Implicit conversion from float -100.02493765586036 to int loses precision