Implicit conversion from float -58.62068965517241 to int loses precision