Implicit conversion from float -23.669724770642205 to int loses precision