Implicit conversion from float -11.386861313868621 to int loses precision