Implicit conversion from float -20.190615835777123 to int loses precision