As much as I love Bader's work, keep in mind that his model is limited by three main factors:

1. All that is predicted for is point production. He is only predicting how many points a player will put up. A "star" is just a player who passes a certain career scoring threshold. Slafkovsky, though I personally am not high on him, may bring things that don't show on the scoresheet. JFresh's prospect equivelency model does not have this limitation and measures stars based on WAR.
2. Top prospects often have obvious potential that is not reflected in their production. Past the second round, or even the late first, you could blindly draft based on Bader's model and be very successful. At the top of the draft, however, you should put far more stock in the eye test. Fantilli is, without question, the second best player available in the draft. Carlsson, who you mention as being relatively poorly rated by Bader, should unequivocally be among the top picks of the draft.
3. All that is measured is point production. Hutson is an excellent prospect, I will not object to that, but just like NTDP teammate Seamus Casey, there is a possibility that his game does not translate due to his size and toolset. Beating up on college hockey is definitely a green flag for any prospect, but the level of play is insanely high in the NHL. Of the 74% that profile like Hutson and turned out as stars, how many were of his stature? I am a fan of Hutson and of Casey, but it is important to understand the reality of being a small defenseman who has yet to make it.

Edit: I'll also throw it out there that NHLe does not differentiate between NCAA divisions, which biases it against BIG10 prospects and towards those in weaker divisions

These numbers completely perplex me.

Cole Caufield has already proven, at least to me, that he's close to star quality already. He's so good that you guys shouldn't entertain any thoughts of trading him. Meanwhile, Hutson isn't even signed yet. Moreover, Caufield and Hutson are both undersized -- Caufield is two inches shorter but 15 pounds heavier, yet Hutson plays the more physically demanding position, which also requires longer to learn. (Some forwards never develop more than one side to their game, as Mike Hoffman's career attests, but the Paul Coffey days are over.) So quantifying their respective chances of "reaching star player status" as Hutson having a 50% better shot than the already-tried-and-almost-proven Caufield is just cuckoo. Let me put it another way: anyone who'd rather have the untried Dragicevic than the nicely developing Kaiden Guhle has a, shall we say, peculiar analytic taste.

You’re right, choosing an unproven commodity like Dragicevic over an established player like Guhle makes no sense if your goal is to ice an NHL team in 2023, but that’s a moot point. I’m no expert on the matter, but if I got this right what the model is saying is that you wouldn’t be fooling yourself to expect a better production from Dragicevic at the NHL level provided his development keeps trending in the right direction – he’s been significantly outperforming Guhle in his D-1 and D0 years.

The Caufield vs. Hutson ratings don’t shock me: Caufield’s 47% chances to become a star player are still pretty darn good. Plus that doesn’t seem far off if you look at his actual production: he had an OK D+1 season, a fantastic D+2 season, but has regressed a bit in his D+3 season. Yes, that was mostly on coaching, but its just logical that the model shows some reserves as to whether he’ll reach star level production. As to whether the Hutson hype will translate to the NHL is totally up in the air, but I believe the model does factor in position played, size and weight.

I agree with everything you say here and also in your Team Description. My discomfort is with the model, not you (or your explanation of it, which seems entirely accurate.)

Let's round some numbers off for ease of calculation: Hutson, 72%; Caufield, 48%. Half of Caufield's 48% is 24%. Therefore, this model suggests to us that Hutson has a 50% greater chance of being a star player than Caufield. The flaw in this analysis is that we already know what Caufield's numbers mean: he's the real deal. So to me, to imply that Hutson has better chances of being a star than Caufield is, to say the least, optimistic. I can see that right now, he could easily be projected to be the next Torey Krug or Sam Girard, but to conclude that he has the same potential to be an All-Star and award contender as Caufield just seems wrong. To put it another way, I find it strange that one could predict a greater than 50% chance that Hutson (or anyone else, for that matter, other than a top 5 pick) would be a candidate for the Calder or Norris Trophies in their careers.

Just my opinion . . . I've already been wrong once today.

Thanks for the feedback. Of course, these models have their limitations, but you have to respect the fact that they're the best available tools to estimate potential outcomes. This being said, I agree with what you’re inferring, a better educated guess would lie at the intersection point of different performance indicators.

On your 2nd point, I believe you’re referring to what can be called “intangibles”? To me, that’s precisely the point of a model like this one: if you want to draft an NHL player, give extra points to mature/gritty/physically imposing players if that suits you, but if you’re looking to draft a star player, stick to the facts and look at past production.