Implicit conversion from float -155.56451612903226 to int loses precision