Implicit conversion from float -93.91304347826085 to int loses precision