Implicit conversion from float -72.77227722772278 to int loses precision