Implicit conversion from float -15.42986425339366 to int loses precision