Implicit conversion from float -17.98365122615806 to int loses precision