Implicit conversion from float -155.6896551724138 to int loses precision