Implicit conversion from float -9.403747870528107 to int loses precision