Implicit conversion from float -11.538461538461533 to int loses precision