Is it possible to get negative values of probability from a cumulative distribution function?
The probability of an event can never be negative. This makes sense as probability is informally defined to be the likelihood of an event happening, and if it definitely will not occur its probability is 0 i.e there is no likelihood of its occurrence