I'm open to the idea that there are other models which show Jackson projecting into a better player than Tatum. Why don't you try pointing to them instead of holding up the odd counterexample? Right now the best information we have available is that Tatum is the better prospect.
I actually wasn’t holding up the odd counterexample. I was directly responding to your drive-by comparison with Kawhi Leonard with what I think are particularly relevant comps, one being the player I hear most often cited as Tatum’s upside, and the other being Celtics’ current best player, either of whom would seem to be better “poster children” for the limitations of the free throw model.
As to what models I think would project Jackson to being a better NBA shooter (you keep using shooter and player interchangeably), I think their college 3pt percentage and FG percentage are a good place to start. Which is to say, I think you are placing far too much faith in models which even their developer admits are only “modestly successful” at predicting NBA 3-pt shooting. In many instances, of course, college percentage turns out to be a more accurate predictor. I happen to wonder if there is a pattern to those instances. For example, and most relevant to Tatum, I would like to know Johnson’s model’s track record when it comes to players it projects to be
better 3-pt shooters in the NBA than they were in college.
Reviewing Johnson’s initial description and application of his model, it appears that track record is pretty bad. Of the 21 players he discusses in his year one post, there were five players whom he projected to improve their FG3% in the pros, and who have subsequently taken at least 100 3PA: Marcus Smart, James Young, Elfrid Payton, Andrew Wiggins, TJ Warren.
[1] All five of them have NBA FG3% closer to their college percentage than their Johnson projection.
Player NCAA3% AJModel NBAFG3%
Smart 29.5% 34.8% 29.1%
Young 34.9% 35.7% 27.6%
Payton 27.2% 31.2% 28.9%
Wiggins 34.1% 35% 32.9%
Warren 31.4% 32.4% 31.2%
So, while I will trust Johnson that the overall data suggests that on average his model is a slightly better predictor than college percentage, I am far from ready to accept the model’s accuracy when it comes to projecting players to improve, as it does Tatum.
[1] Bogdanovich would also fit here, but according to the Johnson model, he shot 31.9% from 3 pre-NBA. Stats I’ve seen actually have him at 35.9%. Using this number, it again would likely have been more accurate than the Johnson model.
That weighting would also put Jackson ahead of Fultz and Ball. My point is not that shooting is the only thing that matters, but that Jackson's poor shooting puts him behind Tatum. I like Jackson's motor, but I wouldn't want to gamble on his shooting and Ainge didn't either.
I’d also love to know how it performs when projecting, as it does with Jackson, a differential of 5 percentage points or more,
[2] based on free throws.
Finally, I should say that I am bullish on Jackson, and one of the reasons is development and utilization throughout his one college season. Freshmen are by definition new to the team, system and coach, and the Johnson model ignores improvement as the player, team and coaching staff assess and use the talent. In his first 18 games at Kansas, Jackson was not being used in any significant way as an outside threat - he averaged 2.1 attempts per game, those attempts were usually breakdowns, and he shot only 23.7% from behind the arc. Over the final 17 games, the team and Jackson used his 3-pt shooting as a weapon far more frequently, averaging nearly 50% more attempts per game, to 3.1.
He shot 48.1% over that second half of the season. I tend to weight those games more than the first 18.
[2] Another data concern here. Johnson’s model has JJ shooting 36%, while his actual KU 3FG% is 37.8%.