This is all speculation on my point. Just getting that out of the way first.
But in doing my very prospect centric training stuff I've noticed that players unlock their PA in chunks every 2 years. How much is unlocked in each chunk is random, some players must have a flag that makes them always unlock it later. Because some player that I've seen in excess of 50 times always suck until after they're 19/20 and then either become a good AHLer, or an NHLer based on their random PA value in that specific save. When your scouts are scouting the NHL draft, the first thing they will see is obviously the player in his current "phase" which is what i'll refer to each 2 year window of PA as from now on. If that player rolls a very high portion of his total PA to be available, but only have a fairly low total PA once he's hit his final phase. That's how draft busts in EHM happen, your scouts mainly/only end up seeing that player's current, and maybe next phase. And his PA is comparably really good for both those phases. But then beyond that there is just nothing, on the flip side you have players with very low portions of PA in their draft phase and following phase. They will scout as being awful and end up being 6th/7th round steals. You sometimes can catch the AI sniping these in like round 2-3. Sometimes the Scouts will see only the current phase, sometimes they will see more phases. And sometimes they will get false positives and just report something categorically false. That's scouting for you.
This is how you sometimes get that player where 6 scouts that looked at him all year long say "like Eric O'Dell" and one say "like Matt Duchene". Either that guy saw into future phases of a draft steal, or he's drunk. I choose to believe the former, and thus gamble on a 14% chance at Duchene rather than a 100% chance at <insert random 4th liner> with my later picks. It generally won't pay off, but when it does it's pretty glorious.
If you pay a lot of attention around the draft you'll also notice there are some extra layers of deception while in the actual draft. If, like I used to, you looked at every single individual prospect ages 17-19, with Determination of 12+, and work rate+teamwork of 7+. Then make a short comment in their note about which traits they have, as well as their % chances of which player they will become. You'll notice that on draft day your scouts will give you completely different feedback than they did the day before. Sometimes they will say he's worse, sometimes better. Then as soon as you draft the player, it will change your scouts reports from their "draft reports" to their report based solely on their current phase. This is why literally every single player you draft outside top 3 looks like a draft bust right after you picked them an look at their scout reports. Don't freak out at this, it happens every year. Over time they will re-establish real scout reports and your player will hit his later phases so your scouts can see what's actually in the kid. Another way to get an inkling of if the game is telling you lies after the draft is to open up trades and look at the prospects star ratings in trades, if it's 3 star it's still up in the air. If it's 4 you know the scouts are drunk, he's a top 6 player for sure. This is also incidentally how you can trade 5th-6th rounders for 2nd rounders on draft day with the AI. They too get the trash scout reports on their own players after drafting them. Which means pretty much every single prospect from mid/late 1st round to 6th round will end up being rated as 3 star in trades. So you can just pull off 5th rounder for 1st rounder 1-to-1 trades sometimes on draft day. Which is so realistic>.>
Oh and star ratings are 100% useless in the NHL, look at the player comparison. The star rating is so vague and arbitrary that it might as well not exist in the NHL. In Europe it's somewhat useful, but in the NHL Draft anyone who isn't a generational talent normally ends up as a 3 star in scout reports no matter how good he probably will become. The fact that the game considers Burmistrov in the 15-16 db a 5 star player puts into perspective how little the star rating means. Crosby and Burmistrov have the same rating if going by stars, let that one sink in. Then never look at star ratings again and rejoice!
My other rival theory on how scouting prospects work is that your scout can get back any result in the player's PA range in the db, so if you scout a -15 (110 to 190) you could theoretically get back any result in that range. The better your scouts the smaller the "scatter" from his actual PA. So that 1 in 7 Duchene would probably be a -14 (90 to 160) having one scout scatter all the way to 160 (and possibly beyond, if he's real drunk) while the others all end up in the low 100s aka. <insert random 4th liner>-territory. I have a lot more faith in the first phase theory though, since that one ties into how I know the training aspect of prospects works in practice (pun not intended).
Ultimately it doesn't matter which it is. From a player standpoint you make the exact same decision regardless of which it is. The question is still if you gamble on the highroll, or go for the safer more boring pick. Since prospect conversion rate to top 6 players in EHM is very, very high if you know how to draft properly. You'll end up with more top 6 (and bottom 6) players than you could possibly fit on your team regardless. So highrolling to try and get Duchene in the 5th round is the better pick in my eyes every single time. And when in doubt, just ask yourself WWTDRWD (What would the Detroit Red Wings Do) and draft a random swede and pray
