Implicit conversion from float -19.163701067615662 to int loses precision