Implicit conversion from float -92.97413793103448 to int loses precision