Implicit conversion from float -98.24057450628365 to int loses precision