A10 Prediction – Modeling Injuries, Transfers, Coaches

Future Tweaks to the Model

I developed my predictive model last spring and the model is still in the experimental stage. Similarly, Ken Pomeroy’s predictive model in the Basketball Prospectus book is humbly referred to as Projections, Version 1.0. There are still a lot of things we can and will improve in future years. Today I want to talk about some of the things I have in mind for the future.

Can we learn anything from this year’s rankings?

Looking back at all my conference predictions, there are a few teams that stand out as surprises. Teams like Michigan St. and Villanova are a little lower in my rankings than where most experts have them pegged. But that is largely because neither of these teams had dominating efficiency numbers last year.

But the ranking that bothers me the most is North Carolina’s low ranking. And I think Seth Davis’ recent review of North Carolina points out the current biggest problem with my model. My model does not adequately account for injuries. My model looks at Tyler Zeller and sees a player that could barely crack the rotation last year. But that is clearly wrong. Zeller was not being held out because he was not good enough. He was not playing because he was injured on multiple occasions.

Now, this is not such a simple adjustment as you might think. The model may under-rate Zeller’s return, but it is important to realize that not all players will successfully recover from injuries and return to a dominant level of play. Are players with ACL tears more likely to suffer future ACL tears? Are players with foot problems (see Zeller) more likely to continue to miss games in the future? Ideally we would have a database of injuries and project how likely players are to recover from each type of injury. But because that database does not exist, my future project will try to look at how well players recover from a “general” injury.

Second, we need to do more to account for transfers. Simply plugging in a transfer into the new lineup is probably not sufficient. And despite the occasional Wesley Johnson type player, we need to do more to understand how often players succeed in their new environments. For every Wesley Johnson, how many Alex Legion’s are there out there?

Third, there are some problems that will be solved when we have a larger sample size. Right now, I am estimating my model based on three years of returning tempo free player data. That is not a lot of data to draw conclusions about unusual situations. For a team that loses 2 starters, and 3 rotation players, the model probably does a very good job. But for a team that returns almost no players (Kentucky), there simply are not a lot of historical examples. I would like to have a statistical reason to treat Kentucky differently, but for now I am mostly making an out-of-sample projection.

Similarly, the three year data set causes problems because of some recent trends. In particular, teams without elite recruits have been having more and more success from 2007 to 2010. Let’s throw out Memphis, Gonzaga, and Xavier, because all three teams have been recruiting at a different level that most non-BCS teams. Look at the non-BCS teams in the top 35 of the Pomeroy Rankings in 2007 vs 2010:

2007
22nd Air Force
25th Butler
28th Southern Illinois

2010
10th BYU
12th Butler
22nd Temple
25th Utah St.
26th Dayton
29th Northern Iowa
34th Old Dominion

There are a lot more teams without elite recruits performing at a high level recently. (BYU’s Jimmer Fredette was not an elite prospect coming out of high school. Butler’s Matt Howard and Dayton’s Chris Wright were top 100 recruits, but they were the only elite recruits on these teams.) Thus the recent data tends to have more confidence in non-BCS teams than may be warranted.

But I also suspect this is somewhat cyclical. While the SEC fell off the map a couple of years ago, and the Pac-10 fell apart last year, I do not believe those leagues are permanently down-trodden. And as those leagues improve again, I think the recruiting data will start to have a little more predictive power, and I’ll start to rank a team like Wake Forest, a little higher than I do this year.

To deal with this cyclicality, I currently make an adjustment that moves non-BCS leagues downward. But I would like to have the data to determine the proper level for this adjustment. Right now, it is rather ad-hoc.

Finally, I want additional data so we can do a better job modeling how different coaches respond to different situations. We know Mike Brey has a special ability to teach offense; we know Bruce Weber has a special ability to teach defense; and we know Jamie Dixon has a special ability to bring young players along quickly. But modeling the interaction between coach and returning player effects will take more data.

A10 Prediction

Correction: If you have been following my blog closely, you may have noticed that Rhode Island showed up in the Biggest Departures Category in a recent post. That had me scratching my head. I knew Rhode Island lost Lamonte Ulmer, but the ranking seemed wrong. I recently went back and checked my code and found the problem. Even though I have the full rosters of eligible returning players, for some reason I included a line of code that classified all of last year’s “seniors” as departing players. Not only was this line of code redundant, it was also wrong. Many schools list players as seniors who are not really in their final year of eligibility. And Rhode Island had just this problem. Delroy James and Ben Eaves were both listed as seniors on kenpom.com last year, but both are listed on the Rhode Island’s roster again this year. I have now re-run the numbers for all conferences, and fixed the previous conference predictions. This only makes a meaningful difference for two teams whose numbers I have presented previously. First, I had mistakenly coded Notre Dame’s Ben Hansbrough as departing. With Hansbrough, Notre Dame is projected to be in the hunt for an NCAA bid. And given the way Notre Dame played without Luke Harangody last year, I think this is a very reasonable prediction. Second, Miami’s Adrian Thomas was also listed as a senior on kenpom.com last year. After fixing the code, Miami is now projected as an NCAA bubble team. I apologize for any confusion.

The first table shows the expected changes for the A10. Fordham has performed at such a hideously low level the last two years that it almost seems unsustainable for an A-10 team. Almost every player Fordham lost was among the worst in the conference, hence the positive number in the “players lost” column. Even for a team like Fordham, they should be able to replace players with efficiency ratings in the 70s with better options. Fordham will continue to be horrible this season, but with a batch of recruits that do not look like they should play in the MEAC, you have to expect at least modest improvement. Fordham will still be the worst team in the A10, but I suspect they will win more than two games this year.

Among the contenders, Temple is the most likely to improve. St. Louis was going to be the most improved team this year. They were a team that played a lot of young players last year and a team that was peaking at the end of last season. But Willie Reed and Kwamain Mitchell are not enrolled in school due to a recent legal issue. I have heard some speculation that at least one of them will return for the second semester, but for now I am assuming neither player comes back. And instead of being a 5th NCAA contender in the conference, St. Louis is another team that should slip back this year.

Thanks to the recent season-ending ACL injury to Brad Redford, Xavier is now expected to take the biggest fall in the A10.



The next table shows the expected changes in offense and defense. Xavier loses its two most prolific offensive options in Jordan Crawford and Jason Love, and both were very efficient as well. Plus they lose the great three point shooting of Brad Releford. While they return some other players who can rebound and defend, the model thinks Xavier’s offense will take a step back.



The next table shows the conference prediction. Temple is a logical favorite. They had one of the top defenses in the country last year, and while they lose a tough scorer in Ryan Brooks, the departing Luis Guzman was hardly an efficient player. But I am a little concerned that Temple may not have the depth to really get better. They gave a number of young players minutes last year, and outside the starting rotation, no one really stepped forward. In expectation, Fran Dunphy should be able to replace Brooks and Guzman’s production, but in practice I’m not sure where those replacements are going to come from. My model views a Villanova – Temple game as a toss-up, and I’m not quite as comfortable making that conclusion. But assuming Temple’s defense is better than Villanova’s defense, as it was last year, the teams should have similar efficiency margins once again.

Dayton loses a ton of players from their rotation. But their two most efficient and critical players, Chris Wright and Chris Johnson, are back. And some of the role players who are returning are also very efficient. (See Luke Fabrizius.) Dayton will depend on a solid recruiting class, led by Juwan Staten, to fill in the missing playing time. I’m a little concerned about integrating so many new faces given that Brian Gregory likes to play a deep rotation. But by leaning on the team’s two stars, Wright and Johnson, Dayton should be able to stay near the top of the A10 standings.

The A10 looks like a three-bid league, but Richmond is clearly in the hunt. At one time, St. Louis was also in the discussion for one of the top spots in the league, but the loss of two of their key players is devastating to their chances of becoming an elite team.

Are you certain BYU is not the MWC favorite?

I understand the arguments against BYU. Here is what Mike DeCourcy said recently. “Set For a Fall: No. 24 BYU. The Cougars have one of the nation’s best players in guard Jimmer Fredette, and they’ll be an NCAA tournament team. But even with Tyler Haws, Chris Miles, Michael Loyd and Jonathan Tavernari, BYU only made it into the NCAAs as a No. 7 seed.“

And I have certainly read the arguments for San Diego St. They bring back five starters. And they were a pretty effective NCAA tournament team too. But my model thinks BYU and San Diego St. are a toss-up for MWC champ, and here is why I think that is right:

First, if you believe in margin-of-victory, BYU was much better than a 7-seed in the NCAA tournament last year. BYU was 10th in the nation according to Jeff Sagarin’s Predictor rating, and 10th in Ken Pomeroy’s ratings, meaning BYU had some of the best margin-of-victory numbers in the country. And it has been well established that margin-of-victory is a huge predictor of future success.

And while BYU loses some of the players that led to those dominant margin-of-victory stats, the key players are back. BYU returns one of the most explosive and efficient players in the country in Jimmer Fredette. But they also return guard Jackson Emery whose 127.0 ORtg on 20% of the shots when on the floor would have people raving if not for Fredette’s success. Plus as will be discussed below, Emery and Noah Hartsock posted some fantastic defensive stats last year. Basically, BYU may not return as many players as San Diego St., but they return the most important players on offense and defense and there is every reason to think they will be dominant once again.

For everyone who thinks that Purdue should still be a top 10 or top 15 team despite the uncertain rotation, then you should be equally in love with BYU. They have the offensive and defensive stars, they just need to fill out the rest of the rotation. And as five straight years of first or second place MWC finishes have shown, head coach Dave Rose is plenty effective in developing players to fill out the lineup.

MWC Prediction

Once again, I start with the expected changes. As with Indiana, Air Force and Colorado St. experience a case of addition by subtraction. While Air Force will clearly miss forward Grant Parker, Mike McClain and Avery Merriex were so ineffective, as a whole the team should be better allocating shots elsewhere. McClain in particular was just a source of dead possessions for Air Force. He turned it over fully 30% of the time and shot just 24% while taking 45 threes. His block rate was less than 1% as the team’s center. Even with Air Force’s recruiting limitations, they can do better.

Man, I feel bad saying that. What did Mike McClain ever do to me? He did shoot 59% on his twos last year, so he did some things right. And yes, he probably has more athletic talent then I will ever have in my whole life. (Just putting that out there.)

Maybe I should not name any Colorado St. players by name. Colorado St. loses four players who could most generously be called role players. Two of them had ORtgs of 56.8 and 68.0. That’s really all I need to say.

New Mexico might be a little under-rated here. They obviously lose a ton of talent, but I think the model may under-rate what they have joining the team. True, they will only have half a season of Drew Gordon, and that will make the early numbers weaker, but his tempo free stats at UCLA were quite impressive. Also, Alex Kirk is only 98th in the RSCI top 100. Given the inconsistency of recruits at that level, the model does not expect much. But Steve Alford has had a pretty good eye for talent lately.

But here is the kicker for the model. New Mexico played three freshmen last year, but all three barely made it on the floor. In other words, they were mostly playing a deep experienced lineup last year that performed near peak performance. That is not the type of team that improves a lot the following season.



The next table shows the expected changes in offense and defense. BYU suffers some critical defections, but Jackson Emery, one of the national leaders in steal rate, and Noah Hartsock, one of the national leaders in block rate are both back. While the offense may take a small step back because they have to integrate so many new faces, because head coach Dave Rose has two critical defenders coming back, he should be able to keep the defense playing at a high level.



The next table shows the final prediction. The MWC looks like a three bid league, with a huge drop-off after the top. But New Mexico may be back in the mix by the end of the season. And Colorado St. returns five fairly efficient starters from last year and may be prepared for a surprise run.

College Alumni in the NBA

In honor of the NBA season tipping off tonight, I thought I would present the former college of all players on the opening day rosters. (I’ve actually seen some other links that have done the same thing recently, but I wanted to download the full rosters on opening day myself.)

The opening day rosters include:
59 International players that did not attend a US college
56 ACC
54 Big East
48 Pac10
41 Big 12
40 SEC
31 High School
29 Big Ten
12 CUSA
11 A10
10 MWC
9 WAC
5 WCC
5 CAA
5 Sun Belt
3 MVC
18 Other

-In addition to the 59 international players who did not attend college in the US, there are some international players that did attend college in the US, such as Australia’s Patrick Mills who attended St. Mary’s.

-Eventually the high school group should shrink because it is much harder for players to jump directly from high school to the NBA. But the 31 high school players still include the best players in the NBA from Kobe Bryant to LeBron James to Dwight Howard. Brandon Jennings is currently the only player in the “high school” club with less than 5 years of NBA experience.

-The rest of the talent has mostly come from the ACC and Pac 10. Because the Big East includes 16 teams, they have a number of NBA alumni, but the Big East has fewer alumni per team. And the Big Ten has the fewest NBA alumni among BCS conferences with only 29 on opening day rosters.

Here are the numbers by team:
14 UCLA
13 Duke
13 Kentucky
12 North Carolina
12 Kansas
11 Connecticut
10 Arizona
10 Texas
9 Florida
8 Wake Forest
7 Ohio St.
7 Georgia Tech
7 LSU
7 Syracuse
7 Memphis
6 Stanford
6 Michigan St.
5 USC
5 Washington

The next table shows the full rosters broken down by college conference and then by college team. Scroll down to see the full table.



If I had a lot of money and power and could build my dream NBA charity event, I would love to make an exhibition tournament and pit some of these alumni teams against one another.

-Imagine Rajon Rondo leading all those young Kentucky players.

-Is North Carolina’s NBA roster starting to look a little bit weak? Vince Carter might be the only “star” left and I’m not sure how much he has left.

-Could a Wake Forest team with Chris Paul, Tim Duncan, and Josh Howard compete with players like Jeff Teague and James Johnson filling out the rotation?

-Allen Iverson is finally retired and the days of Georgetown being able to fill out a reasonable starting lineup are now long gone.

Big East Prediction - Four Starters Back, One Star Gone

We are about a week away from the release of the Basketball Prospectus book which includes more words on all these teams, but in the meantime, here is my model's prediction for the Big East.

Once again I start with the expected changes. Due to lingering injury issues, Anthony Mason was unable to use a meaningful number of possessions for St. John’s last year. And St. John’s brings back virtually their entire roster. (Ironically, this is true for just about the third year in a row.) With almost everyone back, St. John’s should experience a slight improvement in efficiency, and Steve Lavin has a chance to make the NCAA tournament in his first season.

Cincinnati loses a pair of fabulous guards in Deonta Vaughn and Lance Stephenson, but Stephenson was not a very efficient player last year. (See 16 of 73 on threes.) The team also loses back-up interior defender Steven Toyloy who was not a particularly effective interior player either. The model tends to view the loss of ineffective players as a good thing. On the whole, by reallocating minutes and shots on a team that still returns Yancy Gates and Rashad Bishop, Cincinnati may play better than some people think.

My model also likes Oliver Purnell to turn things around for DePaul in his first year as head coach. There is no question he inherits a disaster, but he also knows a thing or two about how to rebuild. I fully expect him to get DePaul to play better defense this year, and I expect him to steal a few games in the Big East, despite his decimated roster.

The next table shows the expected changes in offense and defense. Notice the contrast between West Virginia and Syracuse. While both teams lose substantial offense, Syracuse’s defense takes a bigger hit. Syracuse loses Wesley Johnson, Arinze Onuaku, and Andy Rautins. Meanwhile West Virginia loses Da’Sean Butler, Devin Ebanks, and Wellington Smith. While West Virginia’s trio were actually more efficient offensively, the three Syracuse players posted significantly more blocks and steals, and their ability to change the game on the defensive end of the court will be missed.

The next table shows the prediction for the season. My model picks Pittsburgh to win the Big East with a similar top five to most publications.

In addition to rolling out a detailed Big East preview, Villanova By the Numbers recently posted the conventional wisdom on the Big East standings, and the consensus agrees with Pittsburgh as league champ. But my model departs from the consensus in three key ways.

First, Villanova is ranked substantially lower in my model. Villanova does have a natural replacement for Scottie Reynolds in Maalik Wayns, but the reason for the poor prediction here is Villanova’s poor defense last season. Two years ago they had three great defensive rebounders in Dante Cunningham, Dwayne Anderson, and Antonio Pena, but last year they only had Pena. Perhaps the loss of Reggie Redding and Taylor King will actually help the defense. Redding was often asked to play forward despite his limited size. And despite good defensive rebounding numbers, King had a reputation as a poor defensive player. But that assumes Wright will give more minutes to Mouphtaou Yarou and top recruit Jayvaughn Pinkston. Wright has shown he will not simply use a taller lineup to block a few more shots. For Villanova to play better defensively, those players will need to execute and earn Jay Wright’s trust. If Jay Wright uses a smaller lineup and Antonio Pena is the only dependable interior defender once again, Villanova may lack the defense to be a true NCAA tournament contender. Of course Villanova nearly won the Big East despite their defensive shortcomings last year, so it would be foolish to count Villanova out.

Second, my model ranks UConn substantially lower than some other sources. UConn loses its two most efficient scorers in Stanley Robinson and the highly underrated Gavin Edwards. And UConn loses the team’s most prolific scorer in Jerome Dyson. Kemba Walker will need to bring three freshman top 100 recruits along quickly, and my model suggests that in the Big East, that could be a recipe for a long season.

But the ranking that most makes me scratch my head is Georgetown’s position on this list. The Hoyas are ranked second in my model despite the loss of an NBA lottery pick in Greg Monroe. Statistically, I understand where this calculation comes from. Georgetown’s defense is expected to slip without Monroe, but Austin Freeman, Chris Wright, and Jason Clark, might be the most efficient returning guard trio in the nation. They were all incredibly effective scorers, and based on John Thompson’s ability to teach efficient offense, the model expects another precision attack for the Hoyas.

But this is the first time in about six years that John Thompson will be playing without a dominant big man in the middle. One has to wonder how much of Freeman, Wright, and Clark’s success the last few years was due to the attention Roy Hibbert and Greg Monroe drew in the paint.

And this leads to an important question. How well do teams do when they return the entire starting rotation, with the exception of one star player?

Four Starters Back, One Star Gone – What happens next?

Ohio St. and Georgetown are ranked very high in my model. Both teams return four starters from rotations that were incredibly efficient last year. But both teams also clearly lose their star player in Evan Turner and Greg Monroe. This brings up an interesting question. Historically, how have teams fared that lost their most talented player, but almost no one else from the lineup? Let’s look at some examples:

Texas: Lost Kevin Durant
Off 2007 - 120.6
Def 2007 - 94.6
Off 2008 - 123.8
Def 2008 - 91.8
AJ Abrams, DJ Augustin, Damion James, and Justin Mason were all efficient players with Durant, but all three played better after he was gone. And Connor Atchley improved tremendously after Durant left.

Texas: Lost DJ Augustin
Off 2008 - 123.8
Def 2008 - 91.8
Off 2009 - 112.4
Def 2009 - 91.6
All the players mentioned above played significantly worse in 2009 without their star PG. Only the emergence of Dexter Pittman kept Texas from plummeting further.

Boston College: Lost Tyrese Rice
Note: Boston College fell apart last year and fired their head coach. But according to kenpom.com, Boston College was very unlucky last season and had a very solid efficiency margin.
Off 2009 - 112.9
Def 2009 - 99.8
Off 2010 - 110.2
Def 2010 - 95.8
Without Rice, Rakim Sanders numbers fell off precipitously, and several other returning players suffered smaller drops in efficiency. Only Joe Trapini performed slightly better. The only reason Boston College was able to post decent efficiency stats on the year was the emergence of Tyler Roche and Evan Ravenel as effective role players.

LSU: Lost Anthony Randolph
Off 2008 – 105.4
Def 2008 – 95.6
Off 2009 – 111.2
Def 2009 – 94.5
Without Randolph, LSU saw a number of improvements. Bo Spencer’s efficiency rating increased dramatically and Garrett Temple and Marcus Thornton also played significantly better. LSU went on to win the SEC regular season title.

Georgetown: Lost Jeff Green
Off 2007 – 124.8
Def 2007 – 89.3
Off 2008 – 117.2
Def 2008 – 86.4
Roy Hibbert was forced to shoot more and saw his efficiency rating fall and everyone else on the team saw their efficiency rating fall slightly, with the exception of Jonathan Wallace. Georgetown did not return to the Final Four.

Kansas: Lost Julian Wright
Off 2007 – 117.8
Def 2007 – 82.2
Off 2008 – 125.3
Def 2008 – 82.8
Brandon Rush and Darrell Arthur got slightly better, and Mario Chalmers and Darnell Jackson improved tremendously after Julian Wright left. Julian Wright probably was not Kansas’s best player, but he did use the most possessions among all starters in 2007. Kansas won the national title in 2008.

Duke: Lost Josh McRoberts
Off 2007 – 113.9
Def 2007 – 85.6
Off 2008 – 118.2
Def 2008 – 87.6
DeMarcus Nelson, Gerald Henderson, Greg Paulus, and John Scheyer all played significantly better without McRoberts, but Duke’s defense did slip slightly without McRoberts in the middle.

Conclusion: There is reason for optimism. Several teams on this list did get substantially better in years in which they returned four starters and lost a star player. But I think the successful teams also had had a number of talented players who were ready to take on larger roles. DJ Augustin was ready to shine when he could break out from Kevin Durant’s shadow. Darnell Jackson was ready to shine when he could take Julian Wright’s minutes. I am not as confident Georgetown has a player as ready to step into a larger role and take the Hoyas to the next level.

But the important thing to note is that we have limited data to reach these types of conclusions quantitatively. For now, my model picks Georgetown for second in the Big East and Ohio St. first in the Big Ten. But in a few more years, as the seasons of tempo free player data continue to grow, future data should allow us to draw better parallels.

SEC Prediction, The Biggest Departures, and a BP Book Update

If you have been following the “Players Leaving” column in my tables, you’ll eventually determine that the top ten teams with the biggest losses this off-season are:

10. St. Mary's
Omar Samhan, Ben Allen, Wayne Hunter

9. Mississippi St.
Jarvis Varnado, Barry Stewart, Phil Turner, Romero Osby

8. Baylor
Ekpe Udoh, Tweety Carter, Josh Lomers

7. Marshall
Hassan Whiteside, Tyler Wilkerson, Chris Lutz, Darryl Merthie, Cam Miller

6. Nevada
Luke Babbitt, Armon Johnson, Brandon Fields, Joey Shaw, London Giles, Ray Kraemer

5. Kansas
Sherron Collins, Cole Aldrich, Xavier Henry

4. Syracuse
Wesley Johnson, Andy Rautins, Arinze Onuaku

3. California
Jerome Randle, Patrick Christopher, Jamal Boykin, Theo Robertson, Omondi Amoke, Nikola Knezevic, DJ Seely, Max Zhang

2. Cornell
Ryan Wittman, Jeff Foote, Louis Dale, Geoff Reeves, Jon Jaques, Alex Tyler

1. Kentucky
DeMarcus Cousins, John Wall, Patrick Patterson, Eric Bledsoe, Daniel Orton, Darnell Dodson, Ramon Harris, Perry Stevenson

No team in the nation lost more quality players than the Kentucky Wildcats. The key question is whether the top recruiting class in the nation will be enough to replace those players. But as you may guess from my prediction for North Carolina in the ACC, my statistical model is not confident that Kentucky will be able to make a successful transition.

As I noted in the ACC post, recruits have had a surprisingly inconsistent effect on team performance over the last three years. The key is probably that recruiting is endogenous. Elite recruits are much more likely to go to teams where they are likely to play. Kansas did not have a dominant recruiting class this year, but that was partly due to the fact that Kansas already had a number of under-utilized players who are ready to take on a much larger role for the Jayhawks. Meanwhile Kentucky’s successful recruiting class was partly based on the fact that they had few pieces already on the team that were ready to step in. Kentucky could pretty much guarantee playing time to recruits under all circumstances. But the lack of emerging backups should also temper our expectations for Kentucky this season.

Basketball Prospectus Book Update – An Alternative Model

But if you are not happy with the weight I put on recruits, I am pleased to announce that there is a numeric alternative. As I hinted at previously, Ken Pomeroy has developed his own predictive model for tempo free team performance and you can find it in this year’s Basketball Prospectus Book. While Ken and I use the same basic underlying factors (the Dean Oliver statistics for returning / departing players), his model puts a higher weight on previous seasons and a much higher weight on elite recruits.

So for the next week or so, you can continue to get a few free tempo free predictions from me, but due to the upcoming release of the book, I have provided very few words on the various teams.

But if you pick up the book, you will get
-A detailed statistical analysis of all major conference teams
-Some analysis of all the non-major teams
-Several insightful articles combining stats and basketball
-And Ken Pomeroy’s tempo free predictions for every conference in the nation.

Somewhere John Gasaway is carefully crafting the finishing touches on the publication, but the wait is almost over!

SEC Prediction

The first table shows the expected changes for the SEC. If it were not for John Calipari’s incredible recruiting ability at Kentucky, we could be looking at a repeat of what happened to Indiana a few years ago. (OK, maybe not that bad, but Kentucky lost a lot of talent.) I assume Enes Kanter will not be eligible in these projections, but even with Kanter, Kentucky would have a hard time duplicating last year’s success. That does not mean it cannot happen, but the expected value of any recruiting class is clearly less than 5 NBA first round picks.

While Kentucky is depending completely on recruits to step in, Florida both adds an elite recruit in Patric Young and gains from having given substantial minutes to Kenny Boynton last season. The model expects the freshman guard to develop into a much more consistent player this year.

Georgia’s recruiting class kind of fell apart due to ineligibility issues, but they bring back enough key players to get better this year.



The next table shows the expected changes in offense and defense. My model is aware that Dee Bost and Renardo Sidney will be joining Mississippi St. mid-season. (And Mississippi St. has used some creative scheduling to try to minimize the impact of the games those players will miss.) But even with Bost and Sidney, Mississippi St.’s defense is expected to get substantially worse this year. You simply cannot replace the top shot blocker in NCAA history (Jarvis Varnado) and not expect the defense to take a step back.

Offensively, the loss of Dan Werner is irrelevant to Florida. He was the team’s least efficient player. But defensively, the Gators will miss his high steal rate.



The next table shows my model’s prediction for the SEC. The Gators barely made the NCAA tournament last year but they bring back all their most efficient players and are the clear favorites in the league. Meanwhile, people tend to overlook the talent Vanderbilt brings back. And Alabama was very unlucky last season. The Tide may very well be the favorites in the SEC West.

Personally, I’m not comfortable with ranking Kentucky this low. In my contribution to the basketball prospectus book, I did not trust my own model and picked Kentucky for a higher finish. But even John Calipari admits his team could struggle early in the season.

More on Hummel

Ah, fantastic, we have some disagreement to fill the weeks until the actual season starts. In my last post I said Andy Katz’s decision to drop Purdue from 2nd to 23rd seemed about right. But Ken Pomeroy does not agree. He states that “If you added a healthy Hummel to Temple, I don’t think you would consider the Owls to be national title material.”

I understand the argument, but from a statistical perspective, I do not agree. I think if you put Hummel on Temple they would be among the top teams nationally. First, Purdue was projected as a Final Four contender because they play such incredible defense. But Temple played incredible defense last year too:

Purdue 2010
Adjusted Offense Rank = 70th
Adjusted Defense Rank = 3rd

Temple 2010
Adjusted Offense Rank = 75th
Adjusted Defense Rank = 7th

And from a lineup perspective, I’m not sure Hummel would not make just as big a difference to the Temple lineup as the Purdue lineup.

Consider Purdue’s key returning players this year:

F JaJuan Johnson 107.2 ORtg, 18.2% Defensive rebounding rate
G E’Twaun Moore 102.9 ORtg on 31% of his teams shots

The only other returning Purdue player with an efficiency rating over 95 was Ryne Smith who posted a 104.9 rating while playing 26.9% of the team’s minutes last year. Purdue also brings in just one RSCI top 100 recruit.

Compare that to Temple’s key returning players:

F Lavoy Allen 114.3 ORtg, 23.7% Defensive rebounding rate
G Juan Fernandez 108.7 ORtg, 73 made three pointers
G Ramone Moore 103.3 ORtg and very effective inside scorer as a freshman

Now these individual ORtgs are not adjusted for quality of competition, and the A10 defenses were a little easier to score against, but not as much as you might think. And I have every reason to believe that if you put a player with Hummel’s stat line on Temple, they would be an elite team.

F Robbie Hummel 122.1 ORtg, 20.8% Defensive Rebounding Rate

Even if Hummel’s offense only made Temple an above average offensive team, he is also a huge defensive force. Just look at that defensive rebounding rate! I have no doubt that with Robbie Hummel, Temple would be near the top of the rankings.

Of course, I would also agree that Temple would not be the 2nd place team in the nation. But my model did not pick Purdue for 2nd either. My model liked Purdue 5th with this Final Four:

Duke
Kansas St.
Ohio St.
Pittsburgh

Perhaps the argument could also be made that teams need some outstanding recruits to make it to the Final Four, and even with Hummel, Temple would not have multiple NBA ready prospects. And yes, there is evidence that recruits are important to a deep NCAA tournament run. Wisconsin is almost always among the national leaders in efficiency, but the Badgers have not been able to make the Final Four or stay in the Top 5 of the polls during the season. But I think evaluating a team based solely on recruits is wrong too. My favorite example of this is the Florida Gators, which have traditionally had great recruiting classes, but won two national titles with a group of less heralded recruits. (Plus Butler made the national title game with only one top 100 recruit last year.) Temple may not feel like an elite team, but winning is not all about having McDonald's All-Americans.

But I think that is mischaracterizing what Ken Pomeroy was saying. The key point is that in any given season, the difference between the top 5 teams and the bottom of the top 25 is huge. Whereas the top 5 teams are far out in the tail, teams from 20-40 in the national rankings are usually bunched together at a much lower level of performance. And I agree completely.

But, I’m not convinced that Purdue was so far out in the tail of the distribution that the loss of a player as efficient as Hummel will be easily overcome.

The Robbie Hummel Effect (Revisited)

[Note: I made a minor change to this post on October 29th. I apologize for any confusion.]

By now I am sure you have all heard the news about a devastating ACL tear. Penn St. forward Sasa Borovnjak has torn his ACL and is out for the year. Oh wait, that is not the ACL tear that everyone is talking about? No, the news that has saddened college basketball fans everywhere is that Robbie Hummel is out for the season.

I’ve already presented the tempo free prediction for Purdue with Robbie Hummel:
Predicted Efficiency Rank = 5th Nationally
Predicted Adj Eff Margin = 25.4
Predicted Adj Off = 110.1
Predicted Adj Def = 84.8

One thing we can do is run the model again and see what it predicts if we do not count Robbie Hummel as a returning player this season.
Predicted Efficiency Rank = 24th Nationally
Predicted Adj Eff Margin = 19.4
Predicted Adj Off = 106.6
Predicted Adj Def = 87.1

Andy Katz has moved Purdue from 2nd to 23rd in his top 25, and my statistical model roughly agrees with that prognostication. But you may notice that these numbers are still better than Purdue’s splits without Hummel last season.

I think the key difference is that replacing a player like Hummel is a lot easier in October than in February. With this injury taking place in October, Purdue will still have a number of practices, exhibition games, and early season cupcakes to build a team without their star forward. Making this adjustment in February when their opponents were peaking was a very difficult adjustment for the team to make in such a short amount of time.

Regardless, Purdue may now be the 5th best team in a top-heavy Big Ten, and that is a shocking development for a team that was a Final Four favorite in most preseason publications.

Big Ten Prediction – A surprise pick for champ?

A Note on Midnight Madness

I’m excited for Midnight Madness this weekend, but I caution anyone who tries to learn anything from the inter-squad scrimmage or any of the exhibition games that will follow. Exhibition games are not a great place to evaluate talent. As evidence I provide two anecdotes. Two years ago Maryland’s Jin-Soo Kim (now Jin-Soo Choi) scored 20 points in an exhibition game against Northwood University. The fans were shouting his name and there was serious talk that he was the surprise gem in Maryland’s recruiting class. But Jin-Soo went on to score a grand total of 41 points in his two years with the Terrapins. And if the exhibition games are not a great place to evaluate players, they are not a great place to evaluate teams either. Michigan St. lost to Grand Valley St. in an exhibition in 2007-08, but it did not stop the Spartans from finishing 15th in the Pomeroy Rankings that year. Enjoy meeting the teams, enjoy the atmosphere, and enjoy the theatrics this weekend, but do not take anything too seriously until the real games begin.

A Note on Ken Pomeroy

As I alluded to a few weeks ago, Ken Pomeroy is also in the business of modeling returning talent. I’ll have more thoughts on the various ways to make tempo free predictions in a future post, but for now I will simply steal his punch-line. The Big Ten is a league on the rise.

Big Ten Prediction

Today I roll out the 4th of my tempo free conference predictions, and no league has intrigued me as much as the Big Ten. The traditional consensus seems to have Michigan St. and Purdue as the best teams in the league, but Doug Gottlieb has picked Illinois to win the league this year. So when I first ran these numbers, I was very curious to see what the model would say. But before I get to the full conference prediction, let’s look at the expected change in efficiency margin. I list the losses due to players leaving and the other changes due to recruits, player development, and coaching changes.

First, note that Indiana’s rating in the Players Leaving column is not a misprint. Indiana has a positive rating for Players Leaving. Devan Dumes had the second lowest efficiency rating on the Hoosier’s last year (84.2) and used a ton of possessions. For the Hoosiers, this should be a case of addition by subtraction. By letting other players take Dumes’ shots, the team should be better. But the Hoosiers will improve for a more important reason. They gave playing time to a number of key freshmen last year. And those players are expected to play significantly better this year. That list includes Maurice Creek (who was injured and played in only 12 games last year), Jordan Hulls, Christian Watford, and Derek Elston.

Also on the upswing are the fighting Illini who replace the last few spots in the rotation with three top 100 recruits. The model likes last year’s freshman to develop, and it expects Brandon Paul to emerge for the Illini. Paul took a lot of shots last year, suggesting he has the confidence to become a star, but he was also the team’s least efficient scorer last year. Will he play with more consistency as expected, or will Paul find his spot in the rotation taken by another great recruiting class? Regardless, this is the most talented team Bruce Weber has had since his Final Four squad, and the pressure is squarely on Illinois to play at an elite level this year.

On the opposite end, everything went wrong for Michigan this off-season with Manny Harris, DeShawn Sims, and even Laval Lucas-Perry leaving the team. And Minnesota loses Damian Johnson, Lawrence Westbrook, and Paul Carter. Johnson’s loss really hurts because not only was he quietly efficient on the offensive end, he was a key force on the defensive end for the Gophers.

But despite these defections, the Big Ten is clearly a league on the rise. If the Big Ten finishes 5th in the RPI again this year, it will definitely be a disappointment.



The next table shows the expected change in offense and defense. Wisconsin loses its most efficient offensive player in Jason Bohannon, and the team’s second leading scorer Trevon Hughes. And despite Trevon Hughes’ high steal rate, those losses should hurt the offense more than the defense.

Fran McCaffery will need time to bring in more talented offensive players to Iowa. But based on his success teaching defense at his previous jobs, the model predicts Iowa will improve more on defense than offense.

On the other hand, Illinois should expect a bigger offensive improvement than defensive improvement. While the departures of Dominique Keller and Jeffrey Jordan will not mean much on the offensive end, those players did contribute on the defensive end for the Illini.



And what do all these numbers mean for the projected league standings? My projected winner may be a bit of a surprise. My model likes the Buckeyes to repeat as league champs. (OK, my model is not the only source that likes Ohio St. Blue Ribbon picks the Buckeyes for the Final Four. But this is far from the consensus pick.)

Evan Turner is gone, but the Buckeyes bring in an incredible recruiting class and return some phenomenal players. People just don’t have a handle on how well Jon Diebler, Dallas Lauderdale, William Buford, and David Lighty played last year. (I’m still having nightmares about Jon Diebler knocking down threes by the way.) Yes Evan Turner was good, but it was not all Turner. While the Buckeyes are expected to slip slightly, my model projects the newcomers will fill most of the void Turner left behind.

Some may be surprised to see Michigan St. picked so low, but it makes perfect sense when you look at the numbers. The Spartans were only fourth in the Big Ten in efficiency margin last season, and they do lose some key players in Raymar Morgan and Chris Allen.

To put it another way, Michigan St. was only a five seed in the NCAA tournament last year, and I’m not sure how wins against Maryland and Tennessee in last year’s NCAA tournament make them a clear Final Four favorite this year. Tom Izzo and his players deserve a ton of credit for what they were able to accomplish, but I think people tend to overlook the fact that the Spartans were not a dominant team last year.

Now, nothing about where I’m projecting Michigan St. says the Spartans cannot win the Big Ten or go the Final Four. But Tom Izzo plays a lot of players and builds a team that peaks in March and not November. If his team has another slow start and is not among the nation’s leaders in efficiency margin, it should not come as a surprise.



Penn St. was the second unluckiest team in all of D1 last year, and I think some people are under-rating the Nittany Lions. It will be no shock if Talor Battle keeps Penn St. in the NCAA tournament hunt this year.

And I’ll be cheering for Northwestern to finally break through and make the NCAA tournament, but the model is not confident. As many people have said, the defense has to take a serious step forward for Northwestern to be a serious NCAA tournament contender.

In the end, the Big Ten is a league with five potentially dominant teams, and three bubble teams. Assuming the league plays as expected in non-conference play, Minnesota, Northwestern, and Penn St. will have plenty of chances to earn RPI top 50 wins and earn an NCAA tournament spot. But I don’t think any of them are a lock. Minnesota is projected to finish in 6th place, but the Gophers would only be projected as the 8th place team in the ACC or Big 12. This is my way of saying that the Big Ten may be stronger at the top, but some of the other leagues may still be deeper this season.

Pac-10 Prediction

Once again, I start with the predicted changes. Cal loses a ton of talent this off-season. They not only lose the four players on the team that played the most minutes last year (Jerome Randle, Patrick Christopher, Jamal Boykin, and Theo Robertson), but all four were incredibly efficient scorers. This is going to hurt. On the other hand, things are starting to come together for Sean Miller at Arizona. Arizona played only two upperclassmen in the rotation last year, and the playing time given to younger players should pay dividends this year.



Next I present the expected change in offense and defense. USC loses a number of players, but the return of Nikola Vucecic, the team’s only efficient scorer, and the addition of Bryce Jones should lead to a slight uptick in the offense. Shot-blocking DeAngelo Casto is back for Washington St., and he should contribute to a slight defensive resurgence for the Cougars.



The next table shows the prediction for the conference. While a few teams are ticking upward, the bottom of this league is still very mediocre. One problem is that recruiting did not really turn around this year. The Pac-10 does not have any RSCI top 10 recruits this year, and even trails the Big 10 in top 100 freshmen this year.



The Pac-10 is the type of league where I love having an empirical model, because I do not have a good feel for how this league will perform this year. And the model does point to some sleeper teams like USC. The natural instinct is to say that USC loses four key players from a hideous team last year, and should finish near the bottom of the league. But that’s probably an overstatement. The Trojans return their two most important defensive players in Alex Stephenson and Nikola Vucecic, and Kevin O’Neil has the team believing they can win by playing elite defense. While USC may still have the worst offense in the Pac-10, if the defense stays dominant, the team should be able to finish in the middle of the pack.

Five Second College Football Rant

Earlier this week I read that the Pac-10 had the best record against other BCS conferences this season. (The Pac 10 is 10-4.) And I was curious how the other leagues had done, so I decided to pull the numbers. The next table shows the non-conference records of all the BCS leagues against one another and Notre Dame. (Scroll down and right to see all the BCS leagues.) There are 48 of these games scheduled this season. 41 of these have now been played and 7 are scheduled for week 13.



Still to come in Week 13
Georgia at Georgia Tech
Wake Forest at Vanderbilt
Florida at Florida St.
South Florida at Miami
Boston College at Syracuse
South Carolina at Clemson
Notre Dame at USC

(In the table I also list the losses to non-BCS teams as “other losses”. The only thing that stops the Pac-10 from being the top conference in all the computers is the fact that Pac-10 teams suffered five losses to non-BCS teams while the SEC only suffered one such loss.)

This paltry list is all we really have to compare the leagues to one another. There is not much to learn here, but there is one clear lesson. There is no reason to watch the Big East or ACC. The Big East and ACC are a disappointing 2-11 and 3-10 against other BCS leagues. But you probably already knew that.

So why does ABC keep insisting on airing ACC games every week in the Washington DC market? Are they trying to minimize their viewers? Do they want me watching stuff on my computer every weekend? Regional viewing windows suck.

But guess what? For the NCAA tournament, they are going away soon. I don't think people have any idea how awesome this is going to be.

Big 12 Prediction and a Quick Review of the Model

My model is incredibly simple. The idea is that

Efficiency =
f{Player Talent, Coaching}

Thus the difference between any two seasons is

Change in Efficiency =
f{Players lost, Player development, Incoming recruits, Coaching changes}

When evaluating players lost, I look at the possession-weighted tempo free statistics of the departing players. This is critical because when a team loses a very inefficient scorer (think someone with an 85 ORtg), that is not going to have a negative impact on the team. But if the team loses a player with a 120 ORtg, that will hurt a lot. And because it also matters how often a player shoots, I weight by the percentage of possessions used over the full game. (This incorporates both percentage of minutes and percentage of possessions on kenpom.com.) This year in addition to the tempo free offensive player statistics, I also include the tempo defensive player statistics in the model.

Player development emphasizes the fact that the biggest leap is often from freshman to sophomore year.

And I include incoming recruits which is simply a measure of the average impact of RSCI top 10 and top 100 recruits.

Finally, my model also accounts for coaching changes. On average, new coaches tend to have a negative impact on the offense. (It can take time for the players to learn a new offensive system.) But successful veteran coaches will usually improve the defense in the first year.

Today I present the model’s predictions for the Big 12.

Big 12 Prediction

For the record, I do not include Josh Selby for Kansas and Tony Mitchell for Missouri based on the fact that neither has been cleared academically. I’m currently still including LaceDarius Dunn of Baylor, but after yesterday, that may be a mistake. (These player eligibility issues are becoming more irritating by the day.)

The first table lists the changes we should expect this year based on players leaving and other factors. Texas Tech loses almost no key players, and should be better. But there are a lot of teams that are not so fortunate. Iowa St. loses a ton of talent, and unless I’ve read things incorrectly, transfers Chris Allen and Royce White will not be eligible until next year. That means there is really no reason to expect Iowa St. to be better this year.

And no Big 12 team lost as much talent as Kansas. Kansas will still be very good. After all, they did have the top Sagarin rating in the nation at the end of the regular season. But with Cole Aldrich, Xavier Henry, and Sherron Collins leaving in the off-season, Kansas will take a significant step back.

Texas and Baylor also lose a lot of talent, but Texas adds a pair of Top 15 recruits in Tristan Thompson and Cory Joseph. And Baylor adds Top 10 recruit Perry Jones.



Next I isolate the expectations for offense and defense. While Oklahoma St. loses two incredible scorers in James Anderson and Obi Muonelo, they should return two of the best defensive rebounders in the nation in Matt Pilgrim and Marshall Moses. While the Oklahoma St. offense is expected to slip a little this year, the defense may be better.



Finally, I present the standings. As it turns out, these are eerily similar to how the Big 12 coaches voted. On the one hand, that is very comforting. This model is still in the experimental stage, so it is nice to see the model matches with what others are thinking. On the other hand, the whole point of the statistical model is to identify misperceptions, (things that other folks are over-looking). But in the case of the Big 12, the coaches seem to agree with what the statistics predict. Kansas St. is the favorite, and Kansas and several teams are within striking distance.

For me the interesting story this season will be Texas Tech. Pat Knight has an extremely senior-laden team, and he needs to make the NCAA tournament this year. Texas Tech is currently 7th in these projections, which would put them squarely on the NCAA tournament bubble. But with a few late bloomers, Texas Tech is not that far behind Baylor for 3rd place in the league. On the other hand, with a few injuries or players not living up to their potential, Texas Tech could have a new coach this time next year.

ACC Prediction - The Harrison Barnes Effect?

Do Top 10 Recruits Really Make a Difference?

When I start to share my conference predictions, the first thing that is going to jump into your head is that I am clearly under-valuing elite recruits. But the next table should illustrate the reason. On average, teams that rely heavily on Top 10 freshman do not perform substantially better than the previous season.



Once you control for the other factors impacting the team (players leaving, player development), you do get a positive impact from top 10 recruits. But on average, it is not nearly as large as most people expect.

There are two problems here. First, there are top 10 recruits that were simply disappointing. (Think John Henson for North Carolina or Lance Stephenson for Cincinnati last year.)

Second, consider the alternative. If a top 10 recruit did not show up, they would not necessarily be replaced by a player with a 75.9 ORtg. Top 10 recruits tend to go to programs that have pretty good alternatives. For a quality team like Duke in 2008, what would have happened if there was no Kyle Singler? Well, Taylor King, Nolan Smith, or Brian Zoubek might have earned more playing time. And while Coach K clearly had his reasons for limiting those player’s minutes in 2008, they were all very effective when on the court. In other words, we have a value over replacement player (VORP) issue. For super talented teams, adding a top 10 recruit usually takes playing time away from other very talented players.

Do not misunderstand what I am saying. As Luke Winn and others have shown, top 10 recruits are often difference makers on the court. But basketball is still a team game. To expect one player (Harrison Barnes) to catapult North Carolina into the top 10 is optimistic. The team still loses its best player in Ed Davis and one of its top scorers in Deon Thompson. If Harrison Barnes only replaces Ed Davis’ production, the team will be treading water. For North Carolina to get better, it is going to require that other players develop at a rapid pace. Can Larry Drew finally cut down on his turnovers? Can Tyler Zeller finally stay on the court?

Or will North Carolina’s rebuilding project follow the path that Connecticut took earlier this decade? In 2006, a deep and talented UConn team lost to George Mason in the regional final. Then after seeing numerous players depart for the NBA, UConn went 6-10 in 2007. And it was not until 2009 that UConn returned to the Final Four.

If North Carolina returns to being an elite team it should qualify as a pleasant surprise. It should not be the expectation.

ACC Prediction

I’ll explain my model again in future posts if you missed it this spring. Suffice to say, I am going to try to replicate what most analysts do in their head. I’m going to look at players that leave and players that return and make a statistical prediction.

Technically, I’m going to look at the drop in efficiency we should expect based on the possession-weighted Dean Oliver statistics of departing players. If a team loses a player with an ORtg of 120 who shoots a lot, that will hurt that team’s predicted efficiency rating. If a team loses a player with an ORtg of 120 who shoots a little, that will hurt less. And if a team loses a player with an ORtg of 85, that will not have a negative impact on the team.

The first table shows the predicted changes in efficiency margin for the ACC teams. At the top we see Virginia Tech. The Hokies return a player-of-the-year candidate in Malcolm Delaney and lose almost no one from last year’s team. (I included the fact that JT Thompson is out for the season in the model.) That means Virginia Tech is going to be better.

On the opposite end of the spectrum, Georgia Tech, Maryland, Wake Forest, and Duke all suffered major player losses in the off-season. Duke makes up for those slightly by bringing in a top 10 recruit in Kyrie Irving.



The key thing I added to my model this summer is the defensive player statistics. This allows a more detailed prediction of why efficiency margins may change. The next table shows what the model predicts will happen to each team’s adjusted offensive efficiency and adjusted defensive efficiency.

As an example, Maryland loses some incredible offensive players, but retains some of its best defensive players. Jordan Williams, who was one of the top defensive rebounders in the entire ACC, and Sean Mosley who had the top steal rate on the Terrapins, are back. Thus while Maryland’s offense should be significantly worse, my model predicts a slight improvement in the Maryland defense.

Similarly, Chas McFarland struggled offensively last season for Wake Forest. But the team will miss his defensive intensity in the middle. Wake Forest’s offense is expected to take a small step back this year, but the defense is expected to take a major step back.



Everything I’ve listed above makes intuitive sense to me. But I’m not as satisfied with the projected ACC standings, which I present in the next table. These don’t quite add up to what most people are predicting. For example, Maryland and Boston College seem a bit high, and North Carolina seems a bit low.

But what the predictions really say is that outside of Duke, the ACC is wide open. Virginia Tech should be good, but the difference between the teams in the middle of the ACC is almost meaningless. The final standings will depend on which players surprise us by developing more than expected.



I’ve read some people who like Wake Forest as a sleeper team based on all the top recruits. But based on the poor defense Bzdelik displayed at his previous job, and the lack of returning starters, even a lineup of top 100 freshmen is going to have a hard time playing at an elite level right away. Jeff Bzdelik has been a great offensive mind and I expect him to turn Wake Forest into a dominant offensive team in the next few years. But remember that the ACC tends to eat its young. In any other BCS conference, Wake Forest would be predicted to finish in the middle of the pack. In the ACC, they are predicted to finish in last.

In the end, these ACC standings are not very satisfying to me because they do not conform to popular opinion. But as you will see in the next few weeks, many other conferences come closer to matching expectations.