Home Bracketology Bracketology: Historical Analysis of the RPI for At-Large Teams

Bracketology: Historical Analysis of the RPI for At-Large Teams

by Ian

As we count down the days until Selection Sunday, everyone will be inundated with talk of who is “in” and who is “out” of the NCAA Tournament field. I updated my Bracket Projection today to reflect the results of the weekend’s games. Of course, we won’t know for sure who is in or out until Selection Sunday, and the only way to ensure a spot in the field of 68 is to win your conference tournament. The rest is all projection, but thanks to some great tools like the Team Sheets on Warren Nolan’s site, it has become increasingly easy for the common fan to try their hand at bracketology. Since ESPN has the broadcasting rights to most major conference games, the airwaves will be flooded with analysts citing Joe Lunardi’s Last Four In / First Four Out. To make matters worse, in-game analysts will proclaim a team “definitely in” after winning a single game. As The Bracket Matrix has shown, Lunardi is a slightly-below average bracketologist, he is just fortunate to have a bigger platform than the rest of us. One of the other drawbacks of ESPN is that they routinely force the BPI metric as a system for rating teams. BPI is useful, but like the KenPom rankings, it is not a metric that the NCAA Tournament Selection Committee uses. The NCAA did add additional ranking information (including BPI and KenPom) to the Team Sheets this year under the labels of “Results-based” and “Predictive” metrics.

The RPI is not a perfect metric of analysis, and many statisticians have developed better models for quantitatively ranking teams. That said, the NCAA Selection Committee uses the RPI as one of their metrics, not just as a way to rank teams, but as the framing for the entire tournament selection discussion. The placement of opponents on the Team Sheets is based on the opponents RPI. Every discussion of a team’s Tier 1 wins or losses against Tiers 3 and 4 is all framed by the RPI. Those data points are inherently tied to the RPI because “Tier 1 wins” really means “Home wins over teams in the RPI Top 30, Neutral wins against the RPI Top 50, and Road wins against the RPI Top 75.” Love it or hate it, the RPI frames the entire tournament discussion, so it has to be taken into consideration when discussing potential tournament teams. For the last few years, the NCAA released their RPI calculations on a weekly basis as part of an effort to bring some transparency to the Selection Committee process. I have been tracking the RPI of Tournament teams for the last 10 years (since the 2008 Tournament) and compiled data on how RPI compared to the probability of a team making the tournament.

The NCAA Tournament has evolved in the 10 years since 2008. In 2008, the field was comprised of only 65 members (31 Automatic Bids and 34 At-Large Bids) and there was only one play-in game. In 2011 the field was expanded to 68 teams with 37 teams receiving At-Large bids and 4 Play-in games occurring – 2 between the bottom 4 teams in the field and 2 between the lowest seeded 4 At-Large teams. With the breakup of the Big East and the formation of the American Athletic Conference in 2014, 32 Automatic Bids were awarded and the At-Large qualifiers shrunk to 36. Over the last ten years, there have been 357 At-Large bids awarded by the Selection Committee. Of those 357 bids, 283 (79%) were awarded to teams from Major conferences and 74 (21%) to teams from smaller conferences.

For purposes of this article, I’m defining the “Major” Conferences as the “Big Six” from 2008-2013 (Big Ten, ACC, SEC, Big XII, Pac 10/12, and Big East). “Mid-Major” Conferences are the other 26 leagues that receive automatic bids to the NCAA Tournament. For 2014, I kept the Big East as a “Major” Conference and defined the AAC as a “Mid-Major” conference. As I will note later, this trend was followed in the seeding of AAC teams in 2014 and 2015 but the Committee broke from that tradition and treated the AAC more like a Major Conference in 2016. Based on the seedings received by AAC teams in 2017, the Committee treated them like Mid-Majors once again. During the last decade the highest RPI of any team to receive an At-Large bid was USC’s 73 in 2011. The only other team with an RPI over 65 to get an At-Large bid was Syracuse (RPI 70) in 2016. It should be noted that Syracuse’s worst losses did occur while Jim Boeheim was serving a suspension, which was not accounted for by the RPI but may have been taken into consideration by the Committee. The Committee was rewarded with their selection of the Orange, who made a run to the Final Four. Given that the highest RPI for an at-large team in my data set was 73, my analysis focuses on teams with Top-75 RPIs. As you may expect, the committee has been much more forgiving of Major Conference Schools for lower RPIs than they are of Mid-Major schools.

The Top 30 Guarantee

St. Bonaventure defeated Rhode Island last week. The win lifted the Bonnies into the Top 30 of the RPI, which is good news for their NCAA Tournament hopes.

Over the last ten years, 298 teams have entered Selection Sunday with a Top 30 ranking in the RPI (2 Top 30 teams were ineligible for the Tournament because of NCAA violations). Of those teams, 98 were conference champions that received automatic bids to the Big Dance. For the other 200, all except two received invitations to the Tournament. The only two to miss were Colorado St in 2015 and St Bonaventure in 2016 who both had RPIs of 30. All 192 at-large eligible teams with RPIs under 30 received bids. Major Conferences accounted for 153 of the 198 at-large bids and Mid-Majors for 45.

Currently, there are a few Mid-Major situations where this guarantee should come into play. Rhode Island and Nevada are both leading their respective conferences and have RPIs in the Top 10. Rhode Island should feel confident about their Tournament status, but Nevada’s 2 Tier 3 losses could be cause for concern if they fall in the Mountain West Tournament. In the AAC, Cincinnati and Wichita State both have Top 20 RPIs and should feel safe while Houston’s RPI surged to around 20 when they beat Cincinnati last week. The Cougars have a Tier 3 and a Tier 4 loss on their resumes and need to avoid another bad defeat down the stretch. While Rhode Island is definitely in from the Atlantic 10, it will be interesting to see how the Committee handles St. Bonaventure. The Bonnies RPI just rose into the Top 30 with their win over Rhode Island on Friday but like Nevada they also have 2 losses to Tier 3 opponents. In Conference USA, Middle Tennessee’s RPI has been hovering in the mid-20s all season after the Blue Raiders played a tough non-conference schedule. They have only 1 bad loss but are also short on top-end wins which could also put this guarantee to the test if they fail to win an automatic bid.

Major Conference Surety: Be in the Top 50

UCLA and Utah both have RPIs between 40 and 50 and are fighting for their Tournament lives on the Bubble.

Major Conference schools with Top 50 RPIs have fared very well on Selection Sunday. All 49 schools that had an RPI between 31 and 40 were awarded At-Large bids. Only four Major Conference schools have missed the Tournament with Top 50 RPIs (Ole Miss and Ohio State in 2008 who had RPIs of 48 and 49 and Missouri and Minnesota in 2014 who had RPIs of 49 and 50). This means that 92% (44 of 48) of Major Conference schools with an RPI between 41 and 50 have gone dancing and the only ones that have missed have had RPIs of 48 or higher. There are some interesting cases this year that will put this theory to the test. The most interesting scenario involves four Pac-12 teams. USC (RPI 41) has just 2 Tier 1 wins and no wins over teams that should be considered a lock to make the field (their best are over Middle Tennessee St and New Mexico St). USC has only one loss outside the Top 100 but overall has little exciting about their resume. UCLA (RPI 48) has two excellent wins in a road win at Arizona and a neutral court win over Kentucky but have 2 Tier 3 losses. Washington (RPI 49) has the best wins of the group, a road win at Kansas and home wins over Arizona and Arizona State. However, the Huskies are just 1-4 against Tier 2 opponents and have terrible numbers in the predictive metrics, which skews their overall metric average to be well below that of other bubble teams. Finally, Utah (RPI 50) has a sweep of Washington along with a road win at Arizona State and a home win over Missouri. All four of these teams have 5 victories against Tier 1 and 2 combined and no one has lost more than 1 game outside the Top 100.

Major Conference Gamblers: 51 to 60

NC State has wins over Duke, North Carolina, and Clemson. Will they be enough to get the Wolfpack in the Big Dance?

There is a clear delineation between the Top 50 and those outside the Top 50 in the RPI for Major Conference schools. After the Top 50, only 57% of schools have received At-Large invitations (29 of 51). Most famously, Virginia Tech has been shut out twice (2008 and 2010) with an RPI of 52. Having an RPI in the 50s is a gamble and brings other factors such as strength of schedule and strength of victories into question. Teams with RPIs in this range are a mixed bag and incredibly resume-dependent. In 2009, Boston College received a 7-seed with an RPI of 60 while Florida, Baylor, Georgetown and South Carolina were left out, despite having RPIs between 54 and 58. In 2016, a trio of SEC teams were the highest-ranked major conference schools to be left out. Florida (RPI 54), Georgia (RPI 60), and South Carolina (RPI 62) all went to the NIT while Vanderbilt (RPI 63) and Syracuse (RPI 70) both got at-large bids. Last season, Providence (RPI 56), Kansas State (RPI 57), and Marquette (RPI 61) were the lowest-ranked Major Conference schools to receive At-Large bids. Meanwhile, Georgia (RPI 52) and California (RPI 53) were left out and went to the NIT. Interestingly enough, Providence and Kansas State were included in the “Last Four In” while Marquette received a 10-seed. This season, the gamblers appear to be hailing from the ACC and Big XII. Louisville (RPI 51), Virginia Tech (RPI 55), and North Carolina State (RPI 59) are all battling in the middle of the ACC pack while Texas (RPI 52) and Kansas State (RPI 56) have both performed well in a crowded Big 12.

How did they get in?

In 2016 Syracuse received an At-Large bid with an RPI of 70 and made a run to the Final Four.

For Major Conference schools with RPIs between 61 and 70, only 7 out of 39 possible teams (18%) have been awarded At-Large bids. In 2009, Arizona was one of the last teams in with an RPI of 62 and a 12-seed in the tournament. The Wildcats then proceeded to defeat 5th-seeded Utah and 13th-seeded Cleveland State en route to the Sweet 16 In 2012, West Virginia was awarded a 10-seed despite and RPI of 64 and subsequently got pounded by Gonzaga in a game that I attended at the CONSOL Energy Center. As I already mentioned, USC received a bid in 2011 with an RPI in the 70s and in 2016 Syracuse got a bid with an RPI of 70 and made a run all the way to the Final Four. Last season, Marquette finished with an RPI of 61 but had 8 Top 50 wins, including one over top-ranked Villanova, that bolstered their resume. The teams to keep an eye on this year in this range hail from the SEC. Florida’s recent losing streak has plummeted their RPI to 63, but 6 Tier 1 wins should be enough to keep the Gators in the Tournament conversation. LSU has 7 Tier 1 victories, including sweeps of Texas A&M and Arkansas, but the Tigers are just 1-5 against Tier 2 and have an RPI of 75. Georgia (RPI 66) could also work their way into the conversation if their 3 Tier 3 losses don’t hold them back too much.

The Mid-Major Disadvantage

The West Coast Conference has been a battle between St. Mary’s and Gonzaga, but both have RPIs below 30. Could one of them be in danger of not making the Tournament?

Even though every team with an RPI under 30 has made the Tournament in the last 9 years, there is a sharp drop-off for Mid-Major schools after the 30th spot in the RPI. For Mid-Majors with an RPI between 31 and 40, only 14 of 30 (46.7%) have received At-Large bids. Besides the two teams mentioned earlier (2015 Colorado St and 2016 St Bonaventure) who missed the field with RPIs of exactly 30, 2013 Southern Miss (RPI 31), 2016 San Diego St (RPI 32), 2014 Southern Miss (RPI 32) and 2008 Dayton (RPI 32) squads were the highest ranked teams not to receive an At-Large bid. Illinois State was the highest-ranked Mid-Major to not get a bid last year, the second time the Cardinals have been shut out with RPI of 33 (2008 and 2017). As RPIs increase to the 40s, the likelihood of a Mid-Major team receiving a bid remains about the same as 13 of 30 (43.3%) have received bids. It is worth noting that 9 of these 13 came from either the A-10 and Mountain West, two of the biggest Mid-Major Conferences. Within the same RPI range of 41-50, Major Conference teams have more than double the rate of at-large bid awards as Mid-Majors (91.7% to 43.3%). This is a clear disadvantage for Mid-Major schools.

The biggest discrepancy for Mid-Majors comes in the next segments. Between 2008 and 2015, no Mid-Major school with an RPI over 50 was awarded an At-Large. The 2016 bracket was a bit of an anomaly as two teams from the American Athletic Conference with RPIs over 50 received at-large bids. At the outset I mentioned that I categorized the AAC as a mid-major, which fit their trends in 2014 and 2015. In 2014, SMU (RPI 53) was left out while UConn (RPI 21) was drastically under-seeded as a 7-seed. The Huskies defied that seeding and went on to win the NCAA Tournament. In 2015, Temple (RPI 38), and Tulsa (RPI 45) were both left out of the field while SMU received a 6-seed despite an RPI of 12 (the only team in the Top 13 of the RPI not to be at least a 4-seed). However in 2016 the Committee reversed course. Cincinnati (RPI 47) was awarded a 9-seed, Temple (RPI 59) was awarded a 10-seed, and Tulsa (RPI 61) was the last team in the field. Those numbers do not at all match up with the other mid-major trends and are certain outliers, much more in line with the Major conference RPI trends.  The Committee seemed to return to the opinion of the AAC as a Mid-Major in 2017, seeding Cincinnati (RPI 12) and SMU (RPI 13) on the 6-seed line and leaving Houston (RPI 54) out. This season, three teams from the AAC have high RPIs (Cincinnati 11, Wichita St 16, Houston 19) and Cincinnati was placed on the 2-seed line in the Committee’s initial bracket reveal.

There is a long way to go until Selection Sunday, but some surprising teams may find themselves closer to the bubble than initially expected. For most of the season, the West Coast Conference was expected to be a two-bid league with St. Mary’s and Gonzaga. However, both have RPIs below 30. Gonzaga (RPI 43) has enough quality wins on their resume that they should be okay come Selection Sunday. The lowest-ranked Mid-Majors to receive an At-Large bid had RPIs of 46 (UNLV in 2010, LaSalle in 2013, and Wichita St in 2016). St. Mary’s may not be as comfortable as some may think. The Gaels have an RPI of 32 but have just on Tier 1 victory and 2 losses against Tier 3 opponents. St. Mary’s rates well with the predictive metrics but on the whole has a very similar resume to 2016 when they were left out by the Committee.

In addition to the West Coast Conference teams, Buffalo (RPI 30), Louisiana-Lafayette (RPI 40), Loyola-Chicago (RPI 42), and New Mexico State (RPI 47) will be sweating it out come Selection Sunday if they fail to win their conference tournaments. Only five mid-major teams, Wichita St (RPI 16), Houston (RPI 19), St. Bonaventure (RPI 27), St. Mary’s (RPI 32), and Temple (RPI 44),  are currently ranked in the Top 50 of the RPI and are not leading their respective conferences.

Obviously there are a lot of other factors that go in to the Selection Committee’s decision-making process, but when looking at the rough percentages of teams that have been awarded At-Large bids, there are clear delineations for which teams have increased probabilities of making the Dance. Major Conference teams with Top 50 RPIs are almost guaranteed to get a bid while Mid-Major schools basically need to be ranked 20 places higher than their Major-conference counterparts to feel a similar level of comfort. While RPI can be used for roughly determining who should or should not be in the field of 68, it is not a good indicator for where teams will be seeded. In conclusion, RPI isn’t the be-all-end-all for the Committee, but keeping a close eye on a team’s ranking in the RPI should give a good indication of their chances to make the field. This may change to some extent this year with the addition of other metrics to the Team Sheets, but it is important to remember that the sheets are still organized and sorted by RPI. Of course, every school in the country has the opportunity to receive an automatic bid to the Dance by winning their conference tournament and taking the decision out of the Selection Committee’s hands entirely.

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.