Anyone who has looked at the RPI realizes that the Ivy League had fallen from its recent perch, squarely in the teens, first to the absolute bottom of the list before rebounding to the high 20s.
While the league has struggled this season, the binary win-loss nature of the RPI is currently punishing the Ivies far too much for their close losses, something which should even out and lead to the league's rise over the next month.
Pomeroy tells a different story, as the league sits 15th nationally, primarily on the strength of the overly generous preseason rankings, which are still biasing the team ratings in his system.
The ultimate answer likely lies in between the two. Massey's composite rankings, which look at a wide ranging group of ranking systems, have the Ivy League in 21st, and Sagarin's predictive model has the league 25th. If forced to pick a narrow range within which the Ivies would be likely to fall, those boundaries sound as good as any.
In my own personal Pomeroy-style model, I have the league's average Pythagorean Win Percentage at .3745, which would place the league 24th in Pomeroy's rankings. That happens to fit quite nicely with the range discussed above.
From a historical perspective, that would still rank the league ninth best of the 33 seasons of the Ivy Academic Index Era. If that seems high by the eye test, keep in mind that the last three seasons of Pythagorean averages were 2010 (.3804), 2011 (.4228) and 2012 (.4631). It may just be that we Ivy fans have become so accustomed to solid play that we've forgotten how horrible the league's bad teams were many years.
The same potential deception applies to this season's Ivy race as well. Over the past three seasons, the league has boasted some very powerful squads that seemed to punish non-conference opponents, notching intimidating wins with very few no-show performances. Cornell won at Alabama and St. John's in 2010. Harvard and Princeton combined to beat Colorado, Boston College, Rutgers and Tulsa - all Top 100 teams - in 2011. Last year, the Crimson won the Battle 4 Atlantis taking down Florida St. and beat St. Joseph's.
Sure, this season, Columbia has already posted a signature win at a Top 100 Villanova squad, but the frequent missteps (the Lions losing at home to Marist, for instance) have grabbed more of the headlines. The result is a thesis that the league is comprised of eight flawed teams, making the title chase wide open.
A deeper look at the numbers reveals that it, indeed, is not.
First of all, the Ivies have three bad teams that have no chance of winning the league - Brown, Dartmouth and Yale (all in the 290s nationally in my model). The Bears have played better defense this season, probably due both to new coach Mike Martin's instance that they do such and to the addition of a true interior defensive presence in Rafael Maia. For its part, the Big Green has locked opposing offenses down for vast stretches as well, but Dartmouth has absolutely no offense of which to speak. The Bulldogs weren't expected to fall quite this far after losing Greg Mangano and Reggie Willhite to graduation, but Yale has been horribly inconsistent and generally awful on both ends of the court thus far.
Then, there's the middle crew of Cornell and Penn, which both sit in the 240s nationally in my model. The Big Red's offense has been the highest variance of any in the league by far, and its defensive variance ranks in the league's upper half. That means a different Cornell team every night, which is great for springing a random upset, but horrible for winning a test of consistency like the 14-Game Tournament. The Quakers have been exactly the opposite - consistent, but consistently bad. Penn has had six of its eight individual game Pythagorean Win Percentages between .1700 and .2600. Teams don't win a lot of games playing that poorly.
Finally, there are the league's contenders: Columbia (194th), Harvard (111th) and Princeton (104th).
The Lions' two losses to Marist and San Francisco reminded everyone of how this team can play remarkably poorly just when you think they might be really, really good. In fact, Columbia did that within the same game last night, hopping up by 17 on Bucknell before having its entire frontcourt land in foul trouble and ultimately losing the game by eight.
Similarly, the Crimson has shown its relatively unknown bench players, who have been pressed into service after losing four of five starters from last year, are very talented, but very raw from a game management perspective. A new problem has popped up every night for Harvard, as poor coverage of pick and rolls yielded too many open shots against Vermont and overzealous help defense led to multiple layups and offensive rebounds for Fordham.
The Tigers miss shooters Douglas Davis and Patrick Saunders greatly, as Princeton's offense has struggled for vast stretches of time and turnovers have resulted from the lack of comfort on that end.
So, the Ivy League does have eight flawed teams. Focusing on those flaws, though, misses the positives from each squad that ultimately separate two or three of the squads from the other five or six.
Princeton's defense has been way better than expected, almost on par with that 2009-2010 squad that shut down a high-octane Cornell team twice. Harvard has four of the league's top seven players in offensive rating (allocated points per possession). Columbia has the best inside-out duo in the league in Brian Barbour and Mark Cisco, not to mention the fact that Barbour can take over a game like a Zack Rosen.
Those are strong statements to make in a weak league. Thus, it should be no surprise that four of the league's top five performances to this point (and nine of the Top 15) have come from the Columbia, Harvard and Princeton trio. The only other team with at least two Top 15 games thus far is Cornell, but the Big Red also has three of the league's worst 11 showings.
While it's been hard to grasp where the league truly stands through the noise of this opening month, hopefully the analysis above provides a little clarity. In reality, we're not far from where we started. The Ivies are likely a low-to-mid 20s league. Princeton is no longer the prohibitive favorite it seemed in the preseason, but all that's done is to bring the presumptive second and third place teams (Harvard and Columbia) closer to the title chase. With youth driving the league's production for better or worse this season, what is quite clear is that this lull will likely be a short one, as the Ivies should improve greatly over the next couple of seasons.
Your proposal that the Ivy League abandon AI averages in favor of a higher AI floor is a simple, elegant solution. But it's a solution to a problem which only one member seems to have.
ReplyDeleteHarvard appears to employ academic boosters far more aggressively than any conference peer. Yale and Princeton might support your suggestion because, as the other high AI schools, right now they are the net losers when Harvard takes advantage of a loophole which they have decided to avoid on philosophical grounds.
Why would the other five Ivies want to reject individualized AI averages for each member? The current system is intended to give them a tangible advantage versus HYP to offset the supposed intangible edge which the latter three have in brand name and recruiting appeal.
A higher, uniform AI floor for all eight members is a lose/lose for the other five and its only appeal to Yale and Princeton is to close a loophole which the Crimson is willing to "game" but the Bulldogs and Tigers are not.