Implicit conversion from float -10.858034321372855 to int loses precision