Implicit conversion from float -27.709090909090918 to int loses precision