Implicit conversion from float -56.12068965517241 to int loses precision