Implicit conversion from float -10.236966824644554 to int loses precision