Implicit conversion from float -23.03656597774244 to int loses precision