Implicit conversion from float -133.72013651877134 to int loses precision