Implicit conversion from float -31.614349775784746 to int loses precision