Implicit conversion from float -10.187265917603 to int loses precision