Implicit conversion from float -41.03693181818181 to int loses precision