They are generated as the season transitions to the next year. I've had a few requests to create the draft class during the playoffs and would like to do that but it will take a bit to get that working and tested (don't want to accidentally allow teams to sign the new class the previous year :) )

I think it would be excellent to have access to the draft class earlier, even as early as the start of the regular season (would certainly make potential trades better informed).

As a far off future suggestion, that could be an interesting way to implement a scouting mechanic as a new staff position (Head Scout). Hired from a separate pool (to avoid coaches and scouts bouncing between roles, which would just be silly), they could either share the same attributes as the standard coaching staff or have ratings by position group (ie QB, WR, TE, HB, OL, etc). Players would also be able to specify how their scouting department allocated its time via a 1-10 metric for each position group.

At its first generation, the estimated attribute ratings of players would have a moderately high margin of error, but as the year progresses, the range drift on the draftees would begin to tighten to a more realistic representation (as your scouts work, the players play, combine/pro days happen, etc), the degree to which would reflect your scout's abilities in that area as well as how highly that position was weighted for their evaluation. By end of season spin, this shift would be completed, and however you view the players would be thereafter set until you get them into training camp.

In this system, you may very well end up sitting at the end of the second round looking at a highly rated DE, but since it wasn't a position of need for you last season, you had set your scout to 0 for DEs and thus if you give in to temptation may end up with a guy who after more careful scrutiny ended up not living up to the hype.

The overall impact should be subtle, because no one likes too much gamble, but it would be a very nuanced way of mildly differentiating each player's perception of the draft class, which most definitely reflects the reality of the real league (just ask Jerry Jones about Manziel, compared to all the other QB hungry teams that ended up passing on him).

I'm sure ideas for scouting have been broached before, but that was just where my train of thought led me, sorry!

That's a really interesting idea. I have in my list something similar (where the numbers during scouting have some measure of error), but I like your idea about targeting players more specifically to have a greater range of error based on the position group as you define the priorities. I'm attaching your comments to that todo item.