Implicit conversion from float -12.263083451202277 to int loses precision