Implicit conversion from float -10.374531835206 to int loses precision