Implicit conversion from float -26.285714285714292 to int loses precision