Implicit conversion from float -29.714285714285708 to int loses precision