The NET formula is not only a closely guarded secret but is impossible to reverse engineer. It also seems like it is sometimes impossible to make sense of some of the rankings it produces. But as long as we are talking about the top 40-50, we can take a stab at deconstructing how teams are ranked.
There are only two elements to NET
The Team Value Index is super-secret - we have no idea how it is calculated. The NCAA only says that it reflects a team's performance compared to how a NCAA Tournament caliber team would perform considering opponent and location. So we know is has some relation to schedule strength and win percentage . How TVI relates to a team's Quadrant record is unknown, but the Quadrant system is supposedly separate from NET ranking.
Adjusted Net Efficiency (ANE) is the most important part of NET according to the NCAA and we can estimate that part of the ranking.
Net Efficiency is simple: Offensive points-per-possession minus Defensive points-per-possession. Expressed as points per 100 possessions. How it is "Adjusted" is unknown.
But I suspect that Ken Pomeroy's methodology is not too far off. KenPom adjusts the raw ppp numbers by altering them to how the team would perform against the average D1 opponent -- that is, against a team with average OE and DE numbers. For example, your offensive ppp number is adjusted up against good teams with high DE and down against bad teams with low DE -- that is, "good" and 'bad" are defined as the opponent's OE and DE compared to the D1 average. He also adjusts for game location by adding or subtracting a small percentage (I think 1.4%) for home versus away. Whether he weights recent games higher I do not know. Schedule strength obviously affects the adjustments.
KenPom's ANE seems like a decent proxy for however the NCAA calculates it. Nine of the top 10 NET teams are the same as KenPom's top 10 and in pretty much the same order. (Gonzaga and Arizona are flipped for tenth place.) Of the top 25 NET teams 23 are also in the KenPom top 25 and the rank is within zero to three of one another in all but four cases. Both systems rank Houston (29.46) and Alabama (27.83) #1 and #2.
As we move down the list KenPom's adjusted net efficiency numbers begin to converge becoming much closer together. For instance he has Texas A&M (17.84) at #25 and Indiana (17.79) at #26 while NET has them at #23 and #20. As the ANE differences between teams becomes very small, the TVI contribution to the NET rank becomes more significant. Of the NET teams ranked 26 to 40 all but one are in the KenPom top 40 but 10 are ranked differently by more than three spots. Bottom line, efficiency is a huge factor in the NET rankings when we talk about Tournament-worthy teams.
The top six ACC teams are ranked in the same order on both lists. NC State is 15.07 and ranked fourth in the ACC, for comparison. NET rank is 37, Pomeroy rank is 43.
Takeaways:
There are only two elements to NET
- Adjusted Net Efficiency (ANE) -- the most important piece
- Team Value Index (TVI)
The Team Value Index is super-secret - we have no idea how it is calculated. The NCAA only says that it reflects a team's performance compared to how a NCAA Tournament caliber team would perform considering opponent and location. So we know is has some relation to schedule strength and win percentage . How TVI relates to a team's Quadrant record is unknown, but the Quadrant system is supposedly separate from NET ranking.
Adjusted Net Efficiency (ANE) is the most important part of NET according to the NCAA and we can estimate that part of the ranking.
Net Efficiency is simple: Offensive points-per-possession minus Defensive points-per-possession. Expressed as points per 100 possessions. How it is "Adjusted" is unknown.
But I suspect that Ken Pomeroy's methodology is not too far off. KenPom adjusts the raw ppp numbers by altering them to how the team would perform against the average D1 opponent -- that is, against a team with average OE and DE numbers. For example, your offensive ppp number is adjusted up against good teams with high DE and down against bad teams with low DE -- that is, "good" and 'bad" are defined as the opponent's OE and DE compared to the D1 average. He also adjusts for game location by adding or subtracting a small percentage (I think 1.4%) for home versus away. Whether he weights recent games higher I do not know. Schedule strength obviously affects the adjustments.
KenPom's ANE seems like a decent proxy for however the NCAA calculates it. Nine of the top 10 NET teams are the same as KenPom's top 10 and in pretty much the same order. (Gonzaga and Arizona are flipped for tenth place.) Of the top 25 NET teams 23 are also in the KenPom top 25 and the rank is within zero to three of one another in all but four cases. Both systems rank Houston (29.46) and Alabama (27.83) #1 and #2.
As we move down the list KenPom's adjusted net efficiency numbers begin to converge becoming much closer together. For instance he has Texas A&M (17.84) at #25 and Indiana (17.79) at #26 while NET has them at #23 and #20. As the ANE differences between teams becomes very small, the TVI contribution to the NET rank becomes more significant. Of the NET teams ranked 26 to 40 all but one are in the KenPom top 40 but 10 are ranked differently by more than three spots. Bottom line, efficiency is a huge factor in the NET rankings when we talk about Tournament-worthy teams.
The top six ACC teams are ranked in the same order on both lists. NC State is 15.07 and ranked fourth in the ACC, for comparison. NET rank is 37, Pomeroy rank is 43.
Takeaways:
- KenPom rankings are a decent proxy for NET for highly ranked teams, not so much when efficiency differences among teams become very small.
- ANE and TVI both account for Strength of Schedule in some way but it is not part of the NET formula.
- Win-loss record appears to be less important than efficiency.
- Teams are rewarded for running up the score. (If you play bad teams you can mitigate the weak scheduling by scoring a lot.)
