Implicit conversion from float -40.082987551867205 to int loses precision