Implicit conversion from float -22.440944881889763 to int loses precision