I. WHAT THINGS ARE NOW AREN’T WHAT THEY WILL BE LATER
Every bracket projection that you will read from now until the NCAA Tournament selection special suffers from the same problem: They are not projections. Virtually all bracket architects are looking at resumes and team sheets as of the date of bracket construction; there is no projection in that effort — this is merely casting a fixed reality without anticipating practical environmental shifts. If Jerry Palm or Joe Lunardi are pounding words onto their keyboards, they are doing so in the moment and are not attempting, in any form or function, to actually project what a team’s profile will look like on March 11th.
I don’t know why bracket experts don’t undertake any predictory efforts in their bracketing. The whole point, I think, of assessing a team’s relative chances at participation in a March adventure is to determine the quality of a team’s portrait when the paint dries. This is a difficult task, one that is arguably out of the grasp of the multitudes charged with organizing a bracket complex, but there is, inherently, a greater value in forecasting rather than picking a static point in time and asserting “As things stand now . . . .” This lack of forward prognostication is the propulsion that causes me to shrug so hard that my shoulders attempt to tear themselves from my body every time I look at a bracket before the final weekend of conference tournament games.
II. WE HAVE THE CAPACITY TO FORECAST WHAT THINGS WILL BE LATER, BUT THERE ARE NO GUARANTEES
Predictive models are massively useful in understanding a particular team’s outlook. While far from bulletproof, there are a number of systems available on the internet machine that can quickly and ably assess a team’s future. There are, of course, variables — injuries, sickness, luck, weird matchups, etc. — that influence a team’s “true forecast” that are often left unrepresented in such models, but focusing on what these models can’t do rather than on what they can do (which is sizable and profitable) is just obfuscatory complaining. We have the power and the technology, and there is enough evidence that this stuff works to a sufficient degree of confidence. Why the material isn’t incorporated more often in building stories of potential Tournament fortunes remains a mystery, although one with a lazy and pointless response: “We can’t, for certain, say what a team’s profile will look like, so we simply use what we have — this very second.”
If your financial advisor took this same tack in assessing your investment portfolio while gauging the overall investment landscape, you’d fire him or her on the spot and throw coffee in their face. Nobody is demanding fortune tellers, but using all viable and available information to compose a reasonably defined future is kind of a prerequisite to a robust analysis.
III. THIS IS WHAT LATER THINGS POTENTIALLY LOOK LIKE, BUT IT IS CONTINGENT ON A NUMBER OF VARIABLES (DUH)
As of February 11th, this is the NCAA’s team sheet file: Click here to make something else appear on your internet machine. Here is an image of Syracuse’s team sheet:
Little of this matters, at least in the context of absolute NCAA Tournament inclusion or exclusion. Syracuse will face, in its final six regular season games, six teams that will likely stand as “Quadrant I” or “Quadrant II” opportunities. There is no gruff remaining on the Orange’s schedule, a half-dozen chances to stuff the ballot box and shape its overall resume. Yet, no bracket expert is considering those six games in their bracket projections; all they see and reflect are the games that already appear on Syracuse’s one-page biography. Isn’t it more compelling assert status in the lens of expectation?
The RPI is not designed to do this, and neither is ESPN’s strength of record or KPI. There are models, though, that are devised to serve this purpose: KenPom, Sagarin, and Massey (to name just a few). Using a blend of these three models, this is what a truncated team sheet for Syracuse is expected to look like (through games played on Sunday):
There is admittedly limited context to this: In a process where resumes are compared to determine inclusion, simply looking at the Orange’s projected profile is only witnessing something in a vacuum (full disclosure: I am not paid to build profiles for any kind of bracket construction; I both recognize this paucity of context and also do not give a damn (but am willing to change my position given the sudden appearance of cash money)). But as a direct comparison to its current RPI-dominated team sheet, it shows a number of things: (1) Syracuse is probably going to end the year with a clean performance in Quadrants III and IV; (2) the Orange is going to look competitive in Quadrant II; and (3) Quadrant I value is going to fill out almost 30% of Syracuse’s overall profile.
This isn’t a dark, bleak forecast. Syracuse isn’t comfortably in nor outrageously out. Even pushing through a season with little offensive might and a limited bench hasn’t been the profile tragedy that many are alleging. The Orange has hurdles to cross — 5.5 available bodies is a colossal problem; there’s about a 50% chance that Syracuse finishes with fewer than 20 wins (meaning that the Orange will fail to bag a big kill the rest of the way); chaos in conference tournaments (out of Syracuse’s control) could impact the depth of the true at-large pool, etc. — but this is still a team that isn’t dead and buried. If Syracuse maintains its relative strength compared to the rest of the country — a team positioned somewhere between 40th and 50th nationally — the Orange will remain a team in contention for a bid. With a little bit of positive juice and a beneficial selection committee eye, Syracuse may be fine (albeit extremely nervous).
IV. THE MOST IMPORTANT THING IS THE THING THAT NOBODY ACTUALLY DISCUSSES: THIS IS PROBABLY A DUMB WAY TO POPULATE AND SEED A SINGLE-ELIMINATION TOURNAMENT
PROBLEM A: The RPI is old math run on old ideas considered useful to old people that don’t understand why it’s old math run on old ideas. That it is used so fervently in the selection and seeding of a competition field is laughable, especially considering that humans are far more intelligent around understanding how teams relate to each other and how strength is represented. The NCAA convened a meeting this past summer with a lot of smart people and nodded their heads as the smart people said smart things, which caused the NCAA to reorganize its existing shit pile into different shit piles that still smell like shit. How the NCAA continues to run in fear of human advancement in analytics is one of life’s great stupidities.
PROBLEM B: There is no framework around the important differences between populating the field and seeding the field. These are two wholly different analyses: Inclusion/exclusion should carry one set of considerations (“Is this team deserving?” or “Is this one of the best teams in the country?”) and seeding should carry another (“What is the ordering of the relative strength of the teams included in the field?”). The selection committee appears to blend these two sets of determinations and relies on the bland team sheets to engage in both discussions.
PROBLEM C: There is no framework around the first analysis (team inclusion/exclusion from the field). Is this based on some opaque belief in merit or is it based on team power? If the framework positions itself on the former consideration — body of work — how does the selection committee normalize for disparate opportunities? Saint Mary’s, which is very good and is more than capable of giant slaying, is institutionally marginalized from a team sheet perspective because it doesn’t have the schedule conveniences of a team like like West Virginia. Yet, the Gaels would give the Mountaineers a great game based on advanced metrics. To punish Saint Mary’s because they don’t have the institutional advantages of a power conference team throws deep shade upon a process that rests on perceived merit.
If the framework is organized toward the latter — true skill — the NCAA is underproviding relevant information as its team sheets are hilariously devoid of important advanced metric detail. The RPI was not devised to consider qualitative power, and the team sheets fail to address that in any pertinent form.
When the selection committee chair does his or her annual interview following the release of the bracket, the comments are expectedly contradictory — what may apply to one team seems to run opposite of what applied to another. The reason for that isn’t necessarily idiocy; it’s probably more the residue of having a lack of defined framework to work within.
PROBLEM D: Regardless of the approach to populating the field, there’s still the issue of how the bracket is seeded. The ultimate goal in seeding a tournament is to protect the strongest teams, but seeding a bracket based on team sheets which provide little insight into relative ability results in odd seeding decisions. This, consequently, can put a strong team in a position where it is unfairly forced into a more difficult path than it deserves and allows for weaker teams to operate in the inverse. This creates artificial drama, but if the end goal is to at least try to approximate a bracket construct that seeds teams based on relative strength, RPI-dominant team sheets (and analyses that do not delve into the rich textures of assessed skill), unsuccessfully directs decision makers in the right direction.
PROBLEM E: The constitution of the selection committee is not one of dedicated basketball acumen. These are athletic directors, conference commissioners, and other university administrators that exist in their own universes and aren’t likely spending significant amounts of time thinking about (a) the philosophical underpinnings of how teams relate to each other, or (b) about the college basketball landscape in any identifiable detail. As fiduciaries of constituents in some form, we’re asking individuals without technical expertise to engage in technical analyses, and there’s going to be leakage in their ability to formulate the kind of discussion that needs to occur to elevate the quality of a bracket construction.
In other words, when you give non-experts weak information and require them to operate in an undefined framework, you’re going to get a product that is less than stellar. This isn’t an issue for the NCAA — the NCAA only cares that a field is populated; once that’s done it can start counting money (which is the NCAA’s only true concern) — but it is an issue for pretty much everyone else (coaches, players, fans, Vegas, etc.).