Implicit conversion from float -10.469798657718115 to int loses precision