Implicit conversion from float -9.587628865979383 to int loses precision