Implicit conversion from float -58.27586206896552 to int loses precision