Implicit conversion from float -15.405405405405403 to int loses precision