Implicit conversion from float -18.128834355828218 to int loses precision