Implicit conversion from float -10.399361022364232 to int loses precision