Implicit conversion from float -10.382409177820264 to int loses precision