I would suggest that the PG class percentiles are not the "best of the best". Class of 2017 is showing a best of 95 (outier) with an average of only 78. That would imply that there are lots of 2017's throwing slower than 78 - but can also be the result of some prior year's speeds being pulled forward. There can be a pretty wide age range in these classes (upwards of 24 months from youngest to oldest), but PG does help demonstrate the pretty side range. You can look up individuals and start to get an idea where the 70th, 80th, etc. percentiles sit. I'm not saying the 50% of 2017's throwing slower than 78 are highly successful (many of these may have stopped pitching altogether years ago but still have old speeds tracked) but they do represent some very average players who happen to throw at PG events. There may be some data set elsewhere that may be better - and the data you can readily access off of the PG site is limited, but if you poke around enough on PG, including looking up individual players and scanning tournament pitching speeds, you do start to get a feel for a class' speed on the high end and low end. I always figured something around 75% percentile meant you were hanging in there from a speed standpoint. Looks like 82 (both RHP and LHP) appears to be the 75% for current 2017's.
In looking at my 2019's class percentile rankings, I've been thinking that this still represents an early view - only a small number of his class has actually been gunned at this point. So I've also looked at some 2016s and 2017s for comparison. It didn't occur to me that these percentiles would include throws from 2016s and 2017s when they were freshman (or before?). It would seem logical that PG's class percentiles would drop measurements over a year old or so. And, drop previous measurements for a player when he's gunned later. How does this work?
My understanding is that the system at PG, at the free level, works on an if then computation.
IF you are the class of 2016, and you have a data point (pitching speed/pop time/60 time) THEN you get included into the averages...even if that data point was from 2012! I think as 2017 said, in order to eliminate the extra data you would need a paid subscription.
My son did a PG event at 12u and there was a team from Texas that had 5 pitchers all pitching over 75mph, and 2 hitting over 80! These were NOT normal 12 year olds!