Implicit conversion from float -9.137931034482762 to int loses precision