Implicit conversion from float -88.34394904458597 to int loses precision