Likelihood of extreme returns in number of days if data would follow a normal distribution

Question

I would like to know if the approach of coming up with the solution for following problem is viable:

I have a datatset of 3538 trading days of a stock. What I would like