Implicit conversion from float -15.666156202143952 to int loses precision