How does it make sense that they pick them as a #3 seed while they also say they are the 6th best team?
Short version --
These models predict your seed line not just based on your ranking divided by four and then rounded up. They take the mixture of criteria provided to the commit when making its decisions:
Analytical rating (e.g., BT, KenPom, NET, etc.)
Win and loss quality through the quadrant system (their projections of the same)
Head-to-head matchups
Raw win and loss total
Conference
etc.
Sometimes these can give you conflicting messages (e.g., sorting out where to rank a team that racked up a ton of wins a weaker P5 conference and blew people out so their analytics look good but doesn't have a really attractive win like going into a gym like Hilton or Phog Allen and coming away with a win... versus a team that didn't win as many games in a VERY strong conference... but the wins they do have are stellar and they have few "bad losses" on their CV... with the analytics giving you no strong indication between the two).
(What I've described is roughly what Iowa/Iowa State the first few years of TJ's tenure, by the way.)
Figuring out how the committee is going to weigh those factors is more art than science. And it can change year-to-year in unpredictable ways. And that is before you have to account for the committee making the bracket with locations and moving teams up/down to avoid rematches until the second weekend.
It's hard. The models do their best. I've noticed the
Team Rankings one tends to favor the "old school" wins and losses approach compared to Torvik, which leans more heavily into analytical quality.