Implicit conversion from float -10.851063829787222 to int loses precision