Sunday, February 11, 2018

2/11/18 Bracketology

Teams are ranked in order by seed. Auto bids are in bold and are determined via KenPom projections, with tiebreaker being awarded to the team ranked higher in KenPom’s rankings.

1 seeds: Virginia, Villanova, Xavier, Purdue

2 seeds: Auburn, Kansas, Duke, Cincinnati

3 seeds: Clemson, Texas Tech, Michigan State, North Carolina

4 seeds: Tennessee, Ohio State, Arizona, Oklahoma

5 seeds: Rhode Island, Gonzaga, West Virginia, Texas A&M

6 seeds: Wichita State, Creighton, Miami FL, Florida

7 seeds: TCU, Arizona State, Kentucky, Missouri

8 seeds: Alabama, Florida State, Seton Hall, Butler

9 seeds: Virginia Tech, Nevada, St. Mary’s, Michigan

10 seeds: Texas, Houston, Arkansas, Kansas State

11 seeds: Providence, Syracuse, USC, Louisville, Temple, Washington

12 seeds: Middle Tennessee, New Mexico State, Vermont, Loyola Chicago

13 seeds: UL Lafayette, East Tennessee State, South Dakota State, Buffalo

14 seeds: Murray State, Montana, Northern Kentucky, UC Santa Barbara

15 seeds: Bucknell, Rider, College of Charleston, Florida Gulf Coast

16 seeds: Wagner, Penn, UNC Asheville, Nicholls State, Savannah State, Arkansas Pine Bluff

/**********************************************************************************/
Last 4 In: USC, Louisville, Temple, Washington
First 4 Out: Saint Bonaventure, UCLA, NC State, Baylor
/**********************************************************************************/

New to the bracket: Syracuse, Temple, Murray State, Savannah State

Removed from the bracket: NC State, UCLA, Belmont, North Carolina A&T



Review and Takeaways From Top 16 Reveal

The NCAA Tournament Selection Committee has released its top 16 teams, and the one question I had going into the exercise was how much of a factor having the predictive-based metrics on the Team Sheets for the first time ever would be. The committee’s rankings show that it is playing more of a factor than in years past, and as such, bracketologists like myself will have to adjust our methodology accordingly. Below is the committee’s ranking alongside my projection and the composite projection from bracketmatrix.com, a site that compiles the one hundred or so projections like mine:



The only team that made the top 16 that myself and most bracketologists did not project is Oklahoma. I personally was only able to nail down 7 teams exactly to their seed, far below my expectations (for reference, last season I had 12 of the teams seeded correctly for this exercise). However, this was the mid-term exam; a learning opportunity to prepare for the final examination on Selection Sunday.

The reason I say the predictive-based metrics are playing more of a factor than years past is highlighted by the following selections by the committee:
  • Cincinnati on the 2 seed line – their RPI ranking is 10th, but their consensus predictive-based ranking is 4th. While they do have a record of 12-2 against Tier 1 and Tier 2 opponents, none of the wins stand out when compared to the other teams ranked around them (their best win is a neutral court win over Buffalo). Their SOS is also poor relative to the top 16 teams.
  • Clemson on the 3 seed line – the Tigers are 4th in the RPI, but have a consensus predictive-based rankings of 18th. Their SOS is strong, and they have beaten North Carolina and Ohio State. A 3 seed would be the lowest the 4th ranked team in the RPI has been seeded in the past 5 years.
  • Purdue on the 1 seed line – Purdue is 11th in the RPI, but 3rd in the predictive-based metrics. An RPI ranking of 11 would be the worst RPI ranking for a 1 seed in the past 5 seasons.
  • The absence of Rhode Island from the top 16, who ranks 5th in the RPI. No team in the top 5 of the RPI has ever been seeded worse than a 2 seed in the past 5 seasons.

It sounds like the committee is adding emphasis on road and neutral court wins beyond the already stratified quadrant system. They mentioned how impressive the four 1 seeds’ records away from home was as a basis for their seeding.

I could easily act like a child and blast the committee and their entire process since it doesn’t meet my own biases like some people have already, but being the actuary that I am, I will study the discrepancies between my prediction and the committee’s actual results and will be determined to perform better on Selection Sunday.


Top 16 Projection for 2/11/2018 - Factoring in Saturday's Games

Last season the NCAA Tournament Selection Committee released their top 16 on a Saturday afternoon, prior to the games being played. As people that follow college basketball know, Saturdays are the biggest day for the game, with nearly every team playing. This season, the committee is releasing the top 16 projection on Sunday, and it is uncertain how much of a factor Saturday's results will play into it. I personally think it should be factored in, as one team's profile has materially changed in just 24 hours, showing the challenges of the art of bracketolgoy.

As such, I have revised my top 16 projection in anticipation of the committee factoring in Saturday's results.

1 seeds: Virginia, Villanova, Xavier, Kansas
2 seeds: Purdue, Clemson, Auburn, Texas Tech
3 seeds: North Carolina, Cincinnati, Tennessee, Duke
4 seeds: Michigan State, Ohio State, Arizona, Rhode Island


Teams just missing the cut (aka the 5 seeds): Wichita State, Oklahoma, West Virginia, Texas A&M




The profile that has materially changed is Michigan State. Prior to Saturday, the Spartans did not have a profile deserving of a top 16 ranking. On Saturday, they beat Purdue, picking up their best win of the season. But that's not the only thing that changed. Due to Notre Dame's upset of Florida State and UConn playing Wichita State, the Spartans retroactively picked up two Tier 2 wins on Saturday. That means they went from having 5 total Tier 1 + Tier 2 wins (the lowest of the teams I considered) to 8, all in the matter of 24 hours. Their SOS has improved from 94th to 79th. That is still the worst of all teams considered, and I think that caps them at a 4 seed for now.

The other change was moving Texas Tech up to the 2 seed line, replacing them with Tennessee. Tennessee got blown out at Alabama on Saturday and Texas Tech picked up a Tier 1 win at Kansas State. I would not be surprised to see North Carolina get the final 2 seed, however, given their volume of Tier 1 wins (7 compared to just 4 for Texas Tech).

Saturday, February 10, 2018

Top 16 Projection for 2/11/2018

The NCAA Tournament Selection Committee is releasing their top 16 teams on Sunday. This is the second year the committee has done this practice, and their goal is to shed some light on the selection and seeding process. Below is my attempt to mimic what their ranking will be, as well as what region the teams will play in. Note that this projection is reflective of results through Friday 2/9, which is consistent with the committee’s process.

Teams below are ranked in order by seed.

1 seeds: Virginia, Villanova, Xavier, Kansas
2 seeds: Purdue, Clemson, Auburn, Tennessee
3 seeds: Cincinnati, Texas Tech, North Carolina, Duke
4 seeds: West Virginia, Ohio State, Oklahoma, Kentucky

Teams just missing the cut (aka the 5 seeds): Rhode Island, Michigan State, Miami FL, Arizona



This season the committee adjusted the Team Sheets they use to select and seed the tournament field to place more emphasis on road and neutral court games. Instead of using a Top 50, Top 100, Top 200 RPI breakout as in years past, the committee is using a four tier stratification that adjusts for location of the game. 

Below is the stratification:

RPI Tier 1: Home (1-30)                  Neutral (1-50)                    Away (1-75)
RPI Tier 2: Home (31-75)               Neutral (51-100)               Away (76-135)
RPI Tier 3: Home (76-160)             Neutral (101-200)             Away (136-240)
RPI Tier 4: Home (161-351)           Neutral (201-351)             Away (241-351)

Below is an explanation of my rankings.

Virginia and Villanova are the consensus top two teams – pick them in any order you want, I have no problem with it. After that, I believe there are three teams fighting for the final two 1 seed spots: Kansas, Purdue, and Xavier. Below are the Cliff Notes versions of each team’s RPI profile:

Kansas – Kansas is tied for the most Tier 1 wins with Villanova, each having 9, and they have the most Tier 1 + Tier 2 wins at 13 total. Their strength of schedule (SOS) is #1, due to the strength of the Big 12 conference and having a top 10 non-conference SOS. Four of their five losses have come at home, with one of them coming against a Tier 3 opponent (i.e. a bad loss).

Purdue – Purdue has 5 wins against Tier 1 opponents and 11 wins against Tier 1 + Tier 2 opponents. Their SOS ranks 43rd overall, with the non-conference portion ranking 83rd. None of their three losses fall into the “bad loss” category, and they are a perfect 7-0 in road games, with 5 of those coming against Tier 1 and 2 opponents.

Xavier – Xavier has 6 wins against Tier 1 opponents and 12 wins against Tier 1 + Tier 2 opponents. Their SOS ranks 10th overall and their non-conference SOS ranks 32nd. All three of their losses fall into Tier 1 and come against teams projected to make the NCAA Tournament, including a loss at Villanova for which you cannot really ding them.

I think Xavier has an edge over Purdue and Kansas right now, so it comes down to the final 1 seed, and I believe Kansas is more deserving at this point in time. This is really, really close, and there are very solid arguments for both teams, but ultimately the committee values quality wins and SOS the most, and they will forgive a bad loss if there is sufficient evidence to suggest that was a one-off occurrence. As stated above, Kansas has the most quality wins of any team in the country, which should offset the home loss to Oklahoma State in the committee’s eyes. The other item that hurts Purdue’s chances is its RPI ranking – over the past 5 seasons, the worst RPI ranking for a team to receive a 1 seed was 8th (2017 Gonzaga, 2014 Virginia, and 2013 Indiana). As of this writing, Purdue is 12th in the RPI. Placing Purdue on the 1 seed line at this point in time would require the committee to do something that has not been done in recent history.

After those four teams, I think Clemson is safe as the 6th overall team. Then things start to get interesting. 

I think there are four teams under consideration for the final two 2 seed slots: Auburn, Cincinnati, Tennessee, and Texas Tech. I went with Auburn and Tennessee. While Cincinnati is a very impressive 12-2 against Tier 1 and Tier 2 opponents, including 7-2 in road and neutral court games against that cohort, their SOS ranks 68th overall, which is the second worst of the 20 teams I compared for this exercise. Auburn and Tennessee have similar road and neutral court records against that cohort, but with a SOS ranking of 36 and 13, respectively. I think the SOS difference will knock Cincinnati down a seed line. Texas Tech has the worst non-conference SOS of the four teams in this comparison, ranking 195th nationally, and similar to Cincinnati, I think that pegs them down a seed line.

North Carolina and Duke round out the 3 seeds. The Tarheels have a really bad home court loss to Wofford (ranks 138th in the RPI), but they also have wins over Clemson, Duke, Tennessee, and Ohio State, all four of which are projected to be in this top 16 release. As mentioned in the Kansas/Purdue comparison, the committee will forgive a bad loss if it is backed up with quality wins. Duke meanwhile does not have the number of quality wins as the teams above them, but more than the teams below them.

Next comes West Virginia. The Mountaineers have one item on their resume that no other team in the country can tout: a win over Virginia. West Virginia has five wins against Tier 1 opponents, which is more than Duke, Cincinnati, and Auburn. But what hurts their profile is their non-conference SOS being ranked 191st nationally due to 7 of their non-conference games being against Tier 4 opponents. That has resulted in them being ranked 25th in the RPI. Only once in the past five seasons has a team with that low of an RPI received a top 4 seed (2015 Georgetown). The lowest a top 3 seed has ever been ranked in the RPI over the past five seasons was 18th (2015 Notre Dame and 2016 Texas A&M). So, similar to my reasoning for Purdue being left off the 1 seed line, I am leaving West Virginia off the 3 seed line. I think there is a definitive cutoff at this point, and I think there are five teams vying for the final three spots.

Ohio State’s win at Purdue on Wednesday night got them a much-needed Tier 1 road win. Prior to that win, their best wins were at home over Michigan State and Michigan. The game also boosted their SOS up to 32nd nationally. While they do have a Tier 3 loss, something only Kansas and North Carolina also have of the 20 teams analyzed for this exercise, I think the win at Purdue more than offsets it.

Oklahoma has six Tier 1 wins; only the four 1 seeds have more Tier 1 wins, and so I think this will keep Oklahoma in the top 16 for now. The final selection comes down to Kentucky, Michigan State, and Rhode Island. Here is a snapshot of their profiles:



I went with Kentucky as they have the most Tier 1 + 2 wins and a significantly better SOS, which are the primary measures the committee uses. This means I am leaving out Rhode Island, which currently ranks 5th in the RPI. No team that high has been left outside the top 8, let alone the top 16, but upon digging deeper into the resume I do not think the committee gives them a 4 seed. The Rams played the 3rd toughest non-conference schedule in the nation, due in part to playing teams from one-bid leagues that are projected to win their conference (Charleston, Florida Gulf Coast, Iona, UNC Asheville). Unfortunately for Rhode Island, all of those games came at home. Their two Tier 1 wins are on a neutral court against Seton Hall and at home vs. Providence. That second one is right on the borderline of a Tier 1 and Tier 2 win, showing the need for deeper digging than a straight comparison of records at arbitrary cutoff points.

Now for the biggest surprise: I have left Michigan State out of the top 16. The Spartans are 23-3 on the season and rank 19th in the results-based metric RPI, but are a consensus top 6 team according to the predictive-based metrics KenPom, Sagarin, and BPI. The problem for Michigan State is their resume features just 1 win over team that is currently projected to make the field, a neutral court win over North Carolina. Michigan State is historically known for scheduling tough out of conference – that is not the case this season, as the Spartans’ non-conference strength of schedule ranks 233rd (out of 351) as of Saturday morning. That weak of a schedule means the opportunities to pick up quality wins out of conference just were not there. That is compounded in this particular season due to the middle of the Big Ten conference being the weakest it has been in nearly a decade, resulting in an overall SOS of 94th. For further perspective, that overall SOS ranking is the worst of all 20 teams I considered for this top 16 exercise.

Only Purdue, Ohio State, Michigan State, and Michigan are projected to make the NCAA tournament from the Big Ten conference, and as of Saturday morning the Spartans are 0-2 against those teams. This all leads to Michigan State being just 2-2 against Tier 1 opponents and 3-1 against Tier 2 opponents. The sum of 5 wins against Tier 1 and Tier 2 opponents was the lowest total among the 20 teams considered for this exercise (starting to see a pattern here?). Take the name “Michigan State” off the profile and slap on “Cincinnati” or “Gonzaga” on it and there probably isn’t much “potential 1 seed” chatter from ESPN around a team with that type of profile. This is absolutely name recognition bias on ESPN’s part to be touting Michigan State’s profile worthy of a 1 or 2 seed. Note that last season for the top 16 release I explained precisely why Wisconsin (who was ranked 7th in the AP poll at the time and 10th in the KenPom rankings) was not going to make the top 16, and it was for very, very similar reasons as to why I believe Michigan State will not make the cut this year.

Michigan State has the most intriguing resume to me, and we are all going to learn just how much of a factor having the predictive-based rankings (KenPom, Sagarin, BPI) on the Team Sheets for the first time ever will be. Based on their RPI profile, the Spartans are not deserving of a top 16 rankings – but the predictive-based metrics love them enough to suggest they are one of the 6 best teams in the country. This type of profile is exactly why I am in favor of the committee releasing a top 16 list at this point in the season, as it allows the public to learn and understand exactly what matters to the committee. If the Spartans do in fact make the top 16, then that tells me the committee is putting more weight into the predictive-based rankings than in years past, and I will adjust my methodology accordingly.


Sunday, February 4, 2018

2/4/18 Bracketology

Teams are ranked in order by seed. Auto bids are in bold and are determined via KenPom projections, with tiebreaker being awarded to the team ranked higher in KenPom’s rankings.

1 seeds: Villanova, Virginia, Purdue, Xavier

2 seeds: Auburn, Clemson, Kansas, Cincinnati

3 seeds: Duke, Tennessee, Texas Tech, Oklahoma

4 seeds: North Carolina, Kentucky, Arizona, Michigan State

5 seeds: Seton Hall, West Virginia, Rhode Island, Miami FL

6 seeds: Gonzaga, Florida, Alabama, St. Mary’s

7 seeds: Ohio State, TCU, Wichita State, Creighton

8 seeds: Nevada, Texas, Florida State, Arizona State

9 seeds: Texas A&M, Butler, Louisville, Michigan

10 seeds: Arkansas, Providence, USC, Missouri

11 seeds: Washington, NC State, Houston, Kansas State, Virginia Tech, UCLA

12 seeds: New Mexico State, Middle Tennessee, UL Lafayette, Vermont

13 seeds: Loyola Chicago, East Tennessee State, Montana, Northern Kentucky

14 seeds: Belmont, Buffalo, South Dakota State, UC Santa Barbara

15 seeds: Bucknell, Florida Gulf Coast, College of Charleston, Rider

16 seeds: Wagner, Penn, UNC Asheville, Nicholls State, North Carolina A&T, Arkansas Pine Bluff

/**********************************************************************************/
Last 4 In: Houston, Kansas State, Virginia Tech, UCLA
First 4 Out: Syracuse, SMU, Boise State, Saint Bonaventure
/**********************************************************************************/

New to the bracket: NC State, Virginia Tech, UCLA, Belmont, College of Charleston, Rider, Penn, UNC Asheville, Nicholls State, Arkansas Pine Bluff


Removed from the bracket: SMU, Syracuse, Marquette, Murray State, Northeastern, Iona, Princeton, Stephen F. Austin, Winthrop, Texas Southern


Sunday, January 28, 2018

1/28/18 Bracketology

Teams are ranked in order by seed. Auto bids are in bold and are determined via KenPom projections, with tiebreaker being awarded to the team ranked higher in KenPom’s rankings.

1 seeds: Villanova, Virginia, Purdue, Kansas

2 seeds: Duke, Xavier, Auburn, Clemson

3 seeds: Oklahoma, Cincinnati, North Carolina, Tennessee

4 seeds: Arizona, West Virginia, Texas Tech, Florida

5 seeds: Kentucky, Wichita State, Seton Hall, Michigan State

6 seeds: Ohio State, TCU, Arizona State, Rhode Island

7 seeds: Arkansas, Texas, Creighton, Alabama

8 seeds: Gonzaga, Louisville, Florida State, Miami FL

9 seeds: Butler, Michigan, Nevada, Texas A&M

10 seeds: Providence, St. Mary’s, USC, SMU

11 seeds: Syracuse, Marquette, Missouri, Houston, Washington, Kansas State

12 seeds: New Mexico State, Middle Tennessee, UL Lafayette, Loyola Chicago

13 seeds: East Tennessee State, Vermont, Murray State, Northern Kentucky

14 seeds: Montana, Buffalo, South Dakota State, UC Santa Barbara

15 seeds: Bucknell, Northeastern, Iona, Princeton

16 seeds: Stephen F. Austin, Florida Gulf Coast, Winthrop, Wagner, North Carolina A&T, Texas Southern

/**********************************************************************************/
Last 4 In: Missouri, Houston, Washington, Kansas State
First 4 Out: NC State, South Carolina, Georgia, LSU
/**********************************************************************************/

New to the bracket: Kansas State, Middle Tennessee, Winthrop


Removed from the bracket: Georgia, Western Kentucky, Radford


Sunday, January 21, 2018

1/21/18 Bracketology

Teams are ranked in order by seed. Auto bids are in bold and are determined via KenPom projections, with tiebreaker being awarded to the team ranked higher in KenPom’s rankings.

1 seeds: Villanova, Purdue, Kansas, Duke

2 seeds: Virginia, Clemson, North Carolina, Xavier

3 seeds: Auburn, West Virginia, Oklahoma, Tennessee

4 seeds: Arizona, Cincinnati, Texas Tech, Florida

5 seeds: Ohio State, Seton Hall, Arizona State, Michigan State

6 seeds: Wichita State, TCU, Kentucky, Rhode Island

7 seeds: Creighton, Texas A&M, Arkansas, Texas

8 seeds: Louisville, Gonzaga, Providence, Alabama

9 seeds: Miami FL, Butler, Florida State, Michigan

10 seeds: SMU, Missouri, Nevada, USC

11 seeds: St. Mary’s, Marquette, Syracuse, Washington, Georgia, Houston

12 seeds: Western Kentucky, New Mexico State, UL Lafayette, Murray State

13 seeds: East Tennessee State, Northern Kentucky, Buffalo, South Dakota State

14 seeds: Vermont, Montana, Loyola Chicago, UC Santa Barbara

15 seeds: Bucknell, Northeastern, Iona, Princeton

16 seeds: Stephen F. Austin, Florida Gulf Coast, Radford, Wagner, Texas Southern, North Carolina A&T

/**********************************************************************************/
Last 4 In: Syracuse, Washington, Georgia, Houston
First 4 Out: St. Bonaventure, Notre Dame, South Carolina, Baylor
/**********************************************************************************/

New to the bracket: Providence, USC, St. Mary’s, Georgia, Houston, Loyola Chicago, UC Santa Barbara, Northeastern, Wagner

Removed from the bracket: Notre Dame, UCLA, NC State, LSU, UCF, Missouri State, Towson, UC Davis, Robert Morris