does not go all the way up to 100%. I suspect it tops out at about 80%, max. Might be lower. One could do a study to see how many of each category - as listed before the end of the season - end up going, and I've never done that. But nothing has led me to believe that any given player is a "sure thing" to leave early.