Implicit conversion from float -153.4077079107505 to int loses precision