Implicit conversion from float -111.50943396226415 to int loses precision