Implicit conversion from float -13.846153846153854 to int loses precision