Implicit conversion from float -1.4772727272727195 to int loses precision