Monday, March 13, 2017

Assessment of 2017 Bracket Projection

My favorite day of the year, Selection Sunday, has come and gone, and it’s time to evaluate my performance this year in projecting the committee’s official bracket. Below is how my performance this year stacks up to prior years, and here was my final projection.

Teams Selected Correctly (3 points per team)
Teams Seeded Correctly (3 points per team)
Teams Missed by 1 Seed Line (1 point per team)
Total Team Seeded Correctly/Within 1 Seed Line
Total Points
Bracket Matrix Finish
Bracket Matrix Percentile

*My entry was not included in the bracket matrix in 2014 to 2016 as this blog didn’t exist, but the last two rows show where my projection would have placed if it were included

I was able to get all 68 teams correctly selected for the first time ever, but it was an easy year in selecting the field, as 115 out of 174 members in the bracket matrix got 68 out of 68 teams correctly selected. This led to 17 entries this year breaking the previous record score of 351, with the top score this year being 367! Incredible year for the bracket matrix. Where this year was really challenging was in the middle of the bracket. Most folks in the top half of the matrix we able to accurately get 13 of the top 16 seeds seeded exactly to their seed (unless you listen to Jay Bilas and thought Duke deserved a one seed despite entering the ACC tournament losing three of four games and had a home court loss to NC State that apparently doesn’t count in his world, but I digress). From there, it was a hot mess as the committee threw some curveballs.

Overall, I am very pleased with my performance, as I was one of twelve people in the entire matrix to get either 67 or 66 teams seeded exactly correct or within one seed line of their actual seed. It’s very rewarding to see the hours and hours of hard work pay off with a successful showing.

Time to investigate where and why I missed on teams to improve for next year.

Teams Missed by Two Seed Lines

There were two teams I missed by two seed lines: Wisconsin and Xavier.

Wisconsin finished tied for second in the Big Ten and made a run to the Big Ten title game. They went 4-6 against the top 50 and 13-9 against the top 100 in the RPI. Those 13 top 100 wins typically guarantee at least a 7 seed; however, their non-conference strength of schedule ranked 301 out of 351 teams…that historically drops a team a seed line below where the public believes them to be, especially when said team doesn’t win a game against the best teams on that poor non-conference schedule (as was Wisconsin’s case).
I had Wisconsin as the last 6 seed and Maryland as the top 7 seed, and strongly considered swapping them in the final hour, but ultimately I decided that Maryland’s two losses to sub-100 RPI teams compared to Wisconsin’s 0 losses to that cohort would be the trump card; alas, the committee punished the Badgers for a very poor non-conference showing. If I had swapped those two I would have been rewarded with an additional 3 points (+2 for Maryland and +1 for Wisconsin).

Below is a comparison of the resumes of Wisconsin and the four teams rewarded with a 7 seed by the committee (St. Mary’s, Michigan, Dayton, and South Carolina). After a review of these five teams, I can understand why the committee gave Wisconsin an 8 seed.
Comparison of the Resumes of the 7 Seeds and Wisconsin
South Carolina
St. Mary's
RPI Ranking
SOS Rank (out of 351)
Non-Conf SOS Rank
Record vs Top 50
Record vs Top 100
Record vs Top 200
Avg Opponent RPI
Road Record

I think it’s clear that Michigan and St. Mary’s are deserving of their 7 seed. Michigan has the most top 50 wins of this cohort, thanks in part to their run in the Big Ten tournament, and they played the toughest schedule of any of these teams. St. Mary’s challenged themselves out of conference and beat Dayton and Nevada, two tournament teams, and three of their four losses are to Gonzaga. Then it gets interesting. Dayton, South Carolina, and Wisconsin all have red flags on their profiles. Dayton has 3 losses to teams outside the top 100, South Carolina has an RPI of 43, and Wisconsin has the second worst non-conference strength of schedule of any at-large team in the field (congratulations to Virginia Tech on that honor!). Based on how a lot of the other seeds were handed out (Vanderbilt, Michigan State, etc.) the committee really valued challenging yourself out of conference. As long as the committee is consistent in how they seed the teams, I have no problem with their method, and they were consistent in that regard.

The second team I missed by two seed lines was Xavier. Xavier has struggled since losing Edmond Sumner to injury for the season, winning just four games, with three of those four wins coming against a very bad DePaul team. That fourth win, however, came against Butler on a neutral court in the Big East Tournament. Knowing Butler was going to get a 4 seed, I figured that win, coupled with Xavier’s 15th ranked SOS and 0 losses to teams outside the top 100 RPI, was going to be enough for the committee to reward them with a 9 seed. This one stings, as the very last change I made to my seed list before making it final was swapping Seton Hall and Xavier. If I had not made that change, I would have been rewarded with an additional 3 points (+2 for Seton Hall and +1 for Xavier). Similar to the exercise above, if I compare Xavier to all of the 10 seeds, I am having a really tough time seeing how they are behind Marquette and Oklahoma State, but the reason must be the Sumner injury knocked them down a seed line.


There were a couple teams from mid major conferences this season that many were predicting would receive a seed better than 12. Those two teams were Wichita State and Middle Tennessee.

Starting with Wichita State, the Shockers blitzed through the Missouri Valley this season to a 17-1 record and then won the conference tournament with ease. KenPom and Sagarin rankings love Wichita State and have them ranked inside/near the top 10 in the nation.  The name “Wichita State” has become synonymous with tournament success. All those factors allowed many to predict Wichita State as an 8 seed (ESPN even had Wichita State as a projected 7 for a day after winning the MVC tournament). However, in years past Wichita State played a strong non-conference schedule and the Missouri Valley is typically good for three to four top 100 teams. This season the Shockers played the 215th toughest non-conference schedule and lost to every decent team that they played. Their three top 100 wins came against Illinois State (twice) and Colorado State, and their only win against a team in the field was South Dakota State, a 16 seed. The Missouri Valley was way down in the middle this season, so the opportunity to pick up quality wins just wasn’t there this season.

Middle Tennessee was the team that ruined many people’s brackets last season (myself included) when they pulled arguably the greatest upset in the history of the NCAA tournament in beating 2 seed Michigan State in the first round. Many pieces from that team returned this season, and the Blue Raiders went 29-4, winning the C-USA regular season and tournament title. They played the 19th toughest non-conference schedule and picked up two top 50 wins and four top 100 wins along the way. Those metrics are all as good as or better than when Middle Tennessee picked up an at large bid in 2013 and received an 11 seed, so it was only natural to expect them to receive an 11 seed or better this season. Alas, the committee gave them a 12 seed, which turned into a double whammy to most bracketologists as that meant that all four final at large teams would receive 11 seeds as opposed to two of them receiving 11 seeds and two receiving 12 seeds. I feel the top ranking bracketologists this season will be those that correctly assumed the committee would give Middle Tennessee a 12 seed, and thus would get a full 12 points on all four final at large teams and 3 points on Middle Tennessee due to accurate seeding, whereas the masses will get 9 points for getting just two of the four at large teams correct and missing by one seed line on the other two (would be projected as a 12) and missing Middle Tennessee by one seed line. That’s a 6 point swing right there alone.

The lesson is this: in the future, seed the mid major a seed line lower than you expect/when in doubt, give the top mid majors a 12 a move along. I’m not saying it’s fair, but remember we’re trying to predict what the committee will do.

Quick Thoughts

  • The decision to place Louisville as the final 2 seed is a head scratcher. Most people in the bracket matrix were giving that to either Oregon or Baylor. I went with Baylor thinking that their non-conference performance (went 12-0 and had the number one non-conference RPI with wins over Louisville, Oregon, Xavier) and the strength of the Big 12 would put them over Louisville. I also didn’t expect the committee to place 3 ACC schools into the top 8. Further, Baylor has 9 top 50 wins to Louisville’s 7, and 14 top 100 wins to Louisville’s 13. Both teams are top 5 in SOS. The only area Louisville has an edge over Baylor is in the losses: all 8 of Louisville’s losses came to top 40 RPI opponents, whereas Baylor lost a game to a team outside the top 100. That to me isn’t enough to offset all the good on Baylor’s resume compared to Louisville
  • Purdue as the final 4 seed shocked me. The Boilers are 1-5 against teams in the top 25 of the RPI, with the only win being against Notre Dame on a neutral court. I figured Notre Dame’s run in the ACC tournament and Purdue’s quarterfinal exit would be enough to push Notre Dame over Purdue, as the SOS, top 50 wins, and top 100 wins are all similar for the two teams. Maybe head-to-head result was the factor?
  • A lot of folks are probably shocked that a 15 loss team was rewarded with a 9 seed, but I was thisclose to placing Vanderbilt on the 9 seed line, but decided to keep them as the top 10 seed. Think about this for a second. Vanderbilt had the number one non-conference SOS and the number one overall SOS…given that they play in the SEC, which is probably the fourth best conference, that means that their non-conference SOS was head and shoulders above the field. This showed in their record against various cohorts, as they went 6-8 against the top 50 and 11-14 against the top 100. That’s right; Vanderbilt played 25 games against the top 100. The next closest teams are Kentucky, UNC, and Duke with 23. Looking back at recent years, the worst seed any team has ever received with at least 11 top 100 wins was Texas back in 2014, and they were given a 7 seed. That Texas team only had 10 losses, but had a non-conference SOS near 200, a far cry from number 1 out of 351.
  • If you thought Syracuse was going to get in, I’ll throw some data points at you: they went 12-13 against the top 200, and lost an additional game to a team outside the top 200. Their RPI was 84. Their non-conference SOS was 217. They scheduled poor in the non-conference and lost to many bad teams on said bad non-conference schedule. And, as most people have probably heard by now, they had just two wins outside of the Carrier Dome: a one point win at Clemson and a win at NC State.
  • I am very happy the committee rewarded Gonzaga with a one seed. They more than earned it with neutral court wins over Arizona, Florida, and Iowa State, plus a 3 game sweep of St. Mary’s. Besides, the team they lost at home to is not as bad as NC State ๐Ÿ˜ 

That wraps up the 2017 bracketology season. Thanks to all of you who made it this far and to those who followed me throughout the past couple of months. Here’s to another successful year in 2018!

No comments:

Post a Comment