Wednesday, July 13, 2011

Initial Draft of Preseason 2011-2012 Rankings

It's been almost four long months since Brandon Knight beat Kareem Maddox off the dribble and finger-rolled in the game winning shot that sent Kentucky past Princeton in the First Second Round of the NCAA tournament.

The Wildcats used the 59-57 win as a springboard at the start of a Final Four run, while for the Tigers, it was merely the first loss of many.

First was the known reality that seniors Kareem Maddox and Dan Mavraides had exhausted their eligibility and would be need to be replaced. Second was a more troubling and unexpected departure, as Coach Sydney Johnson left to take the same position at Fairfield.

Ironically, it was the latter which was the easier fix. Princeton once again stayed inside the family, hiring another BCS assistant who was with the Tigers in their mid-90s glory days - Mitch Henderson. Trying to replace a legitimate All-Ivy candidate and the team's best offensive and defensive player in Maddox as well as the team's best three-point shooter in Mavraides will likely prove to be a much taller task.

Princeton's odds of overcoming those losses would be a lot greater if it weren't for the fact that fellow co-champion Harvard returns 100 percent of its minutes and brings in the consensus top recruiting class in the league.

The Crimson is the first Ivy League team since the advent of the Pomeroy Ratings to return all five starters from a Top 100 squad and the first champion of any kind during that span not to lose a single rotation player.  While enough injuries can derail any team (where would Harvard have been with a fully healthy Casey last season and alternatively, where would it have been if it lost both Casey and Wright for the season), the freshman class has enough answers at enough different positions to allay any serious concerns.

Among the other six league teams, Yale is expected to make the biggest jump behind All-Ivy center Greg Mangano and sharpshooter Austin Morgan. All of Brown, Columbia, Cornell and Penn figure to gravitate around, but maybe a little below, the Division I average, while Dartmouth should remain a distant, but steadily improving last.


If you haven't read Dean Oliver's Basketball on Paper or don't have a strong familiarity with Ken Pomeroy's adaptation of the Oliver stats to college basketball, this section might not be for you. For those that have, here is how I arrived at the preseason projections.

At the most simplistic level, I created mock projected season stats by player, as well as team-level opposing stats.

The player-level statistics were easiest to generate for juniors and seniors, many of whom have already logged a ton of minutes and show decent consistency in their output. For more lightly used players, I tended to peg their efficiency just south of the expected team average, though that was far from a universal rule.

Freshmen were far more difficult, though only a handful of them actually matter. Those I charted out separately, using things like shooting percentage, rebounding percentage and other statistics that tend to exhibit correlation from level-to-level as my "knowns" and pegging the unknowns to reasonable levels relative to the historical data that I have. The freshmen that project to be bench players were treated much like the lightly used players above - given what might be considered a "replacement level" rating as well, since the best of them or the lightly used players will be awarded the reserve minutes.

The primary complicating factor involved situations where a large bank of minutes became available without logical replacements to fill them. In these cases, I favored lower efficiency rather than higher, admittedly leaving a lot of potential upside for a breakout candidate, but since these relatively blind estimations comprise a small portion of total minutes and possessions, the potential resulting error shouldn't skew the results too greatly.

The opposing team four factors are easier to derive. Effective field goal percentage allowed is decently consistent across years, unless a team loses a particularly imposing interior defender that racks up blocks and alters shots or a lock down perimeter defender that can affect opposing shooters.

Rebounding percentage is easily altered year-over-year depending on whether a team loses a player that rebounds at a high rate or figures to bestow more minutes upon a better rebounding player. Turnover rate tends to be even less personnel specific with the exception that the loss of a particularly high steal percentage guard can lead to a corresponding decrease in turnover rate forced. Free throw rate tends to bounce around the most, but significant deviations from the average can be predicted (see: Cornell 2011 and its lack of an imposing or even effective frontcourt defender).

With the individual players rolling up into the team stats and the derived opponents' stats, each Ivy team now has all of the Pomeroy stats necessary to simulate the 2011-2012 league season. A future draft will include a full 2011-2012 simulation, once all eight schools release their schedules, as well as player stat updates depending on preseason injuries or departures that become known.


Though it might be a small sample size, it is worthwhile to revisit last year's Season Preview to see how the model and projections fared against the actual results. The individual team previews are accessible through the 2011 Season Preview label on the sidebar of this blog, but I've pulled in the top level stats (Record, Pomeroy Pythagorean Expectation and Offensive and Defensive Ratings) here.

The biggest misses were Yale and Cornell - almost a full tenth of a point off the pythagorean win projection - while the title chase was predicted with decent accuracy and Penn, Dartmouth and Brown landed within four-hundredths of a point of their projected finish.

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 22-8, 11-3, T-1st
Actual Overall/Ivy Record (without Postseason): 24-6, 12-2, T-1st
Expected Efficiency Stats: 0.6767 Win Pct; 99 ORat; 92 DRat
Actual Efficiency Stats: 0.7157 Win Pct; 105 ORat; 97 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 20-8, 11-3, T-1st
Actual Overall/Ivy Record (without Postseason): 23-5, 12-2, T-1st
Expected Efficiency Stats: 0.6524 Win Pct; 105 ORat; 97 DRat
Actual Efficiency Stats: 0.7213 Win Pct; 110 ORat; 101 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 15-13, 8-6, T-3rd
Actual Overall/Ivy Record (without Postseason): 10-18, 6-8, T-5th
Expected Efficiency Stats: 0.4804 Win Pct; 100 ORat; 100 DRat
Actual Efficiency Stats: 0.3990 Win Pct; 103 ORat; 107 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 15-13, 8-6, T-3rd
Actual Overall/Ivy Record (without Postseason): 13-15, 7-7, 4th
Expected Efficiency Stats: 0.4420 Win Pct; 100 ORat; 102 DRat
Actual Efficiency Stats: 0.4281 Win Pct; 100 ORat; 103 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 14-14, 6-8, 5th
Actual Overall/Ivy Record (without Postseason): 11-17, 4-10, 7th
Expected Efficiency Stats: 0.3422 Win Pct; 99 ORat; 107 DRat
Actual Efficiency Stats: 0.2776 Win Pct; 100 ORat; 109 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 11-17, 5-9, T-6th
Actual Overall/Ivy Record (without Postseason): 15-13, 8-6, 3rd
Expected Efficiency Stats: 0.2944 Win Pct; 96 ORat; 105 DRat
Actual Efficiency Stats: 0.3952 Win Pct; 97 ORat; 101 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 12-16, 5-9, T-6th
Actual Overall/Ivy Record (without Postseason): 15-13, 6-8, T-5th
Expected Efficiency Stats: 0.2493 Win Pct; 92 ORat; 103 DRat
Actual Efficiency Stats: 0.3289 Win Pct; 98 ORat; 104 DRat

Expected Overall/Ivy Record/Ivy Finish (without Postseason): 5-23, 2-12, 8th
Actual Overall/Ivy Record (without Postseason): 5-23, 1-13, 8th
Expected Efficiency Stats: 0.1243 Win Pct; 89 ORat; 107 DRat
Actual Efficiency Stats: 0.0872 Win Pct; 88 ORat; 108 DRat



Losing three starters from a team that barely escaped the Bottom 100 nationally wouldn't seem to indicate promise on the horizon. For Brown, it's very likely to be addition by subtraction.

Garrett Leffelman took a huge step back from a breakout junior campaign and while Adrian Williams rebounded nicely from a rough 2010 season, neither brought offensive production that would be deemed irreplaceable. With Matt Sullivan and Stephen Albrecht pushing for minutes this year, as well as a promising freshman class and returning Ivy ROY Sean McGonagill, the guard positions are pretty well covered anyway.

Whether there's enough talent to absorb the loss of swingman Peter Sullivan remains to be seen, but another year of improvement from the frontcourt of Tucker Halpern, Andrew McCarthy and Dockery Walker should provide enough of a boost in other areas to make the Bears a better team overall.


The myriad frontcourt losses will hurt the Lions' gaudy rebounding rates from last season (eighth nationally on the defensive glass), but most of those minutes went to highly inefficient offensive players.

John Daniels and Mark Cisco remain, but not much else at the four and five, which makes the frontcourt rotation minutes a huge source of variability heading into the season. There is definitely room for a freshman to make an impact, if one is capable.

The Lions are pretty well set in the backcourt both in terms of the starting lineup and the bench, and the ability to take pressure off of Noruwa Agho offensively in the form of a reduced possession load could only improve things for Columbia.


The frontcourt was horrible for the Big Red last year and after the losses of Adam Wire, Mark Coury and Aaron Osgood, it doesn't figure to get any better.

Juniors Peter McMillan, Eitan Chemerinski and Josh Figini are the only true returning fours and fives, but none has played more than 20 percent of the team's minutes in either of their first two seasons. That leaves Cornell in the unenviable position of squeezing production out of those three and similarly untested freshmen or moving a more seasoned combo forward (Errick Peck) to the PF spot.

The Big Red will keep bombing in threes at a high rate all season, but unless a couple answers arrive up front, its opponents will be getting their points early and often in a much easier way - layups and free throws.


The Big Green needs a lot of help in a lot of places.

The 323rd shortest team in the country lost one of just three players on its roster who was taller than 6'5, something which had already forced Dartmouth to go small for most of last year. The guard spots are far from settled as well with only Jabari Trotter posting an Offensive Rating that would be considered above replacement level for an average Ivy team.

Help could come from the freshman class, as the Big Green picked up a few combo forwards and a true center in this year's haul. While the recruits are better than what Dartmouth has been landing in the recent past, it's hard to say that any of them are guaranteed impact players. That being said, at least a couple should be able to help the Big Green improve upon its abysmal 88/108 ORat/DRat splits - just don't expect the improvement to be too great.


After slogging through teams with massive holes that have to be plugged with estimations and tenuous extrapolation, the Crimson provide quite the vacation.

Harvard returns everybody from last year's squad and most have displayed pretty consistent ORating trends (last year's Ivy POY Keith Wright, notwithstanding) making it easy to project performance going forward. The stacked freshman class should provide sorely needed bench help and should be a solid improvement over last year's 7-9 spots in the rotation which failed to produce a player that could log more than 20 percent of team minutes or post an ORating above 90.1.


It's hard to argue that Zack Rosen, Miles Cartwright and Tyler Bernardini form a formidable backcourt. The question is what's behind them and in front of them.

With the loss of Jack Eggleston and Conor Turley, roughly 65 percent of the total frontcourt minutes evaporated due to graduation. The eminently fragile Mike Howlett is capable, but can he stay on the floor? Sophomores Fran Dougherty and Cameron Gunter and freshmen Greg Louis and Henry Brooks have potential, but can they fulfill it within the accelerated timetable that the Quakers need?

Even if the frontcourt issues get solved, the backcourt still has depth concerns. Beyond the big three, there aren't any proven guards in the stable, and Rosen, Cartwright and Bernardini can't play 100 percent of team minutes (not that Rosen, who finished 34th in Min% played, won't try during his senior season).


Usage rate is incredibly important in estimating the impact of losses. The graduation of a player that uses just 15 percent of possessions while on the floor, but does so at very high efficiency, is a lot easier to overcome than a player that uses well over 20 percent of possessions, even if at somewhat lower efficiency.

The Tigers had three players use more than 20 percent of floor possessions last year, and two of them - Mavraides and Maddox - are now gone. Ian Hummer and his 25.6 percent usage rate returns, but there isn't too much more room for him to take on extra possessions. That burden will likely fall upon guard Douglas Davis, who has shown relatively inelastic efficiency movement between usage rates in the high teens and low 20s.

Former rotation players Brendan Connolly, Will Barrett, Patrick Saunders, Mack Darrow and T.J. Bray will no longer have the luxury of posting possession rates in the low teens, however, and how they handle the increased offensive workload will determine how far the Tigers will fall from last year's Ivy perch.


The Bulldogs came out of nowhere last year, rebounding from what looked like a crippling loss of forward Michael Sands just before the season started (along with the loss of three starters from the previous season), to finish as a top 200 squad for the first time since 2007.

Greg Mangano went from an All-Ivy defensive presence to an All-Ivy player in an astounding transformation, while guards Austin Morgan, Porter Braswell and Reggie Willhite each proved that they could maintain their offensive efficiency despite a heavier usage rate.

With almost all of its significant minutes aside from Braswell returning, along with a recruiting class stacked with forwards and wings, the Bulldogs should have a lot of options for a small pool of minutes - healthy competition which should send Yale even higher this season.


Here are the initial pythagorean win percentages for each Ivy team, including a rough estimation of the corresponding national rank:

1. Harvard - 0.7980 (64th)
2. Yale - 0.5558 (139th)
3. Princeton - 0.5530 (140th)
4. Penn - 0.4301 (180th)
5. Cornell - 0.3878 (196th)
6. Columbia - 0.3448 (213th)
7. Brown - 0.3318 (217th)
8. Dartmouth - 0.1590 (294th)

The average pythagorean win percentage across all Ivy teams would be 0.4450, which would have made the Ivy League the 15th or 16th best conference, nationally, over the last five years on average.

Using the offensive and defensive ratings that correspond to the above pythagorean win expectations, as well as assumptions for pace and variance/co-variance by team, we can run simulations of the 2011-2012 Ivy League slate.

As one might expect, Harvard is a prohibitive favorite to repeat as Ivy champions, 93 percent to grab at least a share and 84 percent to win its first outright title. Princeton and Yale check in at eight and seven percent to claim at least a share, respectively, with each at three percent to take the solo crown. That the Bulldogs are slightly behind the Tigers in the Ivy race despite being slightly ahead in the pythagorean win percentage is illustrative of the power of co-variance (Yale is once again slated to be more high variance than Princeton, but if it displays positive co-variance as high as it did last year - good offensive performances cancelled out by bad defensive performances and vice versa - it will struggle to combine the great performances on both ends to knock off the league's best teams).

The playoff likelihood of 9.3 percent is less than half of what it was last year, due to the existence of one dominant team this season versus two equally matched teams in 2010-11.

The Projected League Wins illustrate the point that the league is in three (or maybe four) tiers this year:

1. Harvard - 12.0
2. Yale - 8.5
3. Princeton - 8.4
4. Penn - 6.8
5. Cornell - 6.3
6. Columbia - 5.6
7. Brown - 5.5
8. Dartmouth - 2.8

The Crimson and Big Green both claim their own rungs on the ladder, at opposite ends of the Ivy pecking order. Then, there's the middle 2-7, which is separated by just three league wins. The Tigers and Bulldogs are clearly the class of the group, finishing in the top four 89 and 90 percent of the time, respectively. The Quakers and the Big Red are illustrative of the fluidity of that middle clump of teams - Penn is 14.4 percent to claim a share of second, but 7.4 percent to finish in seventh, while Cornell checks in at 9.1 percent and 11.0 percent, respectively.

The league truly is seven deep this year, as the Lions and Bears each have decent shots at cracking the Top 200 nationally, but only roughly 30 percent chances to land in the league's top four.

Even Dartmouth might be improved enough to avoid the 300s in the Pomeroy Ratings, which would be the first time the league had every team finish in the Top 300 since the 2006-07 season.

For now, this is how the league race projects out for the 2011-12 campaign, but there will be an update as summer turns to fall and full schedules (as well as player reports, including injuries) become more readily available.

No comments:

Post a Comment