Outliers should be excluded when producing a trend line, regardless of the method used.
The student got a score of 100 with no sleep.
This data pair was a rare event.
If the outlier was included, it would cause the trend line to misrepresent the typical data relationship.