That dude's take was so bad he like retired from doing stat posting on twitter lol.
If your model tells you something that doesn't make a whole lot of sense, maybe you should consider that your model may have a blind spot instead of just believing it blindly. If it's tell you that a consensus top 20 pick is actually only the 175th best player in the draft and has a <1% chance of making the NHL based on his scoring not being too good in 15 games in a Swiss pro league during a pandemic as a 17 year old, I think a logical person would think "hm, maybe I should dive deeper on this particular player because there's no way my model can be this far off" but too many of these guys fall in love with their models and legitimately have the idea that they're right and everyone else is actually wrong in a situation like this.