NET Ranking for Tournament Selection

[nelsonmuntz]

The five factors behind the NCAA's NET ranking system

Get ready to hear a lot about team value index, net efficiency, winning percentage, adjusted win percentage and scoring margin.

www.si.com

I can't find a thread discussing this last year, probably because it didn't matter for UConn. After digging in a little, I think I hate this rating, and think the NCAA should have stuck with the RPI, which was a lot simpler and more clear cut for coaches and teams.

The 5 components (no indication of how these are weighted):

Team Value Index - An "algorithm" that rewards teams for beating other good teams. Didn't the RPI already do that? This is a full blown "black box", which means we have no idea what the algorithm is doing. It could be adding +10 for just being in the ACC for all we know.

Net Efficiency - Points per Possession Offense minus Points per Possession Defense. This component flat out rewards teams for running up the score. I don't think a lot of coaches realize this.

Winning Percentage - straight forward, except that there is a separate Adjusted Win Percentage factor that incorporates road and neutral. So why have the unadjusted? This component appears to reward teams for scheduling cupcakes.

Adjusted Win Percentage - Why do we need separate Winning Percentage and Adjusted Win Percentage components? Shouldn't it be one or the other?

Scoring Margin - This is capped at 10 points.

I see why the NCAA did this. This is a full blown protection tool for the P6 leagues for bids, and for the blue blood programs for seeding. This is so much worse than the RPI.

Please correct me if I am wrong.

 

[Slasher]

You are wrong. The NET in the end is very similar to KenPom and BPI which focus on how good a team is (offensive and defensive efficiency) as opposed to simply who they played. RPI 3/4 of the formula is who you played and not whether you won or lost. RPI allowed you to game the system simply by not playing any 200+ teams your RPI was automatically good -- whether you were terrible or not. The NET is a combination of predictive measures (like KenPom) and results driven.

 

[BLUEDOGHOUSE]

The five factors behind the NCAA's NET ranking system

Get ready to hear a lot about team value index, net efficiency, winning percentage, adjusted win percentage and scoring margin.

www.si.com

I can't find a thread discussing this last year, probably because it didn't matter for UConn. After digging in a little, I think I hate this rating, and think the NCAA should have stuck with the RPI, which was a lot simpler and more clear cut for coaches and teams.

The 5 components (no indication of how these are weighted):

Team Value Index - An "algorithm" that rewards teams for beating other good teams. Didn't the RPI already do that? This is a full blown "black box", which means we have no idea what the algorithm is doing. It could be adding +10 for just being in the ACC for all we know.

Net Efficiency - Points per Possession Offense minus Points per Possession Defense. This component flat out rewards teams for running up the score. I don't think a lot of coaches realize this.

Winning Percentage - straight forward, except that there is a separate Adjusted Win Percentage factor that incorporates road and neutral. So why have the unadjusted? This component appears to reward teams for scheduling cupcakes.

Adjusted Win Percentage - Why do we need separate Winning Percentage and Adjusted Win Percentage components? Shouldn't it be one or the other?

Scoring Margin - This is capped at 10 points.

I see why the NCAA did this. This is a full blown protection tool for the P6 leagues for bids, and for the blue blood programs for seeding. This is so much worse than the RPI.

Please correct me if I am wrong.

Now we know why LIBERTY is so high. Yes they are good BUT...

 

[Dream Jobbed 2.0]

Did we end up winning a road game last year?

 

[HooperScooper]

Did we end up winning a road game last year?

Just ECU. Does that even count?

 

[nelsonmuntz]

You are wrong. The NET in the end is very similar to KenPom and BPI which focus on how good a team is (offensive and defensive efficiency) as opposed to simply who they played. RPI 3/4 of the formula is who you played and not whether you won or lost. RPI allowed you to game the system simply by not playing any 200+ teams your RPI was automatically good -- whether you were terrible or not. The NET is a combination of predictive measures (like KenPom) and results driven.

So instead, NET rewards teams for lighting up cupcakes at home. The efficiency numbers don't care if you play UHa or Kansas.

 

[Leebo]

So instead, NET rewards teams for lighting up cupcakes at home. The efficiency numbers don't care if you play UHa or Kansas.

Kenpom adjusts efficiency (not sure if this does). Basically, 100 points per 100 possessions against Virginia would be better than 100 points per 100 possessions against Maine. I don't know if this has such a feature, but that's how you would avoid that.

 

[auror]

So instead, NET rewards teams for lighting up cupcakes at home. The efficiency numbers don't care if you play UHa or Kansas.

Good thing all the midmajors end up with easier schedules due to the conferences they play in then. Easy for them to manipulate the unadjusted raw winning factors and scoring margin.

There were 0 teams from outside the major 7 conferences in the top 70 of KenPom's strength of schedule last year.

 

[Slasher]

NET is adjusted based on the opponent just like KenPom.

 

[nelsonmuntz]

NET is adjusted based on the opponent just like KenPom.

Does this go for the efficiency component alone?

 

[Slasher]

Does this go for the efficiency component alone?

Not sure on that.

 

[UConnSean]

The five factors behind the NCAA's NET ranking system

Get ready to hear a lot about team value index, net efficiency, winning percentage, adjusted win percentage and scoring margin.

www.si.com

I can't find a thread discussing this last year, probably because it didn't matter for UConn. After digging in a little, I think I hate this rating, and think the NCAA should have stuck with the RPI, which was a lot simpler and more clear cut for coaches and teams.

The 5 components (no indication of how these are weighted):

Team Value Index - An "algorithm" that rewards teams for beating other good teams. Didn't the RPI already do that? This is a full blown "black box", which means we have no idea what the algorithm is doing. It could be adding +10 for just being in the ACC for all we know.

Net Efficiency - Points per Possession Offense minus Points per Possession Defense. This component flat out rewards teams for running up the score. I don't think a lot of coaches realize this.

Winning Percentage - straight forward, except that there is a separate Adjusted Win Percentage factor that incorporates road and neutral. So why have the unadjusted? This component appears to reward teams for scheduling cupcakes.

Adjusted Win Percentage - Why do we need separate Winning Percentage and Adjusted Win Percentage components? Shouldn't it be one or the other?

Scoring Margin - This is capped at 10 points.

I see why the NCAA did this. This is a full blown protection tool for the P6 leagues for bids, and for the blue blood programs for seeding. This is so much worse than the RPI.

Please correct me if I am wrong.

Isn't the RPI calculation (Win% x 0.25) + (Opponents' Win% x 0.50) + (Opponents' Opponents Win% x 0.25)? So it's basically a lousy SOS metric with 75% of the weight not even considering how you play/the results of your games? That's laughably terrible

 

[pepband99]

Isn't the RPI calculation (Win% x 0.25) + (Opponents' Win% x 0.50) + (Opponents' Opponents Win% x 0.25)? So it's basically a lousy SOS metric with 75% of the weight not even considering how you play/the results of your games? That's laughably terrible

The RPI had one purpose, really - punish teams like JTJr's Georgetown teams, with laughably awful OOC schedules. It worked swimmingly.

The one thing i miss about the RPI is the predictability of it - you could guess moves up and down, based on game results. @nelsonmuntz is right about the black box on the biggest part of the metric. Why the NCAA doesn't publish this algorithm mystifies me, unless they truly have something to hide...

 

[pepband99]

The five factors behind the NCAA's NET ranking system

Get ready to hear a lot about team value index, net efficiency, winning percentage, adjusted win percentage and scoring margin.

www.si.com

I can't find a thread discussing this last year, probably because it didn't matter for UConn. After digging in a little, I think I hate this rating, and think the NCAA should have stuck with the RPI, which was a lot simpler and more clear cut for coaches and teams.

The 5 components (no indication of how these are weighted):

Team Value Index - An "algorithm" that rewards teams for beating other good teams. Didn't the RPI already do that? This is a full blown "black box", which means we have no idea what the algorithm is doing. It could be adding +10 for just being in the ACC for all we know.

Net Efficiency - Points per Possession Offense minus Points per Possession Defense. This component flat out rewards teams for running up the score. I don't think a lot of coaches realize this.

Winning Percentage - straight forward, except that there is a separate Adjusted Win Percentage factor that incorporates road and neutral. So why have the unadjusted? This component appears to reward teams for scheduling cupcakes.

Adjusted Win Percentage - Why do we need separate Winning Percentage and Adjusted Win Percentage components? Shouldn't it be one or the other?

Scoring Margin - This is capped at 10 points.

I see why the NCAA did this. This is a full blown protection tool for the P6 leagues for bids, and for the blue blood programs for seeding. This is so much worse than the RPI.

Please correct me if I am wrong.

In order:

Team Value Index - agreed "black box" aspect sucks.

Net efficiency - i think you're oversimplifying this. A classic "run up the score" scenario will impact the quotient of both those metrics too much, to really be meaningful.

Winning % and Adj Winning % - I like having both, because it minimizes the mid major gaming of the system that the RPI got before. You can, and should, get a bonus for taking on a good team, especially on the road, but it shouldn't be too black and white - you need to win. Yes it's sort of redundant with the algorithm junk, but since they won't publish it, i'll take a clear metric.

 

[nelsonmuntz]

Good thing all the midmajors end up with easier schedules due to the conferences they play in then. Easy for them to manipulate the unadjusted raw winning factors and scoring margin.

There were 0 teams from outside the major 7 conferences in the top 70 of KenPom's strength of schedule last year.

What does KenPom have to do with NET?

The SOS "gaming" under RPI was actually just teams playing better opponents, which was what the RPI was supposed to do. Toledo is a lot better than UHa. That is not conceptual.

The RPI had one purpose, really - punish teams like JTJr's Georgetown teams, with laughably awful OOC schedules. It worked swimmingly.

The one thing i miss about the RPI is the predictability of it - you could guess moves up and down, based on game results. @nelsonmuntz is right about the black box on the biggest part of the metric. Why the NCAA doesn't publish this algorithm mystifies me, unless they truly have something to hide...

We agree on something!!!

There is no good reason for the NCAA to withhold the formula for any part of the NET. It reeks of unfairness and bias.

 

[auror]

What does KenPom have to do with NET?

Use whatever accurate strength of schedule you want. Your point was that mid majors are getting punished by the system because power teams beat up on cupcakes. But on the whole due to conference schedules (especially expanded to 18 games), major conference teams always have harder schedules than basically every midmajor team.

So if you are upset that the NET does not take strength of schedule enough into account for the margin of victory metric while thinking it would benefit the major teams, it actually benefits the midmajors.

The larger point is that the NET is not a conspiracy against midmajors. The Quadrant system on the other hand...

 

[nelsonmuntz]

Use whatever accurate strength of schedule you want. Your point was that mid majors are getting punished by the system because power teams beat up on cupcakes. But on the whole due to conference schedules (especially expanded to 18 games), major conference teams always have harder schedules than basically every midmajor team.

So if you are upset that the NET does not take strength of schedule enough into account for the margin of victory metric while thinking it would benefit the major teams, it actually benefits the midmajors.

The larger point is that the NET is not a conspiracy against midmajors. The Quadrant system on the other hand...

So you know what is in the Black Box part of NET? Please share it with the rest of us.

 

[auror]

So you know what is in the Black Box part of NET? Please share it with the rest of us.

No, but you asserted you knew the intent of the system without knowing it either because of a certain reasoning. I showed how that reasoning was flawed.

In addition, the actual results from last year were close enough to all the other neutral advanced rating systems to assert that if that was their intent, they screwed it up.

 

[Purple Stein]

The RPI had one purpose, really - punish teams like JTJr's Georgetown teams, with laughably awful OOC schedules. It worked swimmingly.

Exactly! People love to dismiss/hate on the RPI, but it was a project of its time.

Ken Pom, etc., are in part responses to the limitations of RPI. 20 years from now, if there's still an NCAA, we'll have moved onto something new.

 

[nelsonmuntz]

Since the NCAA is not using Ken Pom, we need to leave it out of this discussion.

www.warrennolan.com tracks RPI and NET rating. I took a look at the relative conference strength of the Big 12 and Big East.

Big 12's conference strength assessment:

Overall winning percentage: 79-26, .7524
RPI and ranking:: .6000, #1 conference
NET Ranking: #1
Record against other P6 conferences: 16-22
Top 50 teams in NET: Kansas (1), Baylor (7), West Virginia (11), Texas Tech (35), Oklahoma State (38), Texas (42), Oklahoma (50)

Big East's conference strength assessment:

Overall winning percentage: 94-25, .7899
RPI: .5990, #2 conference
Net Ranking: #3
Record against other P6 conferences: 29-17
Top 50 teams in NET: Butler (5), Villanova (25), Georgetown (33), Creighton (37), Seton Hall (41), Depaul (44)

Why is a team like Depaul (12-1), with 5 wins over major conference opponents, ranked much lower than WVU (10-1), which has 1 win over a major conference opponent? Mid-major opponents, and margin over victory. Wichita State and Northern Iowa are good teams and solid wins for WVU, but the rating seems to be heavily influenced by wins over mid or low majors with good NETs (Akron, Northern Colorado, Austin Peay) and lack of really bad opponents (Youngstown State is worst NET opponent at 241). Depaul has played some really bad opponents (Farleigh Dickinson, Alcorn State, Cornell, UIC, Cleveland State). Depaul also only has 1 win over 20 points, while WVU has dropped the hammer 3 times.

It appears that NET has the same issue RPI had with significant rewards for playing mediocre opponents over bad ones, but has added encouraging teams to run up the score into the mix.

I can't explain Texas' 42 NET rating at all. They are winning close games over bad teams. How are they even Top 75?

I get that the NET is not designed to compare conferences, but how can any rating service have the Big East at #3 conference overall, while the Big 12 is #1? RPI has them essentially tied.

 

[HooperScooper]

Since the NCAA is not using Ken Pom, we need to leave it out of this discussion.

I thought they do use it to some extent? My head hurts reading the below linked article but they use multiple metrics for the NCAA Tournament selection, not just solely the NET.

Another change made in 2017 to the team sheets was the inclusion of other metrics. These include the Kevin Pauga Index and ESPN’s results-oriented metric, the Strength of Record. The team sheets also included three predictive metrics: those managed by renowned basketball analytics experts Ken Pomeroy and Jeff Sagarin, as well as ESPN’s Basketball Power Index.

“The NCAA Men’s Basketball Committee has had helpful metrics it has used over the years, and will continue to use the team sheets, but those will now be sorted by the NCAA Evaluation Tool,” Gavitt said. “As has always been the case, the committee won’t solely focus on metrics to select at-large teams and seed the field. There will always be a subjective element to the tournament selection process, too.”

The NET, explained: NCAA adopts new college basketball ranking

The NCAA has developed the NCAA Evaluation Tool (NET) a ranking system that will replace the RPI as the primary sorting tool used to evaluate teams.

www.ncaa.com

 

[auror]

I thought they do use it to some extent? My head hurts reading the below linked article but they use multiple metrics for the NCAA Tournament selection, not just solely the NET.

Another change made in 2017 to the team sheets was the inclusion of other metrics. These include the Kevin Pauga Index and ESPN’s results-oriented metric, the Strength of Record. The team sheets also included three predictive metrics: those managed by renowned basketball analytics experts Ken Pomeroy and Jeff Sagarin, as well as ESPN’s Basketball Power Index.

“The NCAA Men’s Basketball Committee has had helpful metrics it has used over the years, and will continue to use the team sheets, but those will now be sorted by the NCAA Evaluation Tool,” Gavitt said. “As has always been the case, the committee won’t solely focus on metrics to select at-large teams and seed the field. There will always be a subjective element to the tournament selection process, too.”

The NET, explained: NCAA adopts new college basketball ranking

The NCAA has developed the NCAA Evaluation Tool (NET) a ranking system that will replace the RPI as the primary sorting tool used to evaluate teams.

www.ncaa.com

Basically, the committee members get team sheets for every team. On it is listed a bunch of different metrics (NET, KenPom, BPI, etc.) along with the team's entire schedule. The schedule is not listed chronologically, though, it's arranged by the team's performance against the quadrants based on the NET.

So the NET has a large influence by way of being the organizational tool. Now the team sheets are just the start for the conversation by the committee members. But as everyone knows, how a conversation is framed has a big impact on the results.

The Warren Nolan site linked above actually has examples of a replica of what the NCAA uses with current rankings

 

[nelsonmuntz]

I thought they do use it to some extent? My head hurts reading the below linked article but they use multiple metrics for the NCAA Tournament selection, not just solely the NET.

Another change made in 2017 to the team sheets was the inclusion of other metrics. These include the Kevin Pauga Index and ESPN’s results-oriented metric, the Strength of Record. The team sheets also included three predictive metrics: those managed by renowned basketball analytics experts Ken Pomeroy and Jeff Sagarin, as well as ESPN’s Basketball Power Index.

“The NCAA Men’s Basketball Committee has had helpful metrics it has used over the years, and will continue to use the team sheets, but those will now be sorted by the NCAA Evaluation Tool,” Gavitt said. “As has always been the case, the committee won’t solely focus on metrics to select at-large teams and seed the field. There will always be a subjective element to the tournament selection process, too.”

The NET, explained: NCAA adopts new college basketball ranking

The NCAA has developed the NCAA Evaluation Tool (NET) a ranking system that will replace the RPI as the primary sorting tool used to evaluate teams.

www.ncaa.com

My issue is with the NET calculation, and I don't think it makes sense to conflate NET and Ken Pom when they are separate.

The risk of allowing the Selection Committee to use 3+ different computer rankings is it won't take long to find one that has a 19-15 Syracuse team over a 25-7 UConn team. The Committee should use clearly defined criteria from understandable sources that results in objective assessments. The NET has features that are black box and others that seem to reward teams for a) playing a tough schedule even if that are losing to it, and b) beating up weak opponents in paycheck games.

 

[pepband99]

My issue is with the NET calculation, and I don't think it makes sense to conflate NET and Ken Pom when they are separate.

The risk of allowing the Selection Committee to use 3+ different computer rankings is it won't take long to find one that has a 19-15 Syracuse team over a 25-7 UConn team. The Committee should use clearly defined criteria from understandable sources that results in objective assessments. The NET has features that are black box and others that seem to reward teams for a) playing a tough schedule even if that are losing to it, and b) beating up weak opponents in paycheck games.

I would rather they use multiple publicly-available ratings, which are under some scrutiny, given the "black box" piece of the NET.

Kenpom in constructive as a comparison to NET, whether or not it's used by the committee.

 

[gtcam]

It's too early to have our heads spinning with the NCAA's ridiculous conception of fairness and all this back and forth with Nelsonmuntz
UConn just needs to win as many games as possible and win the AAC tournament and all this stuff is moot
The NCAA does what they want - I don't care what garbled calculations they have posted that they claim to use - the bunch of idiots who meet every March in that tournament selection committee "war room" don't have enough brains to come in out of the rain never mind understand or even speak to that calculation.
They put in who the hell they want beyond the obvious auto bids.