Implicit conversion from float -9.324324324324323 to int loses precision