Implicit conversion from float -55.00693481276005 to int loses precision