Implicit conversion from float -15.633802816901408 to int loses precision