NCAA changes NET formula

[Nugget]

The NCAA has announced changes to the NET Rankings formula. https://www.ncaa.com/news/basketball...ge-net-2020-21

"The NET 'will be changed to increase accuracy and simplify it by reducing a five-component metric to just two. The remaining factors include the Team Value Index (TVI), which is a result-based feature that rewards teams for beating quality opponents, particularly away from home, as well as an adjusted net efficiency rating. The adjusted efficiency is a team’s net efficiency, adjusted for strength of opponent and location (home/away/neutral) across all games played. For example, a given efficiency value (net points per 100 possessions) against stronger opposition rates higher than the same efficiency against lesser opponents and having a certain efficiency on the road rates higher than the same efficiency at home. …

No longer will the NET use winning percentage, adjusted winning percentage and scoring margin."

These seem like reasonable, if minor, adjustments since winning percentage and "adjusted winning percentage," whatever that means, would presumably already be accounted for in the Team Value Index, while scoring margin would appear to already be accounted for by the adjusted net efficiency rating (except for the arbitrary cap they had put on scoring margin, which is probably good to do away with).

The release also states that "In addition, the overall and non-conference strength of schedule has been modernized to reflect a truer measure for how hard it is to defeat opponents. The strength of schedule is based on rating every game on a team's schedule for how hard it would be for an NCAA tournament-caliber team to win. It considers opponent strength and site of each game, assigning each game a difficulty score. Aggregating these across all games results in an overall expected win percentage versus a team's schedule, which can be ranked to get a better measure of the strength of schedule."

It's a little unclear, but I take it to mean this is a change to SOS rankings that is different from/in addition to the two components of the NET ranking? I don't quite follow how this is different from them doing SOS before, other than it appears that site of the game had not previously been part of determining SOS, which (if true) is nuts. I'm also unclear of how they were doing it before if not "aggregating [SOS components] across all games" - again, if they weren't doing that before, it also seems goofy.

[scottdude8]

Ugh. What purpose do I serve on this board if I can't be the "NET Whisperer"? If the NET is less complex I'm out of a job!

[uh_no]

The NCAA has announced changes to the NET Rankings formula. https://www.ncaa.com/news/basketball...ge-net-2020-21

"The NET 'will be changed to increase accuracy and simplify it by reducing a five-component metric to just two. The remaining factors include the Team Value Index (TVI), which is a result-based feature that rewards teams for beating quality opponents, particularly away from home, as well as an adjusted net efficiency rating. The adjusted efficiency is a team’s net efficiency, adjusted for strength of opponent and location (home/away/neutral) across all games played. For example, a given efficiency value (net points per 100 possessions) against stronger opposition rates higher than the same efficiency against lesser opponents and having a certain efficiency on the road rates higher than the same efficiency at home. …

No longer will the NET use winning percentage, adjusted winning percentage and scoring margin."

These seem like reasonable, if minor, adjustments since winning percentage and "adjusted winning percentage," whatever that means, would presumably already be accounted for in the Team Value Index, while scoring margin would appear to already be accounted for by the adjusted net efficiency rating (except for the arbitrary cap they had put on scoring margin, which is probably good to do away with).

The release also states that "In addition, the overall and non-conference strength of schedule has been modernized to reflect a truer measure for how hard it is to defeat opponents. The strength of schedule is based on rating every game on a team's schedule for how hard it would be for an NCAA tournament-caliber team to win. It considers opponent strength and site of each game, assigning each game a difficulty score. Aggregating these across all games results in an overall expected win percentage versus a team's schedule, which can be ranked to get a better measure of the strength of schedule."

It's a little unclear, but I take it to mean this is a change to SOS rankings that is different from/in addition to the two components of the NET ranking? I don't quite follow how this is different from them doing SOS before, other than it appears that site of the game had not previously been part of determining SOS, which (if true) is nuts. I'm also unclear of how they were doing it before if not "aggregating [SOS components] across all games" - again, if they weren't doing that before, it also seems goofy.

that makes a lot of sense. While I've never been as harsh on it as some here (it's not meant to be a predictor, it's meant to all of evaluate teams actual ability, reward them for winning on the court and incentivize scheduling in a way that the NCAA wants. I can see how many of the old components aimed to that purpose, but as has been long pointed out, there were some non-sensical decisions...things that ended up being double counted and weighted in strange ways due to the complexity (quadrants, especially...are nonsensical in a continuously valued function)

So this seems like a major step in the right direction. adjusted efficiency is held in high regard, and giving out a bonus for playing and winning the type of games the NCAA wants to see. That said, a lot of complexity can be hidden in those "two factors", and even in net efficiency ratings, where the many quants out there have sometimes meaningfully disparate results for what could all be termed "adjusted efficiency ratings"

So we'll see (hopefully) this coming season how it plays out. In any case, NET has been SUBSTANTIALLY better than RPI. so i'm game.

[House P]

Ugh. What purpose do I serve on this board if I can't be the "NET Whisperer"? If the NET is less complex I'm out of a job!

Don't fret too much. My understanding is that the the job of "DBR Net Whisperer" doesn't pay particularly well - certainly not as well as the job of Lead Bracketologist for ESPN.

At first glance, the modified NET looks to like a combination of a KenPom-like rating (a predictive measure based on adjusted efficiency margin without considering wins/losses) and Wins-Above-Bubble-like rating (a measure of a team's "resume" based only on wins/losses and schedule strength).

Most predictive measures reward margin of victory, so I wonder if they will do anything to discourage running up the score.

[Mtn.Devil.91.92.01.10.15]

*Joe Lunardi stays up all night updating algorithms*

[Indoor66]

*Joe Lunardi stays up all night updating algorithms*

*Joe Lunardi stays up all night making up algorithms*

FIFY

[Mtn.Devil.91.92.01.10.15]

*Joe Lunardi stays up all night making up algorithms*

FIFY

Okay, I think we just became best friends