Both systems have their fair share of major discrepancies from the Associated Press or other "human" polls. I was going to whip up a chart to display some of this information for you, but lo and behold, Lone Star Stats' Kevin Cooksey already did the heavy lifting. Some of the discrepancies I'm referring to (say, Hanner's ranking of Michigan at No. 44) can be accounted for by random kinks in the model. Some of them are harder to reconcile with what we know about the affected teams as of today.
Take UCLA. UCLA is a very good team on paper. Assuming Shabazz Muhammad and Kyle Anderson eventually get eligible, they'll be added to a team with a ton of frontcourt size, another highly ranked recruit in Tony Parker, and a capable backcourt coached by a guy who, let's not forget, coached grinding defensive teams to three straight Final Fours. But as Dan wrote, the computer can't smell fear -- fear of Larry Drew II's attitude issues at North Carolina, fear of Josh Smith's inability to get in shape, fear of the Muhammed/Anderson eligibility nightmare scenario. So you see UCLA ranked No. 5 in aggregated rankings, in the top five as an average of Dan and Ken's systems, and No. 13 by both the AP and USA Today polls.
I'd wager this is working in the opposite direction with Louisville. On paper, the models see a very good defensive team that was downright bad offensively for much of last season. The best predictor for a team's offensive efficiency is its offensive efficiency last season, and Louisville wasn't very efficient on offense last season. Even allowing for some offense, that No. 103-ranked efficiency is bound to provide some drag in computer models. So you see Louisville, many experts' first or second choice to win the national title in 2012-13, ranked an average of No. 9 by Ken and Dan.
As in the case of Indiana, which Pomeroy ranked third, getting too caught up in preseason rankings of any kind is splitting hairs. And besides, being ranked in the top 10 to start the season is compliment enough. If there's a range there, Louisville is well within it. But I do think it's fair to perceive the Cardinals as being a good bit better than the computers say. Maybe I'm fooling myself, but I expect their offense to improve for a variety of reasons -- more touches for Chane Behanan, better polish from Gorgui Dieng, 3-point shooting from Luke Hancock, even better offensive rebounding with Montrezl Harrell coming off the bench, the emergence of Kevin Ware and the breakout of Wayne Blackshear.
Maybe those things won't happen. Or maybe Peyton Siva and Russ Smith won't become more efficient offensive players, won't stop turning the ball over, and it will all become moot. These things are all hard to quantify, and they are merely rough-around-the-edges expectations.
The point I'm trying to hammer home here is one that's been made over and over again by people who use tempo-free basketball statistics regularly and those who decry them as some sort of baseball-level robotic takeover of a once-unknowable-and-therefore-beloved pastime, even if they don't realize how much they agree: You can't get anywhere without your eyes. The statistical models and their in-team breakdowns are incredibly, incredibly useful for understanding what happens when a team takes the court, but they're only part of the battle. You have to combine them with a more innate knowledge that comes only from watching and re-watching the games themselves.
Many readers of this space are already long since on board with this, and will be wondering why I'm rehashing it here. I'll tell you why: Because we've not yet reached the start of another season, and I'm already, preemptively tired of seeing all-or-nothing blather about advanced statistical analysis. Don't feel scared, guys: Your deep, wizened basketball knowledge still serves you well. But when you're looking at points per game and saying a team is good at defense, maybe consider adjusting for pace? It's important. Deal?
Anyway, that's why Louisville fans -- or Michigan fans, or even Indiana fans -- shouldn't get too freaked out about any particular model mapping their team significantly lower than they'd expected this season. We always have a lot of knowledge blanks to fill in, and never more so than in the preseason.