Implicit conversion from float -22.714285714285708 to int loses precision