Implicit conversion from float -421.42857142857144 to int loses precision