Implicit conversion from float -101.18466898954702 to int loses precision