2012 NCAA Basketball Preseason Top 25: Did The AP Whiff On Michigan?

posted in NCAA Basketball

After a long weekend poring over spreadsheets full of stats, ratings, recruiting data, and transfer lists, we’re ready to publish our official college basketball preseason projections.

This article is the first of several college basketball posts we’ll make this week. Today we’ll start with our preseason top 25. Later we’ll publish the full list of 347 teams, along with information on why each team is ranked where they are. We’ll follow that with conference projections.

Finally, we’ll be revealing a cool new feature that we created over the summer, which we’re really excited about. But that’s staying a secret for now!

Just a quick reminder to start: the ratings below are completely data-driven, with no manual fudges. The main inputs to the system are past team ratings, rosters and player stats from the past few seasons (which allow us to calculate returning offensive and defensive value, as we did last year), and recruiting info. So if your favorite team is over or underrated, rest assured it’s not because we hold some grudge against them. We’re just reporting what the historical data is telling us.

Without further ado, here is our preseason top 25. Below the table we’ll discuss a few of the teams for which our projections differ significantly from conventional wisdom, as well as reviewing how our preseason ratings performed last season.

TeamRankings 2012 NCAA Basketball Preseason Top 25

TR RankAP RankTeamConfConf W-LOverall W-LSOS Rank2011-12 RnkChange
13KentuckySEC15-327-4441--
21IndianaBig Ten13-525-5419+7
310FloridaSEC14-425-61610+7
44Ohio StateBig Ten13-523-7192-2
513UCLAPac-1214-425-54054+49
67KansasBig 1213-523-724-2
714Michigan StBig Ten12-623-8273-4
88DukeACC13-522-7814+6
9MarquetteBig East12-621-8513+4
1023WisconsinBig Ten12-622-868-2
1119BaylorBig 1212-621-81715+4
1211N CarolinaACC13-522-7185-7
1315MissouriSEC13-523-6807-6
142LouisvilleBig East12-621-8319+5
1521GonzagaWCC14-224-511525+10
169SyracuseBig East12-623-8386-10
1718UNLVMWC12-424-610524+7
18Saint LouisA-1012-424-58518--
19New MexicoMWC12-422-73517-2
20Kansas StBig 1212-620-82322+2
2112ArizonaPac-1213-522-65245+24
2222Notre DameBig East12-623-75043+21
2317MemphisCUSA13-324-59812-11
24PittsburghBig East11-721-75162+38
256NC StateACC13-521-83939+14
Other AP Top 25 Teams
2616CreightonMVC14-423-68929+3
3320San Diego StMWC11-520-810166+33
395MichiganBig Ten9-917-113131-8
4724CincinnatiBig East9-919-112533-14
5225Florida StACC10-819-113623-29

Three Overrated College Basketball Teams

Before we get into discussing over- and underrated teams, let’s explore one quick question: Should you even care which teams we think are overrated?

The (limited) data suggests you should.

Last year six teams ranked in the preseason AP Top 25 did not make the TeamRankings preseason top 25. Of those six teams, only one even received a single vote in the final AP poll from last season, and three finished 49th or worse in our end-of-season predictive power ratings:

TeamAP preseasonTR preseasonTR finalAP final
Xavier144549no votes
Arizona162951no votes
UCLA174270no votes
Alabama193631no votes
Cincinnati213226(27)
California243533no votes

One more key team not listed here is UConn. The AP poll had the Huskies 4th in the country prior to the start of the season, while he had them 11th. They finished 40th in our predictive ratings, and didn’t receive a single vote in the final AP poll. On balance, if a team was ranked in the AP preseason Top 25 but not in our Top 25, it was bad news for that team.

So what teams are we raising red flags about this year?

Michigan (AP #5, TR #39) — This an enormous difference, bigger than anything we saw last season. A quick check of other data-based projections reveals that most are more down on Michigan than the AP is: Pomeroy has Michigan 12th, and Dan Hanner has them 44th. So we’re not totally going out on a limb here.

Why are the Wolverines so low? The answer seems pretty simple, really. They were 31st in last year’s ratings, which form the baseline for this system, and they have only an average amount of returning minutes compared to the teams that were rated ahead of them. They have a couple borderline impact recruits, but nobody that’s a game changer, and have no incoming transfer help unlike several teams ahead of them. As a result, we are projecting them to be roughly as good as last year.

We wouldn’t be surprised if Michigan is better than the 39th best team in the country. However, it would be a pretty big bucking of past trends if they are really the 5th best.

North Carolina State (AP #6, TR #25) — We’re guessing the issue here is that NC State’s solid finish to last season (their only two losses in their last 8 games were by 3 points to UNC and 2 points to Kansas) is biasing people. Humans are prone to such “recency bias” as we stat geeks like to call it. Our ratings, however, weigh the 18-11 start to the season almost as heavily, so the Wolfpack are starting with a much lower baseline.

They do return quite a lot of value-weighted minutes, but simply returning a lot of minutes on a good-but-not-great team doesn’t make them great. (See Vanderbilt, last season.) On top of that, none of their major contributors last year were freshmen, whom are generally the players who make the biggest year-to-year improvements. So we see a mild improvement in NC State’s rating, but not the huge jump the AP voters are predicting.

Louisville (AP #2, TR #14) — Louisville ranked only #178 in raw offensive efficiency last season, which is very bad for a team aiming for elite status. This season, they lose Chris Smith and Kyle Kuric, two of their most efficient players. That means a heavier load will be placed on the shoulders of a few players that were very inefficient last year, particularly Peyton Siva and Russ Smith.

Unless those players show massive improvements, we just don’t see how the Cardinals’ offense can possibly be good enough for them to rate #2 in the country.

Three Underrated Colege Basketball Teams

Our preseason ratings didn’t do quite as well at identifying underrated teams as we did overrated teams last season, but we still picked out a few sleepers. Half the teams that were in our preseason top 25 but not in the AP’s ended up as borderline top 25 teams at the end of the year:

TeamAP preseasonTR preseasonTR finalAP final
UNLV(38)122823
Belmont(36)1424no votes
St. Mary's(39)163624
Purdue(35)1925no votes
Temple(28)2239(26)
Washington(29)2450no votes

We also picked out Missouri’s improvement before AP voters did. The Tigers were 10th in our preseason poll but only 25th in the AP Top 25. At the end of the year, the Tigers were 7th in our predictive ratings and 3rd in the polls.

Here are three teams we think may be similarly poised to beat expectations in the 2012-13 season:

Marquette (AP #26, TR #9) — Out of all teams that went unranked in the preseason AP poll, Marquette is ranked the highest in our preseason ratings. It’s easy to see why the AP voters see the Golden Eagles declining. Losing Jae Crowder and Darius Johnson-Odom leaves them with a big gap to fill on offense. Our projections see transfer Trent Lockett and rising sophomore Davante Garnder helping to ease the transition. We also project Marquette will maintain a high defensive level; they are above average in our returning defensive value metric, and Lockett had good defensive stats for Arizona State last season.

UCLA (AP #13, TR #5) — It’s important to note that our UCLA rating assumes top recruit Shabazz Muhammad will be cleared to play this season. If he’s not, the Bruins will drop down to a borderline top 25 team.  That seems like a huge drop for one player, but Muhammad is rated as the top recruit by some scouting services. Imagine if Kentucky last year was missing Anthony Davis.

We’re wondering if voters aren’t feeling a little buyers’ remorse from last season. They made UCLA #17 only to see them stink up the joint. The Bruins definitely were a disappointment last season, but they have a Kentucky-esque recruiting class this year. That injection of young talent should be enough to push them to the top of the Pac-12.

Wisconsin (AP #23, TR #10) — Losing Jordan Taylor is a huge blow to Wisconsin’s offense, and frankly, we’re a bit worried that our projection may not adequately account for that. However, the Badgers were actually ranked much higher in defensive efficiency (8th) than they were in offensive efficiency (42nd) last season, and their defense shouldn’t take as big of a hit in Taylor’s absence as their offense will.

Besides the fact their defense could be elite once again, it’s always dangerous to count out a Bo Ryan club — he’s nabbed a protected NCAA seed five of the past six seasons.

So, there you have it, our preseason top 25. What do you think? What other teams do you want to talk about? Are we completely whiffing on any teams? Leave any questions or comments in the discussion thread below.

  • thetomb

    Pomeroy has michigan ranked 12th. you linked to 2012

  • Ron

    Not sure who Peyton Silva is. Peyton Siva plays for Louisville.

  • http://www.teamrankings.com/ David Hess

    Thanks, not sure how that snuck in there. Fixed.

  • http://www.teamrankings.com/ David Hess

    Ha, whoops. I must have been thinking of http://beeradvocate.com/beer/profile/2743/62253 … Fixed.

  • Tony

    That was a good try at being smart about sports, anyway. Better luck next time.

  • http://www.facebook.com/profile.php?id=100001379163813 Steve Scott

    You mention Louisville’s poor offense last year, but what you fail to mention, is their kick-ass defense. Let’s face it, their defense has a whole lot to do, with what got them to the Final Four last year. Just remember, that defense of theirs’, is going nowhere! It will be back again this year, just like it is EVERY year! It hurt a little losing Kuric and Smith, but it wasn’t detrimental to Louisville. They have plenty of offense to choose from this up-coming year, but every team they play against, will still have to overcome their DEFENSE. Just ask a lot of their opponents from last year. Lastly don’t ever forget that old coachs’ saying, that has been around for many decades…”Defense wins ballgames!!”

  • http://www.teamrankings.com/ David Hess

    Oh yeah, the defense should definitely be good again. I guess part of the issue here we don’t think they were one of the top four teams in the country last year just because they made the Final Four, so they are starting with a lower baseline than some people would assume.

  • aaron

    how does this list account for freshman and impact transfers…

  • http://www.teamrankings.com/ David Hess

    For freshmen, we convert the RSCI rank of each player into a “recruit value” that is based on looking how equivalently ranked recruits from past seasons impacted their team’s rating.

    For transfers, we create a “transfer value” using the same player value metric that we use for returning starters. It’s a combination of offensive efficiency, usage rate, and minutes.

    Both of those values are used as inputs in a regression equation (along with past team ratings, info on returning starters, etc) that determines the team’s projected rating.

  • aaron

    with that being said do you think the computer is accurate on Cincinnati? Should we expect a mediocre season or is the excitement around here warranted?

  • http://www.teamrankings.com/ David Hess

    I don’t know that much about Cincy’s particular situation, but it seems like after one game, they played about to our expectations. Obviously that’s not enough to make any real judgment, but that’s about all I have to go on at this point.

    More generally, I will say that I bet there are more teams with fans excited about this season than there are NCAA tournament at large spots available. :)

  • aaron

    im sure…but cincy has been in tourney for last couple years and gradually getting better since starting over after huggins left. This is the most excitement for a Bearcats team since. They lost only Dixon and Gates but I assume since your computer model takes minutes played into LARGE consideration that hurts them quite a bit. Dixon played more minutes than anyone last year. also not sure how they were not in AP top 25 at years end after making sweet 16 and Big East final…..

    any sense here or am I off base with the computer model?

  • http://www.teamrankings.com/ David Hess

    Well, we don’t take straight minutes into account. They are weighted by how effective a player was on the offensive and defensive side of the ball, AND how many possessions a player used. Dixon used up a lot of possessions on offense, and Gates was fairly efficient, plus was a good defender.

    As for the end of year rankings, we had Cincy only 33rd, as the whole season is taken into account, not just postseason play. Generally, that is one of the biggest biases people have when forecasting a team. They weight the last few NCAA tournament games much heavier than any other games, and that doesn’t really seem to be justified, based on the data.

    That said, some preseason computer ratings have Cincy up around #30, so we may be underrating the Bearcats. Luckily they’ll play the season out, and we can find out!

  • FB in the D

    Well, so far Michigan’s living up to billing, but let’s see how they do against better teams.

    One HUGE weakness of objective rankings is lack of matchup sensitivity, for lack of better terminology. This usually shows up in individual games, but If the changes in a team shore up its weaknesses (e.g., physical presence, in Michigan’s case), the effect could be much, much greater than the sum of the parts, and that won’t be apparent until results come in. This is where humans have a BIG advantage, even with the rose-colored glasses.

    Even then (as with the football team, which has been consistently shut down by teams with good run defenses with Robinson at QB), a team’s performance could vary wildly depending upon the way the teams match up in any given game, especially if they’re very unbalanced, like last year’s team and the Robinson-led football team.

  • Hell yeah

    Yeah go syracuse Louisville is over rated

  • FB in the D

    How, if at all, does this account for effects on a team’s overall balance of strengths and weaknesses, as opposed to just their “rankings”?

    Regarding Michigan, for example, 2 fair-to-middling “bigs” could be of much greater value to such a team that had no inside presence to speak of last year, and therefore lived and died by the 3-pointer, than 2 much more highly rated shooting guards would.

    A team is do much more than the sum of the parts, and I can see how the actual value of new recruits can involve factors that may be too complex to plug into an equation.

  • http://www.teamrankings.com/ David Hess

    You’re right that sometimes recruits will fit particularly well with a specific team’s strengths/weaknesses, and these ratings do not account for that. Michigan this year may very well be a case of that.

    However, in general, I would assume that *many* teams try to recruit players that help them patch up weaknesses, rather than players that will sit on the bench because they already have talent at a position. So part of that effect will already be baked into the generic “recruiting factor” that we use.

    For the teams where recruits fit particularly well, I can only say that yes, of course no equation is going to perfectly evaluate every team. However, no human is either. And if you look at our results from last year (in the post above), you can see that we fared pretty well when compared to the AP poll. It’s not like we crushed the AP poll, but we certainly did no worse.

  • FB in the D

    Yeah, your models have indeed done quite well, no argument there. More often than not, a team’s (or program’s, as the case may be) reputation precedes them, and overrides logic, hence the inflated or deflated ratings.

    But things can get rather complex, beyond modeling capability. Though some of the things you’ve done on this site are really good, such as the “similar games” analyses, which account for the likelihood of a given result based on the particular strengths and weaknesses of the teams.

  • http://www.teamrankings.com/ David Hess

    Yeah, totally agree there will be cases where simple modeling just doesn’t cut it. I think on of the main benefits to this approach is that it can be applied across all teams consistently, without becoming an expert in every team’s situation.

  • brandon

    looking pretty dumb today