Implicit conversion from float -92.41379310344828 to int loses precision