NET system adds to data in play for Selection Sunday
Chicago — As conference tournaments have taken center stage the past week, March Madness is heating up around college basketball.
Most of the mid-majors have crowned a conference champion and handed out automatic bids to the NCAA Tournament while the power conferences will be in the spotlight through the weekend.
Will there be wild finishes, upsets and crushing defeats? Sure, that’s why they call it madness.
But it’s all leading up to the real climax of championship week — Selection Sunday.
When the NCAA Tournament committee reveals this season’s 68-team field at 6 p.m. Sunday, that’s when things really get interesting, mostly because the evening gets dominated by complaints about every key aspect of the field — who’s in, who’s out, who’s seeded what, who’s headed where.
It’s a tradition as storied as the tournament itself, and now that there’s a new buzzword — the NET — don’t expect those gripes to disappear.
But what exactly is the NET, how is it being used and how will it affect how the committee puts together the field?
In the simplest terms, the NCAA Evaluation Tool is used to sort the 353 teams around the country and give the committee an idea of where they all rank. This season, it replaced the RPI as the main sorting tool. The RPI was first used in 1981 and starting last season, it helped put teams into a quadrant system in order to try to build a team’s resume. The NET is doing the same thing this season.
Why the move from the RPI to the NET? Because there’s far more data available now to make judgments on how good a team might be, said Kevin Pauga, creator of the KPI — a results-based metric used to rank teams — as well as an assistant athletic director at Michigan State.
“The RPI was 25 percent a team’s winning percentage, 50 percent your opponent and 25 percent you opponents’ opponents,” Pauga explained. “That was built on winning as many games as possible and playing teams that won as many games as possible. We have more advanced tools now. We have better data where we don’t need to use such simple data points. That was built almost 40 years ago in days where schools were still faxing scores and calculating it the next day.
“Using the RPI as a sorting tool means that you’re sorting the quality of your wins based on the opponent's resume not on the quality of the opponent. There’s a fine line between sorting your resume based on the quality of opponent vs. the quality of the opponent’s resume. You’re gonna have teams that their resume is a lot better than the quality of their team and vice versa. If I’m gonna judge how good of a win it is, you want a deeper dive than just a win-loss record. The NET as a sorting tool provides that.”
Along with the NET, members of the committee are provided team sheets, updates daily, for every team in the country. They’re sorted by the NET but also include team rankings in five other key metrics — KPI and SOR, strength of record from ESPN, as well as Kenpom.com, Sagarin and BPI, also from ESPN.
Both KPI and SOR are results-based metrics while Kenpom, Sagarin and BPI are predictive metrics. All are included to give the committee as much information as possible.
“If you’ve got multiple metrics you’re looking at, it allows you to identify and minimize the impact of outliers,” Pauga said. “If you just have RPI on the sheet, what if RPI is the outlier? Then you’re driving a conversation based on an outlier. Where if you’ve got multiple metrics, you can very easily see where those outliers are and what the cause of it is.”
The difference in the metrics is also important, Pauga said. The predictive metrics include things like offensive and defensive efficiency in an effort to determine how good a team is and how they might perform in the future. The results-based metrics lean more toward who a team has beaten and how good that team is.
For example, Gonzaga is ranked No. 2 by Kenpom and No. 11 by KPI.
“The NET has leaned more predictive in nature,” Pauga said. “That second component which is based on raw efficiency means it’s measuring the quality of teams significantly more heavily. As a result, when utilizing it to put teams into quadrants, sorting by the quality of the team instead of the quality of the team’s resume is critical. There’s a fine line between the two.”
Which is all to say this — relax, fans. The committee will not simply follow the NET and drop teams in the bracket. It didn’t do so with the RPI — teams ranked as high as 21 in the RPI have been left out of the tournament field — and the added metrics only add to the data being considered, not to mention the gold old eye test.
“(The NET) has been able to allow the committee members to look at something more contemporary, which is what we wanted when we initially set out,” said Bernard Muir, athletic director at Stanford and chair of the men’s basketball committee. “Each and every game as we go along throughout the season, the NET is something I think has become more reliable for the committee. Each committee member will use this sorting tool as well as other tools we have at our disposal in their own way. That discussion will happen … over the next several days.”
It’s important, too, Pauga said, to understand that individual components like strength of schedule can be used by the committee when splitting hairs between two teams but are already worked into most of the metrics.
“The strength of schedule is an explanation, not a metric, no different than efficiency is an explanation,” Pauga said. “Be careful not to cherry pick one part of a resume. … Let the multiple data points drive your decision making as opposed to taking your perception and finding one or two pieces of data that support it, because there could be four or five other points of contention that could counter that argument.”
So, it all comes down to the committee room and how they choose to apply all the data provided. Essentially, it’s the same as it’s always been just with some more information at hand.
“People are naturally scared and unsure of what they don’t know and understand, and rightfully so,” Pauga said. “But overall, I don’t know that the overall process changes all that much, if at all. It’s there is now a better way to sort teams into quadrants and quantify the quality of each win and loss.”
Nothing but NET
Here are the top 25 teams in the NCAA NET rankings, through Wednesday’s games.
1. Virginia, 28-2
2. Gonzaga, 30-3
3. Duke, 26-5
4. Houston, 29-2
5. Kentucky, 26-5
6. Tennessee, 27-4
7. North Carolina, 26-5
8. Michigan State, 25-6
9. Texas Tech, 26-5
10. Michigan, 26-5
11. Virginia Tech, 24-7
12. Purdue, 23-8
13. Wofford, 29-4
14. LSU, 26-5
15. Wisconsin, 22-9
16. Buffalo, 28-3
17. Auburn, 22-9
18. Nevada, 28-3
19. Florida State, 25-6
20. Kansas, 23-8
21. Louisville, 20-12
22. Mississippi State, 22-9
23. Iowa State, 20-11
24. Kansas State, 24-7
25. Villanova, 22-9