Implicit conversion from float -82.22222222222223 to int loses precision