Implicit conversion from float -8.50187265917603 to int loses precision