Implicit conversion from float -8.698224852071007 to int loses precision