Implicit conversion from float -155.45454545454544 to int loses precision