Implicit conversion from float -9.070631970260223 to int loses precision