There is one big caveat here and that is that we implemented the new training rules only one season before we started using college generated players. This means that I can’t really separate the two effects. I.e. The differences I’m about to show could be at least partially driven by the new training program that allows for things like training rookies and training current ratings up to 10 points per training and over multiple categories. That being said, I think we should consider the differences between the two groups as the difference between game generated players under the old training regime versus college generated players under the new training regime.
Points
This always has been and always will be a league that is all about buckets. The first graph below plots the average points per game of players playing more than 15 minutes per game for both computer and college generated players. The second does the same thing but with points per minute.
Points per game
Points per minute
As you can see. New players come in averaging more points per game and increase at roughly the same rate with the exception of the second year. This makes me wonder if the new fad of sitting super raw super prospects for at least one season is impacting both the first and second year averages. The first year is higher because those guys are sitting on the bench and the second year doesn’t grow as much because those raw guys get thrown into the mix. It’s a thought that may be worth pursuing in a future article. You can see essentially the same pattern for points per minute except that the game generated players don’t really catch up towards the end like they do in the points per game graph.
Rebounds
Rebounds per game
The rebounds per game picture suggests that the game generated players were superior rebounders. This does not surprise me. How often do we see guys like Moses Barger, A in offensive rebounding, B in post defense, and a D in defensive rebounding. This is frustrating characteristic with the new players that did not really exist with the old players. If you were tall and strong and played power forward or center you could rebound. Period. Note again that we see that little second year dip for the college generated players.
Rebounds per minute
The picture here isn’t as clear. The back and forth of these two lines is likely a function of the fact that I have much more data on game generated players so the line is smoother. It will be interesting to see how this pans out moving forward.
Assists
Assists per game
Arguably the most interesting thing here is the persistence of the second year dip. Other than that we just see college generated players growing way faster. This looks like a training effect to me. We get these raw point guards (unless you’re trying to draft one this year. They apparently didn’t have any point guards in the entire NCAA) and shell out points to get them up to speed quickly.
Assists per minute
This picture is a little clearer but also consistent. These new players are better passers that get better at passing faster. This also isn’t surprising, the randomness of the college players gives a lot of high passing potential ratings to big men. That was exceptionally rare with the game generated players. We’ve also had some college generated drafts that were loaded with point guards.
Steals
Steals per game
Ahhh, they don’t make ‘em like they used to. These college generated players can’t steal well at all. I may long for the days of Mark Macon and pretend like they don’t make ‘em like Macon but the fact is the performance at the top is still the same so this vastly inferior performance must be driven by lower performance in the middle or at the bottom.
Steals per minute
You can see the steals per minute graph tells essentially the same story. These new players just don’t generate turnovers like the old ones.
Blocks
Blocks per game
It looks like the college generated players block at about the same rate as the game generated players. I wouldn’t put too much weight on the last two years where the new players have more blocks per game, those are the observations with the fewest data points so they are pretty noisy.
Blocks per minute
Essentially the same story here.
Attributes
After all that it made me wonder, are we just imagining that these players are more raw? They don’t perform like they are more raw. So, let’s have a look at the attributes rather than performance.
Total current attributes
The initial year they look about the same but after than the game generated players have total attributes about 25 points HIGHER than those of the college generated players. The second year dip thing is still curious. There has been a fair bit of complaining (some by yours truly) about the athleticism of the new prospects so split the attributes by skill and athleticism.
Total skill attributes
Total athleticism attributes
Interestingly, old players have both higher current and higher athleticism attributes. This seems weird. If the old players have the higher attributes, why are the new players better scorers and maybe better passers. Perhaps the more interesting split is by offense and defense.
Total offensive attributes
Total defensive attributes
Aha. Essentially all of the difference between the game and college generated player attributes is accounted for by defensive attributes. Hence, it is very possible that the new players are not putting up better scoring numbers because they are better scorers but rather because they are going up against weaker defenses.
Minutes
Finally I was curious if the new players were getting more minutes sooner than the old players. This is another way to measure how much they are developing, are they earning playing time.
Minutes per game
The college generated players are playing slightly fewer minutes per game in the early years of their careers. Whether this is driven by worse athleticism, defensive liabilities, or fouling I can’t really say.
So there you have it. College generated players are having a bigger impact on offense earlier but that may be because they are all so bad at defense.