Implicit conversion from float -52.36790606653619 to int loses precision