Brackets 101: Everything you need to sound smart on Selection Sunday

Welcome to March!

It’s that time of year when any remaining stragglers among you finally start getting serious about the impending NCAA tournament. Which generally consists of a whole lot of “who’s in, who’s out” talk between now and next Sunday.

As a licensed Bracketologist (it’s an official word now), I want to make sure you’re the smartest person at the bar or your kid’s soccer game when your friends start arguing about Wichita State’s credentials. (And last year, I came in fifth out of more than 150 Selection Sunday projections tracked by the site Bracket Matrix, so hey, you’re in good hands.) While most fans are well-versed by now in terms like “RPI,” “SOS” and “quality wins,” I’ve found that many of you who write or tweet at me upset about my latest mock bracket often gloss over a particularly glaring flaw in their team’s resume or fixate on criteria that simply aren’t that important.

For today’s lesson in Bracketology 101, I attempt to get to bottom of this budding crisis.

Extremely important: RPI Top 50 record

My single biggest takeaway from both the 2015 and ’16 brackets, as well as last month’s committee Top 16 unveiling, is that no criteria is a better predictor of both seeding and selection than teams’ records against RPI Top 50 opponents. In fact, it’s the first column that jumps out at you on the official team sheets the committee uses.

What does that mean for this year’s bracket? Mainly that teams like 23-6 Butler (10-2 vs. Top 50) and 23-7 Florida State (9-3) will likely draw higher seeds than other metrics would suggest, while a team like 25-4 Cincinnati (3-3) will find itself up against a five-seedish ceiling.

It’s also why a mediocre Power 5 team like 17-13 Syracuse (6-7) and 16-14 Vanderbilt (4-7) may still be safer bets on the bubble than a respected mid-major like 25-5 Illinois State, which has played just two Top 50 games (going 1-1).

Extremely unimportant: A team’s own RPI rating

That may seem contradictory to everything I just said above, but RPI numbers are used far more in evaluating the strength of a team’s opponents than in actually selecting/seeding the teams. Looking back at last year’s bracket, St. Bonaventure, ranked No. 30 by RPI on Selection Sunday, got left out, while Tulsa, ranked 58th, got in. St. Joe’s, ranked 21st, got an 8 seed. Iowa State, ranked 23rd, got a 4 seed.

The only time I pay mind to a team’s RPI rating is when it bumps up against historical benchmarks. Few teams get left out if they reach the low-20s and few get in past the mid-60s. Syracuse last season set the new modern low by getting in at 71st; the previous mark was 67th.

Syracuse is threatening to break its own record this year, currently ranked 79th. That hasn’t kept me from including the Orange among my last four in. However, I don’t consider teams like No. 91 Georgia Tech or No. 93 Indiana to be viable contenders.

Becoming important: KenPom/Sagarin/KPI, etc.

The selection committee recently held a summit of advanced statisticians, with a goal of possibly replacing the RPI with a more advanced metric going forward. Already, though, we’ve started to see signs that analytics are creeping into the process, mainly when it comes to certain bubble teams.

In 2015, many were aghast when 20-13 UCLA — 2-8 against the RPI Top 50, 5-10 against the Top 100 — got in. Conversely, 27-6 Colorado State was considered a snub. Afterward, committee chair Scott Barnes strongly indicated a deciding factor for both teams was efficiency rankings like Ken Pomeroy’s, which had the Bruins in the high 40s but the Rams in the low 60s.

Last season, 24-8 Wichita State — No. 47 RPI, 1-4 vs. the Top 50 — got in entirely due to analytics. The Shockers entered the tourney ranked 12th in the country on KenPom (but even then, still had to play in Dayton). And a Top 25 KenPom ranking likely helped Vanderbilt get in at 19-13 and 2-7 vs. the Top 50.

This year’s most likely KenPom darling is ... Wichita State, again, if it does not win its conference tourney. The 27-4 Shockers have risen as high as 10th recently. His numbers are not as kind, however, to 21-8 USC, stuck down in the low 60s.

Completely irrelevant: Head-to-head results from individual games

This is one I get a lot. “How can you have Minnesota seeded higher than Michigan State when the Spartans beat them twice?” To that, I would point you to the cautionary tale of 2006 Hofstra.

Everyone remembers that year’s Cinderella story — 23-7 George Mason, which got one of the last at-large bids and, as a No. 11 seed, made it all the way to the Final Four. Lost to history was the fact that fellow CAA member Hofstra, 24-6, beat the Patriots in both the regular season and conference tournament but saw its own bubble burst.

What much of the general public may view as a no-brainer tiebreaker, the committee treats as one or two nights out of a long, four-month season. What matters far more is a team’s overall resume.

Mostly irrelevant: Conference records

At some point in time, many fans and media convinced themselves that going .500 in one’s conference was a critical factor for the committee. Or that a team that goes 10-6 in conference will always get first dibs over one that’s 9-7.

Maybe that was true when the leagues all had eight-to-10 members and played round-robin schedules. Today, the committee realizes that two teams from the same 14-team conference can play drastically different schedules, and thus those league records rarely come into play.

But much like RPI ratings, it’s worth paying attention to historical benchmarks. While it’s conceivable that a bubble team like TCU (currently 6-10 in the Big 12) or Ohio State (7-10 in the Big Ten) could get in at four games under .500, the fact is it hasn’t happened in 19 years. (Florida State in 1998 was 6-10 in the ACC). Less uncommon is a 12-6 major conference team getting left out (2012 Arizona and 2014 Georgia among them.)

Extremely (and sneakily) important: Non-conference schedule strength

Without fail, at least once a year, the committee leaves out a team most consider to be right on the bubble because it didn’t heed its annual missive that teams should test themselves out of conference. As such, be on guard for teams ranked 200th or lower in RPI-measured non-conference schedule strength.

Notable recent snubs include 24-9 South Carolina, ranked 247th, in 2016; 23-9 SMU (302nd) in 2014; 27-6 Drexel (212th) in 2012; and 21-13 Colorado (331st) in 2011.

As of now there does not appear to be anyone quite as glaring, but 19-11 Providence (183rd) might want to keep winning to be safe.

No longer important: Recency

When we see a team slumping toward the finish we assume the committee might give it a lower seed than they otherwise might have, but in reality, it really doesn’t matter whether your losses come in November, December or February. In fact, several years ago the committee officially struck “Record Over the Last 10 Games” from its team sheets to send the message that every game of the season is equally important.

In 2015, Indiana started 15-4 only to go 5-9 the rest of the way — including four losses in its last five games — leading many to speculate the Hoosiers would miss the cut. In reality, the Hoosiers made it fairly safely as a No. 10 seed thanks to their four Top 50 wins from early in the season.

Perhaps that might calm the nerves just a little for fans of Northwestern, which at 18-4 in late January appeared heading for a certain first-ever tourney berth only to drop five of seven since. The 20-9 Wildcats could use another quality win to feel safe, but they’re not as close to the bubble as some might think simply because there are teams below them that have been struggling all season.

... So there you have it. By all means, argue away over the next week-and-half in defense of your favorite team or perhaps against your archrival, but just be sure you’re arguing about the right things. You’ll sound like a prophet come Selection Sunday.