So if perimeter defense is just shot contest ratings shouldn't the corresponding statistic to look at be something like this.
https://stats.nba.com/players/defens...PerMode=Totals
Only filter applied is for Games Played to be 41 or more but you could filter by a certain number of contested shots instead simply as a way to ensure the data you look at has the small sample outliers taken out.
But with how I have it filtered, Gary Harris has the biggest differential between opponents FG% vs what they shoot when he is defending (shots greater than 15 ft) with a -10.4% differential, yet he only has an 82 perimeter D rating as shown above.
To compare, Eric Bledsoe who has a 93 perimeter defense and sports a -0.3% differential. At the other end of the spectrum Trae Young had a +2.0% differential and has 55 perimeter defense.