Implicit conversion from float -9.816176470588246 to int loses precision