KenPom before a game was played predicted us to finish right at .500 in conf play, which is astounding, and has us now up to #14 in the country. Why the massive jump from last year? Let's look more closely at the first four games to see if anything pops out.

Opponent KenPom final or current rank in parentheses
2022:
W Idaho St(#245) 60-56, barely beat a terrible team that went 11-21, lost the second half.
L @SDSU(#14) 82-75, easily our best game to start the year, led much of it
W Missouri St(#156) 66-64, I think Hall saved our butts with a clutch layup at the end, one point lead at halftime
W Nicholls (#263) 87-73, this was our first relatively easy win of the year, but still not a terribly impressive result at home against such a bad team

2023:
W Houston Christian (#358) 110-63, demolition like a D3 exhibition game
W SDSU (#16) 74-65, was a pretty close game until we wore them down at the end
W SE Louisiana (#217) 105-48, demolition like a D3 exhibition game
W Morgan St. (#348) 93-50, demolition like a D3 exhibition game

Why at first glance maybe we should be worried we are overrated:
We have clearly played a weaker schedule so far this year, so that is certainly helping us look better, maybe better than we are? KenPom says beating a tough team at home isn't nearly as impressive as a neutral or road victory, so it's possible our close loss against SDSU in 2022 was as valuable to the computers, or more so.

So why are the stats loving us then?:
KenPom says that there is valuable information to be drawn from margin of victory. He's shown that you can with some accuracy predict the second match of a home/away pair based on MOV in the first matchup, and there are significant differences in team strength registered by 20 vs 30 vs 40 point margins. The danger there is just reading too much into possible noise from a single game, but that the data gets better with more games input at such margins. We have won 3 games by more than 40 points, and beat the predicted spread in all four games. That is causing the KenPom system to notice us.

I'm not sure if this relates to KenPom's formula, but in one research paper I read they decided to use defensive rebounds, turnovers, and assists as the best three stats predictors of game outcomes in points. Here are those numbers for BYU for each game of the four, both years.

(NOTE: my mistake I used total rebounds here instead of DREB, will fix it later)
2022 RB-AS-TO:
42-9-23
37-10-20
42-10-13
54-19-21
175-48-77 total

2023 RB-AS-TO:
59-27-9
42-13-14
52-28-5
41-24-11
194-92-39 total

Twice the assists, half the turnovers! That's a remarkable turnaround. Slightly better rebounding too. Can all of that be just explained away by the easier schedule this year? I don't think so, and apparently the computers don't either. That's because the difference in expected margin of victory for the most average D1 team over a team ranked #221 vs a team ranked #307 is only about 6.5 points less. It's not 30 points less. In other words, if BYU were the same team this year as last we wouldn't be beating Idaho St. by 4 and then beating Morgan St. by 43. We'd be beating Idaho St. by 4 and beating Morgan St. by something like 12.

We're not the same team.