Implicit conversion from float -14.602184087363497 to int loses precision