Implicit conversion from float -31.240601503759393 to int loses precision