Implicit conversion from float -48.69565217391306 to int loses precision