I never expected that one day I’d have North Carolina State’s back in a discussion out teams getting hosed on NCAA Tournament selection, but when you choose a life of data analysis, anything is possible.
It was suggested that the reason North Carolina State missed the 2019 NCAA Tournament was due to their non-conference slate. I decided to look at the eight teams that either missed the tournament with a positive WAB (per the KenPoms) or made the tournament as an at-large selection with a negative WAB.
| School |
Seed |
EW% |
Win % |
Result |
WAB |
Pythag |
| UNC Greensboro |
0 |
76.34% |
82.35% |
6.01% |
2.04 |
0.7245 |
| North Carolina St. |
0 |
61.13% |
66.67% |
5.54% |
1.83 |
0.8626 |
| TCU |
0 |
59.46% |
60.61% |
1.15% |
0.38 |
0.8293 |
| Clemson |
0 |
58.94% |
59.38% |
0.44% |
0.14 |
0.8881 |
| Florida |
10 |
56.06% |
55.88% |
-0.18% |
-0.06 |
0.8898 |
| Seton Hall |
10 |
61.12% |
60.61% |
-0.51% |
-0.17 |
0.8088 |
| Arizona St. |
11.1 |
69.80% |
68.75% |
-1.05% |
-0.34 |
0.7749 |
| St. John’s |
11.1 |
66.73% |
63.64% |
-3.09% |
-1.02 |
0.7219 |
UNC Greensboro got royally hosed, but it’s worth noting that North Carolina State had an almost equally good claim to selection. It’s certainly the case that North Carolina State had a stronger argument than either St. John’s or Arizona State in terms of resume, and in terms of team quality. North Carolina State would have been the clear favorite over either on a neutral court, strongly so in the case of a matchup against St. John’s. Let’s focus specifically on the non-conference slate.
| School |
Seed |
NC EW% |
NC W% |
Result |
NC WAB |
Pythag |
| UNC Greensboro |
0 |
76.97% |
81.82% |
4.85% |
0.53 |
0.7245 |
| North Carolina St. |
0 |
82.53% |
92.31% |
9.78% |
1.27 |
0.8626 |
| TCU |
0 |
81.04% |
92.31% |
11.27% |
1.47 |
0.8293 |
| Clemson |
0 |
75.61% |
76.92% |
1.31% |
0.17 |
0.8881 |
| Florida |
10 |
67.20% |
61.54% |
-5.66% |
-0.74 |
0.8898 |
| Seton Hall |
10 |
67.03% |
75.00% |
7.97% |
0.96 |
0.8088 |
| Arizona St. |
11.1 |
72.51% |
75.00% |
2.49% |
0.3 |
0.7749 |
| St. John’s |
11.1 |
79.65% |
92.31% |
12.66% |
1.65 |
0.7219 |
An average bubble team would expect to win 82.53% of their games against North Carolina State’s non-conference schedule. That’s the softest slate of the eight teams listed here (and 299th overall if you’re curious). TCU and St. John’s also had fairly easy non-conference schedules, followed by UNC Greensoro and Clemson. Seton Hall and Florida had very tough non-conference schedules. In Florida’s case, the result was poor as they went 8-5 while a bubble team would expect to go around 8.74-4.26. Not a great result. As for North Carolina State, they went 12-1 while a bubble team would have expected to go 10.73-2.27. Considering the difference in opposition, North Carolina State should have entered conference play roughly two wins ahead of Florida. A much easier schedule, yes, but not four wins easier. Also, please note that St. John’s crushed a soft schedule. It was their much tougher league schedule that tanked their WAB.
Let’s look back at the full season results:
| School |
Seed |
EW% |
Win % |
Result |
WAB |
Pythag |
| UNC Greensboro |
0 |
76.34% |
82.35% |
6.01% |
2.04 |
0.7245 |
| North Carolina St. |
0 |
61.13% |
66.67% |
5.54% |
1.83 |
0.8626 |
| TCU |
0 |
59.46% |
60.61% |
1.15% |
0.38 |
0.8293 |
| Clemson |
0 |
58.94% |
59.38% |
0.44% |
0.14 |
0.8881 |
| Florida |
10 |
56.06% |
55.88% |
-0.18% |
-0.06 |
0.8898 |
| Seton Hall |
10 |
61.12% |
60.61% |
-0.51% |
-0.17 |
0.8088 |
| Arizona St. |
11.1 |
69.80% |
68.75% |
-1.05% |
-0.34 |
0.7749 |
| St. John’s |
11.1 |
66.73% |
63.64% |
-3.09% |
-1.02 |
0.7219 |
UNC Greensboro had by far the easiest schedule overall. A bubble team would expect to win more than three-fourth’s of their games against it. UNC Greensboro bettered than, winning 82.35%. Did that matter to the committee? Not enough.
North Carolina State ended up facing a much tougher schedule than Arizona State or St. John’s. The ACC was brutal last season. They ended up winning two-thirds of their games when par for a bubble team would have been 61.13%. In short, their record was superb given who they faced and where they faced them.
Florida’s schedule was even harder, but they performed basically at a normal expectation for a bubble team. The same can be said for Clemson and TCU, although I’ll note that overall TCU’s results should be considered materally better than Florida’s last season.
Arizona State’s schedule was significantly weaker North Carolina State’s. More importantly, they perform particularly well against it.
St. John’s prayers were answered on selection Sunday, but they shouldn’t have been. They had middling results given the competition they faced. They were one full win below the Mendoza line.
I would not recommend selecting teams to the NCAA Tournament based on their power rankings, be they KenPom or otherwise, but if that was a consideration it’s not one that was favorable to St. John’s. (I should note that seeding is a different matter. Taking power rankings into account is helpful for balancing the bracket.)
I am left to conclude that the Wolfpack was left out of the dance due to who they faced in their non-conference games. That’s absurd. As Bart Torvik noted:
The problem for rewarding or punishing a team purely for SOS is that you are now rewarding losses to good teams and punishing wins against weak teams. That was also the cardinal sin of the RPI. Defeating a team should never be considered a negative. Losing to one should never be considered a positive. Considering SOS as a metric unto itself breaks that. That’s unconscionable, and as Bart noted, braindead.
SOS matters. WAB (and Parcells) takes that fully into account. Your strength of record, AKA, body of work should determine selection to the field. SOS should never be considered in and of itself to reward or punish teams.
Two notes:
- In terms of resume a loss is never good and a win is never bad. That doesn’t apply to power ratings, where playing almost even with Kansas will generally boost your rating. Power ratings and resume ratings are completely separate things.
- Treating SOS as a metric unto itself isn’t just awful, it’s biased. If it were to be codified as part of the selection process, power conference teams would have the power to dominate it and exclude mid-majors with less scheduling power from the process.
For more details:
https://www.heraldonline.com/sports/college/big-south/winthrop-university/article222906070.html
Key quotes:
(ETSU Head Coach Steve) Forbes said that almost no team from those seven leagues (P5 + Big East & AAC) will play non-conference road games against low/mid-majors. That’s hampering the other 25 leagues’ already slim chances of getting multiple teams into the NCAA Tournament, because their chances at quality wins is lowered.
And the NCAA’s new metric, called NET and developed to replace the RPI, isn’t helping low/mid-majors’ cause much either. It gives more weight to neutral court wins than road wins. There is almost no incentive for bigger schools to travel to smaller ones.
“We’re getting scheduled out of the tournament, because we’re not gonna have enough Quadrant 1 wins to even have a chance,” said Forbes.
The data bears this out:
Power 5 teams played 88 percent of their non-conference games in 2017-18 at home or on a neutral court.
So in short, using SOS is effectively a cudgel against schools like UNC Greensboro, East Tennessee State, Northern Iowa, and their ilk. It’s unfair and should be repudiated by the committee.
Remember, SOS matters. Stength of record takes it fully into account.
sethburn 