Implicit conversion from float -8.214285714285708 to int loses precision