Implicit conversion from float -10.260223048327134 to int loses precision