As much as the NCAA is trying to advance the hype of its prize basketball event with the release of a bracket preview roughly a month before Selection Sunday, this year’s early look really didn’t tell us a whole lot that we didn’t know already.
The NCAA’s reveal of its top 16 teams at this point of the season should’ve provided little to be fired up about, nothing further than light debate about whether a team might be a line higher or lower.
Three of the four No. 1 seeds were a shoo-in, and the fourth (Xavier) was a worthy choice. Cincinnati was a deserving 2 seed, while Michigan State down at a 3 seed is a sign of the Big Ten’s slippage as much as anything, but also of a team that had been less than impressive in a diminished conference up until the last week.
Some thought North Carolina’s seed was too low, but there’s little reason to feel sorry for a team already with seven losses getting a 3 seed. Clemson might’ve been a little high, Texas Tech a little low. Oklahoma or Ohio State could’ve been out, Gonzaga could’ve been in.
In the end, though, there was very little notable about it. If people were looking for cunning insights on the committee’s methods and emphases this year, there wasn’t anything new there, either. Moreover: none of this matters much. There is still a month’s worth of games to give us a better sample size to reason this all out.
Regardless, the bracket preview still had some people steamed. Quite a bit of the anger was misdirected, if frankly as predictable as it has become tiring.
Even on a day where the 16 teams picked and their seeds shouldn’t have been too surprising to anyone, the RPI was yet again, somehow, targeted as a major problem over and over on social media.
(Of course, it didn’t help with TV analysts playing to their social media audiences, claiming that the RPI values a win on the road at a MAAC team the same as one at home against Villanova because they end up in the same grouping quadrant. Believe it or not, committee members do read the contents inside those quadrant numbers. But hey, it’s always more fun to stoke outrage with half-truths than to have, you know, the whole truth.)
We get it-about all of us have had moments where the RPI annoyed us. Crying wolf too much and when it’s not warranted, though, eventually just weakens arguments and many times gets them tuned out. As they should be.
Fact: of the 16 teams chosen for top four seeds in the bracket preview, 14 were in the RPI top 16. And 14 of them also were in the Ken Pomeroy rankings top 16.
Yet somehow this was an(other) example that the RPI is awful, terrible, and the Pomeroy rankings are apparently miles (and miles and miles) better, if one is to read Twitter? That makes zero sense. And to head off nitpicking, we’ll note again: most reasonable people had little problem with the teams picked or seedings on Sunday.
If one has a preference for one rating or another, that’s fine. But while the RPI is not perfect, what seems to regularly get ignored is that neither is any other rating. While we’d love to see the importance of strength of schedule reduced in the RPI, we’d also love for predictive metrics to reduce the impact of teams running up points and efficiency ratings at home against bad teams, ones that result in bottom-tier major conference teams oft overrated. (More on this in a bit) And from our view, a mathematician’s own assumptions do not belong in a power rating that the NCAA might use.
As far as we’re concerned, though, most of the outrage about the RPI is wasted. At the end of the day when we look at teams in one ranking and then another, it’s generally a wash. Most ratings’ pecking orders look pretty good to us. Every one of them have some teams with head-scratching ranks. It’s hardly enough to lose sleep over.
Others apparently do, though. That it comes even after a meaningless bracket projection shows it’s certainly time for a balanced philosophical discussion about power ratings and preferences for resume- or performance-based numbers. Because there is a real discussion to be had, if we can just acknowledge it.
Some people are comfortable with power ratings incorporating margin of victory and similar factors. If one reads social media, they might think everyone believes this, or at least they should.
Obviously for those looking to wager on games, the record of predictives is a plus. Unquestionably, efficiency statistics have been a great find for the sport as a whole, and there’s not a thing wrong with people coming up with different formulas to try to evaluate teams in a sport with 350+ teams and where every one of those teams will only play about 20-25 of them in a given season.
The case for predictives has been made repeatedly for a number of years now and needs little more introduction. But if they really are infinitely better, then why hasn’t virtually everyone been sold on them by now?
Hint: it probably goes deeper than people being ‘stuck in their ways,’ ‘behind the times’ or ‘afraid of change.’ And slapping those labels on people who haven’t bought in yet probably isn’t going to help convince them.
It seems rarely acknowledged but must be: there is a sizable contingent that doesn’t want margin of victory or factors related to it involved in ratings, maybe even a silent majority. Some have concerns about teams running up scores, but for us a concern of equal or maybe more weight is it’s a faulty assumption to think every college basketball game is played full-bore, the same way for 40 minutes.
Perhaps more than any other sport, the end result of basketball games can be heavily influenced by late-game situations such as teams fouling to try to get the ball back or one squad sitting its starters earlier than the other or getting less production from the end of its bench than another team’s. A one-point game with a minute left can become a 10-point win; a 25-point lead with eight minutes to play can become a 12-point win. They’re not even close to the same thing, so forgive us if a computer trying to treat them as such-or even attempting to build in adjustments for them-makes us uncomfortable.
That’s a main point, but not the only one. A charge also regularly leveled at the RPI is that teams can ‘manipulate’ it with scheduling. It’s true, but only to a point-there also is a fair amount of luck involved in scheduling. In most cases a team can’t know for certain whether that buy game will be against a team ranked No. 240 or 340. And no one can tell us Vanderbilt planned for anything near the No. 1-ranked OOC strength of schedule last year against this slate.
There also certainly is a case to be made, though, that teams can jack up ratings in more detailed metrics by dominating bad teams at home, and then being part of conferences where everyone else does the same thing. It’s hard not to ask questions when the prominent Pomeroy and Sagarin ratings have teams like Vanderbilt and Wake Forest (with their matching 10-16 records) in the top 90, a 14-12 Northwestern at 63rd or an Indiana with almost the same record (15-12) at 61st or 73rd.
Given these teams’ struggles this year-many of them even in very friendly, home-heavy non-conference schedules-that the RPI has all four teams at least 25-40 spots lower certainly looks more on target than predictives, and in turn fairer to teams who don’t get to play home-loaded schedules out of conference. That is, unless one is looking for rankings that replicate biases towards leagues playing the vast majority of their non-conference games at home, and that we never really find out how they’d do if they consistently played, say, the likes of Kent State or Southern Illinois on the road.
From the info that is available, from here…when we see Indiana losing at home to Indiana State (with a 6-9 mark in the MVC) and IPFW (6-6 in the Summit) and when Wake Forest is losing at home to Georgia Southern (7-6 in the Sun Belt) and Liberty (7-7 in the Big South), it’s a pretty poor assumption that both would be winning most every other game home and away against teams from those conferences. Which is precisely what their high predictive ratings currently suggest.
With the ample financial, exposure and perception advantages that teams like these in the highest-rated conferences already have, do they (and, in turn, their conference opponents) really also deserve a bump in so-called advanced metrics, too?
A primary goal of any changes in metrics should be to get as accurate of a read as possible on where teams with severe scheduling disadvantages actually rate. And we’re not sure we’ve seen an advanced metric that does. Frankly, in many cases the RPI-even with its overemphasis on strength of schedule-is fairer to those schools than predictives are right now.
The other part of the discussion is the human aspect of the NCAA Tournament selection process, and whether, how or if predictives diminish this.
Not everyone is looking for power ratings to win them a wager or select the field all by their lonesome, or for numbers to jump through hoops, sing all four parts in a church hymn harmony or rocket to the moon and back. They just want solid, basic numbers, simple enough for a reasonably intelligent layperson to digest, and then to let reasonably intelligent people evaluate the numbers and the information inside them.
Given the choice between a computer evaluating a 12-point victory and a human being, personally would choose the human 100 times out of 100. Also, while there are ratings that aim to compute exactly which teams would beat each other and thus statistically should be in the tourney, are there really that many people out there looking for a computer to do all the picking and seeding of teams?
The human aspect is what makes the NCAA tourney projections so much fun, and Selection Sunday too. It also can make it infuriating, but that’s why people get so invested in it.
How much more do we really need from the numbers? Do people want numbers doing the vast majority of the heavy lifting of the fun that is bracketology right now? Some may think they do; we’re not so sure they’d enjoy the result. And if anyone thinks different computer formulas are going to reduce or eliminate complaints about the field, they’re in the Twilight Zone.
These are all things that should be debated, and that the NCAA’s committee looking at possible changes in its ratings needs to consider. We can only hope the people looking into this for the NCAA are far less emotional about this issue than many fans and (worse) media.
Those considering different metrics should listen to social media and outraged media. And then they need to also listen to the other side of those arguments against predictives and for more resume-based and, yes, even basic metrics.
Is there a resume-based metric that is better than the RPI, fair to all schools, and also protects the many parts of the selection process that make it so appealing? Possibly.
Can a consensus be reached on predictive measurements? Or does the process even need significant change? Maybe the committee has already had pretty good information, makes pretty good (if imperfect) decisions from it, and people just like to complain, and when they complain it’s often with an agenda?
Whatever the decisions, there should be no rush. We don’t need a ratings formula that gets changed yearly in attempts to appease annual whining from fans and coaches who feel their teams were wronged. (See: the BCS) We also don’t need one that is giving major conferences even more advantages than the many they have already.
The selection committee’s job isn’t to be hip. Its job is to select and seed teams for a basketball tournament. No matter how often we’re told otherwise, it’s really not that hard, and they generally do a solid job of it. And it’s never going to make everyone happy, either, no matter how desperately they may try.
