Implicit conversion from float -98.07339449541283 to int loses precision