Implicit conversion from float -9.051724137931032 to int loses precision