Implicit conversion from float -371.7391304347826 to int loses precision