Implicit conversion from float -25.714285714285708 to int loses precision