Mock Madness: Inside look at the NCAA tournament selection process
INDIANAPOLIS -- Before our group of 20 media members and conference officials arrived at the NCAA's mock selection exercise last week, we each received a binder stocked with vital information. Among the topics covered:
• How to ensure the most Big East teams make the NCAA tournament.
• How to give Duke the easiest path to the Final Four.
• How to make backroom deals so the school each athletic director represents makes the tourney.
I'm kidding, of course.
NCAA Senior Vice President Greg Shaheen created the mock selection exercise specifically to blow up those myths. "You won't look at it the same way after this 24-hour period," he said, "because it is so different than what you expected."
He's correct. After spending much of Thursday and Friday selecting a tournament field using precisely the same rules, information, computer equipment and software as the real selection committee, it's highly unlikely any of us will ever again accuse the committee of favoring one conference over another or of favoring BCS conferences over non-BCS leagues. The Duke haters also can ditch their conspiracy theories. Not only did we avoid favoring the Blue Devils, we almost used them to create a Region of Death that wouldn't have been fair to Duke or any other team. But we'll get to that later. As for backroom deals, the methods used to select teams render those next-to-impossible unless a committee member has the consensus-building skills of a Benjamin Franklin.
The exercise also exposed one potential flaw in the process. The committee relies entirely too much on the RPI in the room. It should be noted that the NCAA sends committee members other rankings on a regular basis, so they have easy access to them. It also should be noted that we rarely discussed the RPI when debating teams. Still, almost every piece of data committee members see in the selection room has the RPI attached. Teams are judged by their record against opponents in the top 50 of the RPI. Their nonconference schedules are ripped to shreds if they include too many teams with RPI ranks above 100. This appears helpful at first, but using just the RPI can lead the committee to some potentially erroneous assumptions. For example, during our two days in the room, the RPI told us Memphis was the nation's 26th-best team.
This is a major flaw only if committee members don't use the browsers on their NCAA-provided Dell laptops to break free from an RPI-induced stupor. As the stand-ins for Big 12 commissioner Dan Beebe, SI.com colleague Andy Glockner and I frequently visited
But, as the seconds grow more precious during the selection process, it's easy to fall back on the RPI because it is plastered all over the video screens each member uses to view side-by-side comparisons of similar teams under consideration. Bottom line: If a committee member goes into the room determined to use other sources along with the RPI, he'll probably vote well. If he only uses the information provided, the RPI will have too much influence.
The RPI is gone by the time the committee actually begins placing teams into the bracket. That might be the most amazing part of selection weekend -- and the part that might need tweaking in the future. The bracket that gets torn apart by millions from Sunday night until the tournament tips off on Tuesday is assembled in a matter of hours. One year, Shaheen said, the committee had 45 minutes to bracket the field.
We put together our bracket in about two hours. While the rest of the exercise was time-compressed because the real committee has two extra days to determine the teams in the field, that part happened essentially in real time.
So how does the real committee get from chicken piccata at Iaria's restaurant on Wednesday -- "Sansabelts are involved because of the flexible garment needs," Shaheen said of the meal -- to a finished bracket in Greg Gumbel's hands at 6 p.m. Sunday? In pretty much the same way a ragtag group of writers and conference wonks
The biggest difference? Unlike us, the real committee won't share the results in real-time on Twitter. (To see more of the gory details, check the timelines of
As I reached the top of the staircase at the Hotel Conrad on Thursday, I noticed a familiar face. Whizzing past and dragging a suitcase was the real Beebe. The actual committee met at the Conrad this week for a principles and procedures refresher and to help familiarize new members Steve Orsini (SMU's athletic director) and Scott Barnes (Utah State's athletic director) with the process and with the equipment they'll use during the selection process. I figured I should grill Beebe about his selection and seeding likes and dislikes so Glockner and I could properly get into character, but by the time I got downstairs, Beebe had already hailed a cab and disappeared. The man is quick.
A few minutes later, Shaheen and Ohio State athletic director Gene Smith -- the actual committee chair -- ushered us into the selection room. Each fake committee member had a laptop. Stationed throughout the room were sets of three flat-screen monitors. Throughout the process, these monitors would show the teams up for consideration, side-by-side comparisons of specific teams and "nitty-gritty" reports on each team that feature 16 different factors including RPI rank, records (home, away, neutral and against RPIs 1-50, 51-100, 101-200 and RPI top 100), average opponent RPI rank and strength of schedule. (Notice how often the RPI pops up there?)
When Shaheen started at the NCAA in 2001, there were no screens. All the information was provided on paper. During the 2001 selection meeting, the committee used 100,000 sheets for copies. When Shaheen plugged his laptop into a dial-up connection to get realtime scores, committee members chided him for bogging down the process. Ten years later, committee members have instant access to almost any piece of information they might want.
Once we got settled, Shaheen and Smith explained the process. After reports from members on the conferences they were assigned to monitor all season (never their own), committee members submit ballots containing the names of the teams they consider surefire at-large teams and the names of teams they believe the committee should consider for at-large bids. As the fake Beebe, Glockner and I submitted 16 locks and 35 under-consideration teams. Fake Beebe was the most skeptical of the bunch; one fake committee member submitted 32 locks.
(Before you start sending nasty e-mails, consider this. We made our bracket using games played through Wednesday night. The NCAA simulated conference tourney results for us, but it did not simulate through the remainder of the regular season. We didn't know at the time that Nebraska would beat Texas or that Purdue would beat Ohio State.)
Shaheen offered a piece of advice about the under-consideration teams that didn't sound sage until hours later as the fake committee debated the merits of the teams near the bottom of the at-large pool. "Don't put teams here to feel good about putting teams here," Shaheen said as he stood at a large white board pointing at the under-consideration group. "Put teams here that you believe could ultimately be one of the best 37 in the country." At first, Glockner and I thought it would be best to include every team we thought might have even the slimmest chance of earning an at-large bid. It turns out that including teams that realistically have no chance only bogs down the process.
Why? Because the committee debates and votes multiple times on every team on the under-consideration board. By the time the committee gets down to the BCS conference teams with .500 conference records and zero quality wins or the non-BCS conference teams with good records but zero quality wins, those debates often become counterproductive and frustrating.
So how does the committee determine which teams receive at-large bids? It begins with the surefire at-large teams. Any team that receives all but two eligible votes on that first ballot is in the tournament. Using the information available at 1 p.m. Thursday, we punched the dance card for 21 teams. Unfortunately, the computer thought we'd only picked 18 teams. We noticed this because Duke, a certain at-large team, didn't appear anywhere on the list. After some checking, the list was corrected. That may sound inconsequential, but remember, this is the same software and hardware used by the real committee. If it happens next month, they'll have to notice and correct it.
At that point, we switched from selection to seeding for the teams we had just placed in the tournament. We were asked to pick the best eight from among the 21. On fake Beebe's laptop screen, the 21 team logos appeared with a bubble next to each one -- except there was no bubble next to Texas or Kansas. As fake Beebe, we weren't allowed to vote for Big 12 members. Ryan Feldman of HoopsReport.com and Doug Harris of the
A computer at NCAA headquarters about a mile away very sloooooooowly tallied the top eight vote-getters. "We just learned that we're using a Speak and Spell back at the office," Shaheen cracked. Eventually, a new screen popped up on fake Beebe's laptop. The eight logos were arranged vertically next to a set of bubbles numbered one through eight. On eight laptops, fake committee members were asked to rank each team one through eight. The results would be added, and the team with the lowest sum would be the tournament's No. 1 overall seed. The next three lowest sums would get true seeds No. 2 through No. 4 -- the field eventually will be seeded No. 1 through No. 68 before bracketing begins -- while the other four teams would be thrown back in the pool. The entire process would be repeated to get true seeds No. 5 through No. 8.
(This seems like proper time to suggest to CBS that a feed all weekend of Shaheen and tournament associate director David Worlock would be comedy gold. I'd pay at least $50 bucks on pay-per-view to hear their running commentary devoid of any bracket-revealing context. Worlock unleashed some zingers during our time there. A favorite is this mantra: "The message," he said, "is always, 'Please have your conference tournament over by Saturday.' ")
The laptops of fake Beebe and fake Gene Smith (SI's Davis and Steve Scheer of CBS Sports) had one extra item on the screen. A red stripe informed the fake committee members that they were ineligible to vote. "You are restricted from participating due to conflicts with teams in this election," the message inside the stripe reads. Beebe couldn't participate in a vote involving Texas and Kansas, and Smith couldn't participate in one involving Ohio State. The real Smith had explained this process earlier. Whenever the Buckeyes are discussed next month, he won't be in the room. "I'll be in the other room," he said, "eating chocolate pretzels."
A warning to the real Beebe: If you partake in the chocolate pretzels every time you get bounced from the room because you're ineligible for a vote, you'll weigh approximately 427 pounds by the time Gumbel reads the bracket. Because Texas and Kansas were potential No. 1 seeds and because Baylor, Oklahoma State and Nebraska were potential bubble teams, Big 12 teams seemed to always be up for discussion or a vote. After seeing the red stripe pop up on fake Beebe's screen for a fourth consecutive vote, Glockner suggested a new revenue stream for the NCAA. "They should get Red Stripe beer to sponsor the red stripe," Glockner said. "Hoo-ray, Beebe!" Meanwhile, I wondered out loud whether Big East commissioner John Marinatto -- whose league might wind up with 10 or 11 teams in the tournament -- would ever be allowed in the room if he served on the committee.
After the fake Beebes returned to the room, we learned the committee had made Ohio State the No. 1 overall seed, followed by Texas, Pittsburgh and Kansas in that order. Next we moved on to the next four. The biggest debate came between San Diego State and BYU, which were competing for the sixth and seventh true seeds behind Duke. Several members called for a side-by-side comparison of the Aztecs and the Cougars, but the room stayed fairly quiet. As a few fake committee members quietly discussed the merits of the Aztecs and Cougars, Worlock reminded everyone that the actual committee would have that conversation for the entire room to hear. "Sidebar conversations are counterproductive," Worlock said. The volume of the debate increased immediately. BYU took the No. 6 seed. In a later vote in which fake Beebe did not participate, we flipped Pittsburgh and Texas, only to flip them again during the seed scrubbing process the following day.
After seeding a few more of the locks, we returned to the selection process. Selection works the same way as seeding. The teams under consideration appear on the screen, and committee members select their top eight. Then they rank the top eight vote-getters from No. 1 through No. 8. The top four vote-getters are placed into the field, while the bottom four are thrown back into the under consideration pool. "We go layer by layer by layer," Shaheen said.
Those layers make it difficult for committee members to think about how many teams from each conference have made the field. By the time one set of teams is placed into the field, it's time to debate the merits of eight more teams. This constant churn makes considering conference difficult. During the selection, I tried to figure out how many Big East teams we were putting in. I couldn't, because things were moving too fast.
The layers also make backroom deals almost impossible. For an AD or commissioner to broker a deal, he would have to first convince a majority of committee members to vote for a particular school during the "choose the eight best" vote. Since he wouldn't need to make a deal if the team were deserving, we'll assume it would be the eighth-best of the group that requires a numerical ranking from each voter. To get that team into the top four and into the field, he would have to convince several committee members to rank that team high -- probably in the top three -- to counteract the honest voters ranking that team No. 8. He'd also have to do it without being in the room for the major portion of the debate. On top of that, the voting would have to clear NCAA staffers, who probably would notice irregularities.
But if an athletic director wants his team to make the tourney, here's the secret. He needs to teach his coach how to schedule. The committee weighs nonconference schedule heavily because it is the one thing a program can control. If a BCS conference team brings in the dregs of bad leagues in the nonconference and hopes to coast on its conference record, it had better dominate its conference (hello, Alabama, which didn't make our bracket). If a non-BCS conference school wants to make the tourney, it had better win big in conference and play -- and win -- in exempt tournaments against tough competition early in the season. There is an alchemy to scheduling. The coaches who can do it well have a huge advantage over their colleagues.
As we filled out the field, mock conference tournament results rolled in. Some team probably got pushed out of the field when Portland beat Gonzaga and St. Mary's to win the West Coast Conference Tournament, but we had plenty of time to adjust to that. We had far less time to adjust to a shocker in the Atlantic 10 final on hypothetical Selection Sunday.
As we reached the point in the exercise corresponding to Sunday morning, Shaheen explained what the real committee members realize as they digest their bacon and eggs. "You're aware," Shaheen said, "that 150 million people are about to watch what you're getting ready to do that day."
The previous night, the Fourth Quadrant Committee -- which sounds like a regulatory body that should be discriminating against the X-Men -- determined true seeds for the champions of the leagues that won't provide any at-large teams. Next month, actual committee members will do this on Saturday night. For us, NCAA staffers filled in the blanks. This helped later as we filled out the lowest seed lines on the bracket.
We counted our at-large teams and then added the number of automatic bids remaining. We had at least one too many at-larges. No matter what happened in the conference tournaments, Michigan State wouldn't get an at-large bid. Marquette still had a chance. Since two at-large teams were playing in the Big Ten and SEC finals, we didn't have to worry about any surprises. But with Richmond facing Xavier in the A-10 final, we had to stay alert. If Xavier won, the Musketeers would take the league's automatic bid and Marquette would make the field as the final at-large team. Unfortunately for Marquette, Richmond pulled the upset. The Spiders took the A-10's auto bid, meaning Xavier snagged the at-large bid Marquette would have gotten.
We scrubbed the seeds again, debating to ensure we had each team in the spot it deserved to occupy. At this point, we engaged in two debates the actual committee is certain to face. First, we had trouble slotting North Carolina. Some in the room believed the switch to Kendall Marshall made the Tar Heels an entirely new team. Others believed North Carolina's complete body of work merited a lower seed. In the end, the Tar Heels were true seed No. 17. The other debate also involved an ACC team. Florida State lost forward Chris Singleton to a broken foot last week. The Seminoles are hopeful that Singleton can return for the tournament, but that return is questionable. We had to decide if Florida State deserves to make the tournament even if Singleton doesn't play. In the end, the Seminoles were one of the last four at-large teams in; they wound up matched against Gonzaga in a play-in game in Dayton with a No. 11 seed on the line.
When we finished scrubbing, we had a list of teams ranked No. 1 through No. 68. This true seed list (shown to your right) is what the committee uses to bracket. Only after all the teams are ranked are they placed in an s-curve format, and even that doesn't guarantee each team will wind up on a corresponding seed line. For example, Wisconsin should have been the fourth No. 3 seed according to the true seed list, but the Badgers had to be dropped to the No. 4 seed line to comply with bracketing principles and procedures. The committee has resisted releasing the true seed list each year, but Shaheen isn't entirely opposed to it in the future. If the committee wants to be more transparent without harming the integrity of the process, releasing this list along with the bracket would help.
With all the teams selected and seeded, we began bracketing. Shaheen reminded us that it was 4 p.m. -- two hours until the selection show on CBS. Sometimes, Shaheen said, the committee is still awaiting results of a conference tournament final. That requires the committee to make contingency brackets. Shaheen said the committee had six brackets going at once a few years ago.
One was all our fake committee could handle. The first three seeds were easy. Shaheen would click on each team's name, and a list of distances between the team's campus and the regional sites appeared. Because Ohio State was the No. 1 overall seed and Newark, N.J., is the closest regional site to Columbus, Ohio, the Buckeyes went into the East Region. Naturally, Texas went to San Antonio. Pittsburgh was next, so the Panthers went to New Orleans. Kansas, the fourth No. 1 seed, went to Anaheim, Calif.
Next, we stiffed Duke. It wasn't intentional. Actually, we tried to help the Blue Devils by placing them in the region closest to campus. Unfortunately, that was the East Region. So we had the No. 1 overall seed and the top No. 2 seed in the same region. Next, we placed BYU in Anaheim. We had to make sure the Cougars played in a Thursday/Saturday regional because the school -- which is owned by the Mormon church -- does not play on Sundays. (The committee learned that the hard way in 2003.) After we finished slotting the No. 2 seeds, the protests rang out. "You can't have Duke in Newark," someone said. "That isn't fair to Duke or to Ohio State."
So we voted to move Duke to New Orleans and put Georgetown in Newark. The Hoyas were our second Big East team. We had five on the first four seed lines, and that wreaked havoc because the principles and procedures require the committee to make every attempt to avoid matching conference foes until the regional final. I'd explain exactly how we figured out where to put the Big East teams to make it work, but my head would explode, and so would yours.
Later, we revisited the Duke move and made things even more of a mess. "Sean McManus is on the phone," Shaheen said, reminding us that the president of CBS Sports expected a finished bracket soon. We decided to leave Duke in New Orleans and plowed ahead.
In the first year of the 68-team field, we managed to make the First Four games relatively manageable for the teams and for the NCAA's television partners. The networks want one at-large vs. at-large game on Tuesday and one on Wednesday. That way, they aren't stuck with an entire night of low-major vs. low-major. Georgia and Butler will play Tuesday in Dayton, and the winner will face Arizona on Thursday in Washington. Florida State and Gonzaga will play Wednesday in Dayton, and the winner will face Syracuse on Friday in Chicago.
Unfortunately, we bracketed as slowly as the NCAA's computers tabulated votes. By the time we bracketed the final play-in games (Maine-Northwestern State and Texas Southern-Buffalo), Andy Rooney had probably finished complaining about those dangfangled new eight-track tapes. Despite our sloth-like pace, Shaheen smiled as we looked over our masterpiece. He knew a few days in Bracket Hell had generated some empathy toward the committee. Now, on the real Selection Sunday, we might not tear apart the bracket so viciously -- and so incorrectly.
He's correct. We probably won't. But we will look hard at Gene Smith's hands for chocolate pretzel residue.