Implicit conversion from float -25.813953488372093 to int loses precision