Implicit conversion from float -11.716738197424888 to int loses precision