Implicit conversion from float -93.02107728337236 to int loses precision