Implicit conversion from float -18.052434456928836 to int loses precision