Implicit conversion from float -30.285714285714292 to int loses precision