Implicit conversion from float -9.327052489905796 to int loses precision