This is the annual "What is NET and how do you affect it" thread. It should also explain why NET rankings are sometimes odd early in the season.
Since 2020 the NET consists of two metrics (NE and TVI) that are combined in some secret way to rank teams. Although the algorithms are secret, we can make some good guesses about what makes for a good score. I'll try to explain without being too wonky and maybe clarify some misconceptions that I see on the forum.
First, understand what is not part of the NET: When a game is played doesn't matter -- November games count the same as February games and injuries are ignored. Surprisingly, won-loss record is also not part of the NET. It is possible to have a good NET rank with a mediocre W-L record. And Intangibles like injuries or streaks, and metrics like W-L records and winning margins are left up to the Selection Committee to consider -- they are not inputs to the NET.
Part 1: Net Efficiency (NE)
NE is all about tempo-free scoring offense and defense -- specificzally points scored per possession (Offensive Efficiency) minus points given up per possession (Defensive Efficiency) usually presented as per 100 possessions. The numbers are adjusted to reflect the opponent's OE and DE averages as of the date of the game. KenPom ranks teams by this method.
It is helpful to think of Net Efficiency as the margin by which a team would be expected to beat (or lose to) the average D1 team (not including a few points added for home court advantage). It's a fairly accurate game predictor -- it predicts the winner of the 9,000 or so D1 games almost exactly 75% of the time, which is a bit better than anhy of the other predictive metrics. Since the methodology already exists, I assume the NCAA is using something very close to KenPom's Adjusted Net Efficiency formula. A minor adjustment might be needed because team efficiencies improve as the season progresses, so that normalizing to the D1 average adds a recency element that the NCAA may ignore.
Gaming Net Efficiency is easy. Just run up the score. What's important is understanding that OE and DE are being compared to all of the other team's opponents. You want to win by more than the rest of the other team's opponents in order to maximize adjusted efficiency. So clearing the bench with a big lead is not a good strategy for maximizing NE, but it doesn't have a huge effect if limited to a handful of games. Net Efficiency is more about winning games decisively against good opponents and totally blowing out the bad ones.
You can see where we're going here. You need to "win better" than your opponent's opponents. The risk to playing mediocre or bad teams is that other schools will beat them just as decisively as you do. If you had a crystal ball you would schedule the best teams in bad conferences (to limit their losses) and never schedule games with average or below-average teams in bad conferences because it matters to Net Efficiency how your opponents do in their other games.
Part 2: Team Value Index (TVI)
This is the other part of NET. The TVI formula is top secret but the NCAA describes it as “based on game results” considering only the opponent, the location and the winner as inputs. My guess is that it is some sort of recursive formula based on the opponent's NET and a location-based weighting that focuses on winning rather than scoring. Perhaps initial NET ranks are somehow assigned and then adjusted as games are played -- which would explain why NET is not published until teams have played 8 or 9 games. That would also explain why NET rankings can seem so odd mid-season and settle down as more games are played. I also suspect that the formula might be biased in some way toward playing good teams -- for instance, the credit for losing to a good team might be greater than for beating a mediocre one, or the formula may give no credit at all for Quad 4 wins. (I should point out here that the Quads reflect NET ranks, not the other way around.) Beating the #5 team at home or the #75 team away both count as Quad 1 wins, but surely have significantly different effects on TVI.
So how do you game the TVI part of the formula?
Schedule the best OOC teams that you can beat. Winning a Quadrant 2 game is better than winning a Quadrant 3 game and losing to a Quadrant 1 team is not as bad as losing to a Quadrant 2 team. I think, in general, the better your opponents are, the better your TVI rank. And don’t schedule any Quadrant 4 games. (I have to think the NCAA likes the idea that chasing NET forces teams into more competitive matchups to drive TV viewership.)
You should also play some beatable teams on their home court or on a neutral court. Playing a lot of home games probably is not as helpful to NET as winning away games against the same opponents. If you schedule a team ranked 125, for instance, it’s either a Quad 2 win or a Quad 3 win depending on where you play. Scheduling Quad 3 and 4 teams, or playing bad teams on your home court, might be helpful to team development but not helpful to TVI. Participating in neutral court tournaments is a positive.
I hope that quick explanation is useful.
Since 2020 the NET consists of two metrics (NE and TVI) that are combined in some secret way to rank teams. Although the algorithms are secret, we can make some good guesses about what makes for a good score. I'll try to explain without being too wonky and maybe clarify some misconceptions that I see on the forum.
First, understand what is not part of the NET: When a game is played doesn't matter -- November games count the same as February games and injuries are ignored. Surprisingly, won-loss record is also not part of the NET. It is possible to have a good NET rank with a mediocre W-L record. And Intangibles like injuries or streaks, and metrics like W-L records and winning margins are left up to the Selection Committee to consider -- they are not inputs to the NET.
Part 1: Net Efficiency (NE)
NE is all about tempo-free scoring offense and defense -- specificzally points scored per possession (Offensive Efficiency) minus points given up per possession (Defensive Efficiency) usually presented as per 100 possessions. The numbers are adjusted to reflect the opponent's OE and DE averages as of the date of the game. KenPom ranks teams by this method.
It is helpful to think of Net Efficiency as the margin by which a team would be expected to beat (or lose to) the average D1 team (not including a few points added for home court advantage). It's a fairly accurate game predictor -- it predicts the winner of the 9,000 or so D1 games almost exactly 75% of the time, which is a bit better than anhy of the other predictive metrics. Since the methodology already exists, I assume the NCAA is using something very close to KenPom's Adjusted Net Efficiency formula. A minor adjustment might be needed because team efficiencies improve as the season progresses, so that normalizing to the D1 average adds a recency element that the NCAA may ignore.
Gaming Net Efficiency is easy. Just run up the score. What's important is understanding that OE and DE are being compared to all of the other team's opponents. You want to win by more than the rest of the other team's opponents in order to maximize adjusted efficiency. So clearing the bench with a big lead is not a good strategy for maximizing NE, but it doesn't have a huge effect if limited to a handful of games. Net Efficiency is more about winning games decisively against good opponents and totally blowing out the bad ones.
You can see where we're going here. You need to "win better" than your opponent's opponents. The risk to playing mediocre or bad teams is that other schools will beat them just as decisively as you do. If you had a crystal ball you would schedule the best teams in bad conferences (to limit their losses) and never schedule games with average or below-average teams in bad conferences because it matters to Net Efficiency how your opponents do in their other games.
Part 2: Team Value Index (TVI)
This is the other part of NET. The TVI formula is top secret but the NCAA describes it as “based on game results” considering only the opponent, the location and the winner as inputs. My guess is that it is some sort of recursive formula based on the opponent's NET and a location-based weighting that focuses on winning rather than scoring. Perhaps initial NET ranks are somehow assigned and then adjusted as games are played -- which would explain why NET is not published until teams have played 8 or 9 games. That would also explain why NET rankings can seem so odd mid-season and settle down as more games are played. I also suspect that the formula might be biased in some way toward playing good teams -- for instance, the credit for losing to a good team might be greater than for beating a mediocre one, or the formula may give no credit at all for Quad 4 wins. (I should point out here that the Quads reflect NET ranks, not the other way around.) Beating the #5 team at home or the #75 team away both count as Quad 1 wins, but surely have significantly different effects on TVI.
So how do you game the TVI part of the formula?
Schedule the best OOC teams that you can beat. Winning a Quadrant 2 game is better than winning a Quadrant 3 game and losing to a Quadrant 1 team is not as bad as losing to a Quadrant 2 team. I think, in general, the better your opponents are, the better your TVI rank. And don’t schedule any Quadrant 4 games. (I have to think the NCAA likes the idea that chasing NET forces teams into more competitive matchups to drive TV viewership.)
You should also play some beatable teams on their home court or on a neutral court. Playing a lot of home games probably is not as helpful to NET as winning away games against the same opponents. If you schedule a team ranked 125, for instance, it’s either a Quad 2 win or a Quad 3 win depending on where you play. Scheduling Quad 3 and 4 teams, or playing bad teams on your home court, might be helpful to team development but not helpful to TVI. Participating in neutral court tournaments is a positive.
I hope that quick explanation is useful.
