shizzle787
King Shizzle DCCLXXXVII of the Cesspool
- Joined
- Oct 19, 2015
- Messages
- 12,531
- Reaction Score
- 21,034
Why the NET and computer numbers don't matter as much as you think...
- Thread starter shizzle787
- Start date
shizzle787
King Shizzle DCCLXXXVII of the Cesspool
- Joined
- Oct 19, 2015
- Messages
- 12,531
- Reaction Score
- 21,034
I was getting at something but it wasn't about UConn but the rest of the middle of the Big East.
- Joined
- Oct 22, 2013
- Messages
- 781
- Reaction Score
- 1,499
Thanks for the video. I love talking college basketball & Jerry Pom is very knowledgeable, so it was a good listen.
Putting on my critic’s hat, here are a couple of points about whether & how much the computers and the NET matter:
1. Jerry told us how decisions are made by the committee. First of all, he’s never sat on a committee and never will, so how does he know? Second, as a long time observer, I’ve noticed that committee members in post-deliberation interviews talk about how individual each committee’s dynamic is. So, I don’t think it’s possible to make broad generalizations about the committee’s decision making.
2. Jerry was just wrong on some of his facts. He made the claim that the NET strongly correlates with scoring margin and that if you check it, you’ll find that the top ten on the NET are all top ten in scoring margin with the exception of Purdue. Do I checked it. Not true. Only 5 teams showed up on the top 10 of both. So, I extended it to the top 15 just to cover the possibility of some near misses. Only one school was added - Purdue - making it only 6 of the top 15 that overlap. I wasn’t surprised because power ratings like Sagarin and Team Rankings do in fact make heavy use of scoring margin with adjustments for home court. In my experience of comparing them, they do not correlate well with the NET. The NET is a very unique formula with its own formula. Jerry Pom is a competitor, so I guess it’s good fir business to bash the competition.
3. It was interesting to listen to the conversation about whether Alabama should be a 1 or a 2 but either way about how strong they are. Either way, we beat the pants off Alabama by 15 points on a neutral court.
Putting on my critic’s hat, here are a couple of points about whether & how much the computers and the NET matter:
1. Jerry told us how decisions are made by the committee. First of all, he’s never sat on a committee and never will, so how does he know? Second, as a long time observer, I’ve noticed that committee members in post-deliberation interviews talk about how individual each committee’s dynamic is. So, I don’t think it’s possible to make broad generalizations about the committee’s decision making.
2. Jerry was just wrong on some of his facts. He made the claim that the NET strongly correlates with scoring margin and that if you check it, you’ll find that the top ten on the NET are all top ten in scoring margin with the exception of Purdue. Do I checked it. Not true. Only 5 teams showed up on the top 10 of both. So, I extended it to the top 15 just to cover the possibility of some near misses. Only one school was added - Purdue - making it only 6 of the top 15 that overlap. I wasn’t surprised because power ratings like Sagarin and Team Rankings do in fact make heavy use of scoring margin with adjustments for home court. In my experience of comparing them, they do not correlate well with the NET. The NET is a very unique formula with its own formula. Jerry Pom is a competitor, so I guess it’s good fir business to bash the competition.
3. It was interesting to listen to the conversation about whether Alabama should be a 1 or a 2 but either way about how strong they are. Either way, we beat the pants off Alabama by 15 points on a neutral court.
- Joined
- Feb 2, 2012
- Messages
- 3,001
- Reaction Score
- 12,330
Seton Hall is now on Lunardi's Bubble in his updated Bracketology today. So BE has 6 viable tourney teams at the moment. St Johns & Nova are both top 75 in Kenpom so they are decent teams even if not tourney teams. Nothing weak about the BE this year.
As for Palm, his take on the NET is little contradictory....he says your opponents NET matters because it shows the quality of your wins but your NET ranking really doesn't matter to committee. So how does that make sense? The NET ranking isn't the end all be all but it matters.
- Joined
- Aug 17, 2011
- Messages
- 18,059
- Reaction Score
- 104,036
In recent years the NCAA has set up a mock selection with the media so they can show everyone what they do. I couldn't find a more recent article than this but I know they've had these more recently also.
Behind the bracket: a mock NCAA tournament selection
A lack of dominant teams could make this year's NCAA men's basketball selection a challenge.
www.usatoday.com
- Joined
- Oct 22, 2013
- Messages
- 781
- Reaction Score
- 1,499
I didn’t mention it in my post, but I thought exactly the same thing about him accepting the NET as a valid measure of strength of opponents for quad purposes but not a valid measure of the strength of an individual school to qualify for selection, I agree that it makes absolutely no sense. Complete non sequitor.
shizzle787
King Shizzle DCCLXXXVII of the Cesspool
- Joined
- Oct 19, 2015
- Messages
- 12,531
- Reaction Score
- 21,034
Seton Hall is 13-9. At a minimum they need to go 7-2 the rest of the way. They are on the very fringe of the bubble. At the end of the day, W/L records do count.
nelsonmuntz
Point Center
- Joined
- Aug 27, 2011
- Messages
- 45,623
- Reaction Score
- 37,334
RPI was better than NET. It was simple, straightforward, and measured wins and losses. No matter how anyone cuts it, Palm is right. Scoring margin drives the NET. Run up the score on a schedule of bad teams and you have a good shot of getting a bid.
It is ironic that the NET, a supposedly superior tool, is actually less decisive in bid selection now than the RPI was in the 2000's. The RPI would pretty much dictate which teams got in and, for the most part, where they got seeded, with a handful of exceptions every year. Now, even the committee does not trust the NET, because they will jump around to select and seed the teams outside of where the NET would indicate.
It is ironic that the NET, a supposedly superior tool, is actually less decisive in bid selection now than the RPI was in the 2000's. The RPI would pretty much dictate which teams got in and, for the most part, where they got seeded, with a handful of exceptions every year. Now, even the committee does not trust the NET, because they will jump around to select and seed the teams outside of where the NET would indicate.
- Joined
- Aug 26, 2011
- Messages
- 899
- Reaction Score
- 6,066
I think scoring margin provides useful information for predictive purposes, as long as it is weighted correctly.
- Joined
- Oct 22, 2013
- Messages
- 781
- Reaction Score
- 1,499
Thank you for sharing your knowledge on this subject, Nelson.
I’m not that smart, so could you please explain how the NET formula is driven by scoring margin. I’d really appreciate the help. I’m linking an article which explains the formula so that we have a common starting point.
NCAA NET Rankings Explained and How to Use Them When Betting College Basketball
- Joined
- Aug 29, 2011
- Messages
- 13,343
- Reaction Score
- 23,375
RPI badly over rated SIS. NET still does that, just not as much. But it over rates margin of victory to counteract the improvements in reducing the influence of SOS. SOS is itself a flawed tool that makes bad teams in good leagues seem better than they are and good team in weaker leagues seem worse than they are. These computer generated models are just pseudo science. The basketball equivalent of using leeches to cure disease.