Implicit conversion from float -15.766423357664237 to int loses precision