Implicit conversion from float -93.06818181818184 to int loses precision