Implicit conversion from float -15.412844036697251 to int loses precision