Implicit conversion from float -27.015706806282708 to int loses precision