Implicit conversion from float -15.980528511821987 to int loses precision