Implicit conversion from float -9.441964285714278 to int loses precision