This post is off topic:
i'm genuinely interested in seeing stats measuring how possible reading is in mafia. I have seen some players track their stats by submitting a guess before the end of the day and then checking how often their guess was correct, but they tend to lack consistent or clear methodology and so its hard to get anything substantial from those. They'll also typically include games with mechanics in them, and so things like cop results could end up skewing the results a lot.
There was also a kind of discord werewolf variant played where one player is given a gun and has to choose to shoot one of two players, and wins if they shoot the guilty player. In this game there is literally no winning strategy, mathematically speaking, as all players are operating on no information aside from testimony. Apparently in this variant some players had high enough winrates that they couldn't be explained by random chance, although in real life studies surrounding lying have generally shown that people can not tell when others are lying, which would suggest this kind of game shouldn't be winnable in a vacuum and that something else, like my misunderstanding the mechanics or players throwing as a joke sometimes could explain the high winrate.
Regardless, I've found very little in the way of whether or not its possible to read people in mafia at all, let alone whether or not its possible to read people based on their opening post, so I'd be surprised if there were stats that existed that rejected the latter proposition.