Implicit conversion from float -9.827586206896555 to int loses precision