FAQ: NCAA Selection Process

Despite a lot of publicity in recent years, the NCAA tournament selection and seeding process remains unclear to many people.

Ironically, the selection of the fields for the NCAA men's ice hockey tournament is one of the more clear-cut processes in collegiate athletics. As opposed to basketball, where considerable subjectivity is involved in selection and especially seeding, the ice hockey selections and seedings are a streamlined process which are, more or less, cut and dried. Subjectivity does enter the picture in seedings, but not nearly to the extent it's perceived.

PWR Explanation - A detailed look at how the selection criteria and comparison process works.

Between the Lines, 3/22/00 - This column is devoted to discussing the meaning of PWR and why it's relevant.

PWR Analysis: Selection And Seeding - A look at the 2000 field.

Understanding the Process - USCHO's first article, original published in 1997, that helped shed light on the selection process.

KRACH - An advanced mathematical model that ranks teams more fairly than PWR.

What is KRACH, and why is it better?

INTRO

Q: You say it's "cut and dried," but you qualify that with the term "more or less." Why?

A: There is essentially no subjectivity in selecting the field of teams (though there was, and continues to be, a little discretion in the process of seeding). That's why, for the sake of brevity, it has often been stated that the process was 100 percent cut and dried, since that miniscule part of the process that wasn't determined by a formula was still run by a specific set of steps.

Recently, with the advent of Atlantic Hockey and College Hockey America, those rules have been bent slightly, though in reality, it never comes into play anymore. See "What about the AHA and the CHA?" below.

Q: Who is on the selection committee

A: The current makeup of the men's Division I ice hockey committee can be found at the Division I Men's Tournament page. This page also includes a bracket, lists (such as past tournament winners), and other data.

BASICS

Q: How many teams are in the tournament?

A: 16 (as of 2003).

Q: How is the field broken down?

A: The tournament is split into four regions — East, Northeast, West and Midwest — with four teams in each. The No. 1 seed plays the No. 4, and No. 2 plays No. 3 in the first round. The winners play each other in the Regional Final, with those four winners advancing to the Frozen Four.

Q: How are the regions aligned?

A: The regions are split into East, Northeast, West and Midwest. The four top-ranked teams in the tournament are awarded No. 1 seeds and can come from anywhere, in accordance with the PWR.

The remaining 12 teams are then "banded" into groups of four, corresponding to Seed Nos. 2-4. An attempt will first be made to keep teams closest to their home region, but there is an emphasis on avoiding first-round intra-conference matchups. Factors such as maximizing gate revenue, and limiting travel costs can also be taken into consideration. As a result, seeds can be flip-flop with same-numbered seeds of other regions.

Though not in its stated list of guidelines, the committee has made a concerted effort in the two years the 16-team field has existed to maintain a strict bracket. In other words, teams are given overall seed numbers, 1-16. The brackets are arranged 1-6-8-9, 2-15-7-10, 3-14-6-11, 4-13-5-12. Adjustments are then made for factors such as avoiding intra-conference matchups, and host schools playing in their home region. But otherwise, the committee has attempted to adhere to this philosophy and a strict reading of the PWR list. Whether that continues in the future, is anyone's guess.

Q: How are the tournament teams determined?

A: Each established conference receives one automatic bid. Currently this includes all six Division I conferences. The rest of the teams are selected through a series of mathematical and other criteria. Either way, there is no subjectivity in the process of selecting teams for the tournament. See below for details.

Q: How do you receive an automatic bid?

A: Starting with the 2000-2001 season, the committee has elected to revert back to the practice of awarding only one automatic bid to each conference. Also starting with 2000-2001, the MAAC (now known as Atlantic Hockey) was awarded an automatic bid; and in 2002-2003, the CHA received its automatic bid. That places the total number of auto bids at six. Each of the conferences have elected to award their automatic bid to their postseason tournament champion.

Q: What is the "Colorado College rule"?

A: In 1994, Colorado College won the regular-season WCHA title, but did not receive a berth to the NCAA tournament. That's when the rule was put in awarding a second automatic bid to a conference's regular-season champion, and was thus nicknamed the Colorado College rule. That rule has since been rescinded.

Q: What is the "Clarkson rule"?

A: The so-called Clarkson rule said that any team which won its regular-season and conference tournament championship, would automatically be awarded a first-round bye in the NCAA tournament. This rule is no longer in effect.

Q: How is RPI calculated?

A: As of 2006-2007, the committee has again altered the method of calculating RPI — now 25 percent is your own winning percentage, 21 percent your opponent's winning percentage, and 54 percent your opponent's opponents winning percentage (25-21-54). "Opponent's winning percentage" is the average winning percentage of each opponent — not the total winning percentage based on the sum of all wins, losses and ties. This change from previous years should have the effect of more strongly emphasizing teams' own winning percentage in the RPI formula, because opponent's opponent's winning percentage — which is now more heavily considered — is an amalgamation of hundreds of games that tends to hover near .500 no matter what.

CHOOSING AT-LARGE TEAMS

Q: So certain teams earn their way into the tournament automatically. Are polls used to determine who else gets in the tournament?

A: Absolutely not. Polls — the opinions of selected voters — have no bearing whatsoever.

Q: But PWR and RPI are polls, no?

A: No. PWR (PairWise Rankings) and RPI are mathematical tools. RPI is one of the four criteria that makes up a single comparison between two teams. The number of comparisons won by one "Team Under Consideration" (or "TUC" — defined as a team in the top 25 of the RPI) vs. all other TUCs is that team's PWR rating. The TUC criterion was altered for 2006-07 in two ways: first, it used to be that all teams at .500 or above in RPI were TUCs. This was replaced by the top 25 rule. Also, previously teams winning their conference tournaments (and therefore receiving automatic bids to the national tournament) were then made TUCs even if they weren't before. This provision was removed, almost certainly as a direct response to the scenario that almost occurred in 2005-2006, in which Holy Cross could have made the NCAA tournament even by losing to Bentley in the Atlantic Hockey title game. That was because Bentley winning the tournament would have made it a TUC by rule, altering Holy Cross' record against TUCs in such a way as to let the Crusaders into the national tournament anyway. Ironically, of course, Holy Cross — which did win the AHA title after all — beat Minnesota in the first round of the NCAA tournament, thereby justifying its inclusion regardless.

Basketball relies heavily on RPI and judgment calls. Hockey takes RPI a lot farther, taking opinion out of the selection process.

See our detailed PairWise Rankings explanation.

Q: So how are the at-large teams selected?

A: Through a series of computations that help compare teams to each other (thanks to Keith Instone for some history):

Each Team Under Consideration (TUC) is compared to every other TUC on the basis of four criteria: Rating Percentage Index (RPI), head-to-head record, record vs. other TUCs and record vs. common opponents.

The "record in last 16 games" criteria was removed as of 2003.

A more thorough explanation of the comparison process is available elsewhere on USCHO.

The information the committee receives is in the form of a grid showing how each team compares to each other team using the above criteria. The Pairwise Rankings (PWR) were created by U.S. College Hockey Online to summarize this grid by totaling up the number of comparisons won by each team. PWR is not directly used by the committee, but the end result is nearly identical.

There is evidence to believe, however, that the committee now does also see a straight Pairwise Rankings-like list, or standings, if you will, and starts from there. Life has finally imitated art to a degree.

Q: So PWR is used to decide who is in, right?

A: Yes and no. The selection committee does not use the PWR in the form that you see it on USCHO. The important distinction is that the committee doesn't actually calculate a final grand total of comparison wins. They take the grid of comparisons, and compare one team to another using a series of head-to-head comparisons among that subgroup of teams. In essence, the effect of each method creates the same results. So while the committee doesn't call its method the PWR, the results are very similar.

The only thing missing from this process is the bonus points for "quality wins." The RPI bonus is a fixed and finite amount set in advance, though the actual amount is kept secret from the public. See the bonus section below.

Note: There is reason to believe the committee has now — thanks to unintentional evolution related to the gradual ubiquity of PWR, and how PWR has so pervasively seeped into everyone's mindset — begun to simply use the list of total comparisons won (i.e. PWR), and foresaken the grid.

Q: Can you give an example?

A: Why, sure.

The committee identifies the point at which all teams below a certain line are beaten in comparisons by all teams above the line. Thus everyone above the line is in the tournament. If everyone above eight beats everyone below eight, those eight are in. Then if everyone below 15 is getting beat by everyone above 15, those teams are out. The teams ranked nine through 15 are then the "bubble teams," to use familiar NCAA basketball terminology.

Note: The committee may not be that methodical in determining a "bubble," though, normally, pulling out a subgroup of debateable teams is a rather obvious process.

At this point, we must remember that the committee doesn't actually know who is eight or 15 or two or whatever on the PWR list. They never use the PWR list in that way. (Or do they? See Note at end of previous section)

They just receive a big mish-mosh printout of each team's comparison with every other team — pretty much like the raw data on USCHO. By going through and comparing teams to each other, it becomes obvious pretty quickly that certain teams are beating the other teams in so many comparisons, and thus it's obvious they are in the tournament — or vice versa.

What's left — the teams that have neither been admitted nor eliminated at this point — is the aforementioned "bubble." The committee looks at the grid of comparison wins among the bubble teams, and the teams which tally up the most wins are in.

And that's why we say that what the committee does is essentially PWR.

Q: Is choosing the field based on the opinion of the committee at all?

A: In practice, not really. In theory, only to the extent that AHA and CHA teams are considered under special limitations. See below.

Q: What about this "good wins" or "bonus points" criteria? Isn't this just a subjective fudge factor?

The so-called bonus for "good wins," instituted for the 2003 tournament, is not subjective. Teams get bonus percentage points added to their RPI for defeating teams that finish the season in the Top 15 of RPI. For 2006-07, the bonus criterion was altered to give points only for a road win, as opposed to the old system, under which home and neutral-site wins also counted, albeit less than road wins. This further limits the importance of this factor. The amount of bonus points awarded has not been made public by the NCAA, but it is a fixed and finite amount.

Because the committee chooses to keep the amount of the bonus secret, it opens itself up to people believing its used as a fudge factor. You'll have to trust us that it's not.

Q: Now, see here. Studly College lost in the quarterfinals of their conference tournament to Pocket Protector University. How can they still make the tournament?

A: No more importance is placed on the conference tournament games than any other games. Winning or losing those games means just the same as winning or losing any other game — except that if you win the whole tournament, you've got an automatic bid.

So losing a tournament quarterfinal would mean the team could drop in PWR, and maybe even miss the NCAA tournament because of that, but it wouldn't be solely based on those quarterfinal games, nor should it be.

Q: OK, but this team played such a weak schedule — they shouldn't be allowed in, should they?

A: Again, that will generally sort itself out in the criteria. Strength of schedule is a major factor of RPI, as is record vs. other Teams Under Consideration (i.e. the better teams).

There is, however, an exception, which is our next item.

Q: What about the AHA and the CHA?

A: Starting with the 1998-1999 season, the committee voted in the right to override the results of comparisons in instances where a team is from a conference with a drastically poor overall RPI. In essence, the PairWise system has trouble dealing with situations where a strength of schedule is intuitively known to be weak, but is too insular (i.e. too many games solely against teams from its own conference) to draw a valid conclusion from the results.

For example, teams in weaker conferences who only — or mostly — play teams from that same conference will win more comparisons than would otherwise happen, due to the fact that both winning percentage and strength of schedule are enhanced. By never playing "stronger" teams — and, more importantly, because their opponents never play the "stronger" teams — there is no way of telling just how accurate their strength of schedule indicator is.

Specifically, this addresses the situation in the AHA, most of whose teams play few non-league games against teams from the four major conferences, and to a lesser extent, the CHA. What it means in practice is that the committee has the authority to deny tournament bids to schools from these leagues even if they meet the PWR criteria for qualifying. In practice, this exception is no longer applied, since all six conferences tend to play one another with substantial regularity these days.

There are better ways to achieve the desired result, by "evening out the playing field," so to speak. Much more sophisticated mathmatical processes than PWR do exist, and USCHO publishes one of them, known as KRACH.

Q: So seeds are also done this way?

A: Not exactly, but more or less (there it is again). See below.

Q: Why is PWR wrong?

A: It hasn't been ... yet. However, because of the new "good win" criteria implemented in 2003, and because the NCAA doesn't publish exactly how much this impacts RPI, the theoretical possibility now exists that PWR won't correctly predict the NCAA tournament field.

SEEDING

Q: How is the tournament seeded?

A: First, the field of 16 is grouped, or "banded," into groups of four; four No. 1 seeds, four No. 2 seeds, four No. 3 seeds, and four No. 4 seeds. The process for doing this follows a similar process for selecting the at-large teams.

The No. 1 seeds are decided by taking, in essence, the top four PWR teams. They are given overall seeds of 1-4. They are then placed as close to their home region as possible, with preference going to the top overall seed. The remaining teams are also seeded as close to their home region as possible, but may be moved by taking into account factors mentioned below.

Q: How do they decide who gets moved?

A: In the past, when there was only two regions, it was a mandatory part of the procedure that two teams from each region would be moved. This was to ensure the flavor of a "national" tournament. Now, with four regions, the teams will be more spread out, and there is no explicit mandate to move teams. Instead, it appears the committee has made it an ultra priority to create perfectly "serpentine" brackets, without much regard to region (i.e. 1-16-8-9, 2-5-7-10, etc...).

Nevertheless, teams will be moved in order to intra-conference matchups in the first round. That remains a priority. As does the need to put schools hosting a regional into that regional. These are the two sacred cows for the committee.

Team are placed into bands of four. Four No. 1 seeds, four No. 2s, etc... Teams can be moved freely within those bands, but not otherwise. The committee has the option to do whatever it wants within those bands. It can set things up geographically, it can choose to help out a region for attendance factors, reduce travel costs, avoid second-round intraconference matchups, or any reason it wants.

It appears, however, the committee is determined to stick to a strict serpentine bracket whereever possible.

In the past, the seeding process was always subjected based on some of the rationale mentioned above, such as determining who will be a better draw in a region. This would often move better teams out of region, just because they weren't good draws in their home region. Nowadays, however, the commitment seems square on bracket integrity, regardless of other factors — though they are certainly still weighed.

Q: Who decides these guidelines?

A: These guidelines and procedures are discussed regularly with the entire body of Division I teams. The committee has the final say, but tends to abide by the sentiment of the member teams.

Q: What happened to the idea of flip-flopping the bottom teams of each region?

A: This was done in order to maintain a "national flavor" to the tournament. Since there was only two regions, the regionals could become too dominated by teams that always played each other during the season. With four regions and four teams in each, there is much more flexibility to move things around, and there is much more diversity built in. By simply mandating that intra-conference matchups should be avoided, the effect of a national tournament is there without needing to also mandate arbitrary switches.

Q: Why is everyone so gung-ho on avoiding intra-conference matchups?

A: Well, this is a matter of opinion, and not everyone agrees on the degree in which the committee should go to avoid these matchups. But we can explain the reasoning.

Since there are just four major conferences in college hockey and only 16 teams in the tournament, the likelihood of seeing first-round matchups similar to regular-season or conference tournament games is very high. The tournament should be a time for different teams to play each other.

In the case of second-round matchups, the committee weights advantages and drawbacks before moving teams around. For example, if two teams did not meet in their conference tournament, the committee may be more likely to allow a second-round matchup of teams from the same conference. Or perhaps the committee is concerned about attendance, a very real issue. Or, for another example, two teams may have played recently, but split the season series, in which case seeing them play again might be a plus. It appears, however, that since the advent of the 16-team tournament, the committee is not concerned about going out of its way to avoid second-round intra-conference matchups.

Q: What is the seeding of a team that wins an automatic bid?

A: Earning an automatic bid has no direct bearing on seeding. In the past, if a team won both the regular-season championship and conference tournament championship, taking both of the conference's automatic bids, that team would receive an automatic bye in its home region. But, starting in 2000-2001 and the awarding of only on automatic bid per conference, this is no longer the case.

Q: Four teams get No. 1 seeds and there are four major conferences, too. Doesn't each conference have to get a No. 1 seed?

A: No, although it often works out that way.

Q: Does winning a conference tournament improve your seeding?

A: Not directly. As in the case earlier, when we talked about conference tournament games, winning a tournament helps only in the sense that it means you won several games in the process.

Q: PWR shows that Team A beat out Team B. Why did Team B get a higher seed?

A: Once the teams are placed once and for all in their proper regions, to order them you compare them head-to-head, not by the overall number of comparisons won. (Note: This is probably obsolete. The committee, it seems, does just keep them in PWR order when giving teams overall seed numbers.)

FAQ: NCAA Selection Process

Related Articles

INTRO

BASICS

CHOOSING AT-LARGE TEAMS

SEEDING

INTRO

BASICS

CHOOSING AT-LARGE TEAMS

SEEDING