(03-03-2020 09:49 AM)Nameless Wrote:  

(03-03-2020 07:00 AM)bearcatmark Wrote:  Our victim complex is getting out of hand

Seriously. It's like no one realizes everyone but Wichita State tanked their OOC, which drops everyone's NET. Even some of the teams Wichita beat that would be tourney caliber in recent years (VCU, South Carolina, Oklahoma State) haven't had good seasons, which prevents their NET from being as high as it could've been had those teams performed up to expectations.

UConn would be top 50ish at 18-11 had we not sucked against Saint Blows. UC would be solidly in the field had they not lost to Colgate and BGSU, as it stands they're still almost top 50 even with those horrible losses. Houston's best OOC win is @South Carolina (65) and they're a top 20 NET team.

Where is there evidence that the fix is in?

We didn't try to tank our non con, just look at this.

vs South Carolina (N)
@ Oklahoma State
vs Oklahoma
vs VCU
vs Mississippi

I would have said 2 Q1 chances, and the rest being Q2. Instead, only Oklahoma St. is a Q1 chance, while Oklahoma is in the 40s and VCU is in the 50s. Oh yeah, and Ole Miss is a Q3 chance.

(03-03-2020 10:40 AM)FadedCrown Wrote:  

(03-03-2020 09:49 AM)Nameless Wrote:  

(03-03-2020 07:00 AM)bearcatmark Wrote:  Our victim complex is getting out of hand

Seriously. It's like no one realizes everyone but Wichita State tanked their OOC, which drops everyone's NET. Even some of the teams Wichita beat that would be tourney caliber in recent years (VCU, South Carolina, Oklahoma State) haven't had good seasons, which prevents their NET from being as high as it could've been had those teams performed up to expectations.

UConn would be top 50ish at 18-11 had we not sucked against Saint Blows. UC would be solidly in the field had they not lost to Colgate and BGSU, as it stands they're still almost top 50 even with those horrible losses. Houston's best OOC win is @South Carolina (65) and they're a top 20 NET team.

Where is there evidence that the fix is in?

We didn't try to tank our non con, just look at this.

vs South Carolina (N)
@ Oklahoma State
vs Oklahoma
vs VCU
vs Mississippi

I would have said 2 Q1 chances, and the rest being Q2. Instead, only Oklahoma St. is a Q1 chance, while Oklahoma is in the 40s and VCU is in the 50s. Oh yeah, and Ole Miss is a Q3 chance.

You misread him. He excluded us.

I agree that we have no one to blame but ourselves. Some of it is just bad luck (i.e. scheduling good but the teams don't perform as well as expected) and others it's just tanking it. We had several crack at Top 50 wins this year and we all collectively lost a lot of them. When you don't at least snag a few BIG wins in the non-con, don't expect any favors to be paid down the road. We're reaping what we sowed in the non-con. And then you have some teams that did well in the non-con like Memphis (even though they had no flashy wins) but then tanked in the conference. That also kills extra bids.

I definitely don't think the NET is the answer and it always gives you pause when you have an organization not releasing the exact formula (they released basic principles of it but not the exact formula). But still it's better than just a standard RPI. I also think the quad thing gets a little out of hand. They shouldn't just look totally at quad wins because it's deceiving. A team could have 8 Q1 wins against a bunch of teams rated 50-70 and then another team only have like 4 Q1 wins but they're all Top 25. You got a look at the whole picture and whole body.

I think the Big Ten is a sham though with all of this. Losses count too and I'm just not cool with the concept of teams playing .500 ball throughout the course of the season and sub. .500 in the conference and still be considered on the bubble. If the conference is too "tough" for ya, then get out. Or maybe we shouldn't be having 16-team mega conferences. Nobody wants to watch a scrub P5 tournament.

(03-03-2020 09:49 AM)Nameless Wrote:  

(03-03-2020 07:00 AM)bearcatmark Wrote:  Our victim complex is getting out of hand

Seriously. It's like no one realizes everyone but Wichita State tanked their OOC, which drops everyone's NET. Even some of the teams Wichita beat that would be tourney caliber in recent years (VCU, South Carolina, Oklahoma State) haven't had good seasons, which prevents their NET from being as high as it could've been had those teams performed up to expectations.

UConn would be top 50ish at 18-11 had we not sucked against Saint Blows. UC would be solidly in the field had they not lost to Colgate and BGSU, as it stands they're still almost top 50 even with those horrible losses. Houston's best OOC win is @South Carolina (65) and they're a top 20 NET team.

Where is there evidence that the fix is in?

Memphis didn't tank its OOC. We went 11-2 without Wiseman. We tanked the AAC schedule, lol.

(03-03-2020 10:54 AM)ShockerFever Wrote:  I agree that we have no one to blame but ourselves. Some of it is just bad luck (i.e. scheduling good but the teams don't perform as well as expected) and others it's just tanking it. We had several crack at Top 50 wins this year and we all collectively lost a lot of them. When you don't at least snag a few BIG wins in the non-con, don't expect any favors to be paid down the road. We're reaping what we sowed in the non-con. And then you have some teams that did well in the non-con like Memphis (even though they had no flashy wins) but then tanked in the conference. That also kills extra bids.

I definitely don't think the NET is the answer and it always gives you pause when you have an organization not releasing the exact formula (they released basic principles of it but not the exact formula). But still it's better than just a standard RPI. I also think the quad thing gets a little out of hand. They shouldn't just look totally at quad wins because it's deceiving. A team could have 8 Q1 wins against a bunch of teams rated 50-70 and then another team only have like 4 Q1 wins but they're all Top 25. You got a look at the whole picture and whole body.

I think the Big Ten is a sham though with all of this. Losses count too and I'm just not cool with the concept of teams playing .500 ball throughout the course of the season and sub. .500 in the conference and still be considered on the bubble. If the conference is too "tough" for ya, then get out. Or maybe we shouldn't be having 16-team mega conferences. Nobody wants to watch a scrub P5 tournament.

I agree with everything you said, especially the parts about the lack of transparency and the mega conferences.

(03-03-2020 10:58 AM)TripleA Wrote:  

(03-03-2020 09:49 AM)Nameless Wrote:  

(03-03-2020 07:00 AM)bearcatmark Wrote:  Our victim complex is getting out of hand

Seriously. It's like no one realizes everyone but Wichita State tanked their OOC, which drops everyone's NET. Even some of the teams Wichita beat that would be tourney caliber in recent years (VCU, South Carolina, Oklahoma State) haven't had good seasons, which prevents their NET from being as high as it could've been had those teams performed up to expectations.

UConn would be top 50ish at 18-11 had we not sucked against Saint Blows. UC would be solidly in the field had they not lost to Colgate and BGSU, as it stands they're still almost top 50 even with those horrible losses. Houston's best OOC win is @South Carolina (65) and they're a top 20 NET team.

Where is there evidence that the fix is in?

Memphis didn't tank its OOC. We went 11-2 without Wiseman. We tanked the AAC schedule, lol.

Yea, I should've expanded upon my post to mention a couple teams weren't performing well in conference that did fine out of it. Even your OOC wasn't great though, as you suffered from the same unfortunate luck that Wichita did. Wins against Ole Miss, NC State and Tenn aren't as impressive as they were in years past. That loss to Georgia hurts as well. Had you won that game and been 4-1 OOC against teams from major conferences, and owned wins against Cincy and Houston in conference, you may be on the right side of the bubble right now.

If the NET is truly biased against the P5, then explain these teams.

BYU - 9 NET, 11 KenPom, 15 AP.
St. Mary's - 31 NET, 36 KenPom, RV AP.
Purdue - 35 NET, 25 KenPom, NR AP.
ETSU - 40 NET, 66 KenPom, RV AP.
Oklahoma - 41 NET, 34 KenPom, RV AP.
Indiana - 54 NET, 37 KenPom, NR AP.

Again the solution to this issue is quite easy and didn't require a change from a results based rating to a predictive one, which is ridiculous to begin with since we play the games for a reason.

One simple rule and the tournament would be incredible.

You must be .500 or better in conference including your tournament to qualify for an at large berth.

The tournament is about winners, the conference tourney winners and the best of the rest. You can't be the best of the rest and not be in the best of your conference. Do this and you will see a lot of the complaints about ratings and the committee die down. It sets a floor.

(03-03-2020 01:38 PM)Foreverandever Wrote:  Again the solution to this issue is quite easy and didn't require a change from a results based rating to a predictive one, which is ridiculous to begin with since we play the games for a reason.

One simple rule and the tournament would be incredible.

You must be .500 or better in conference including your tournament to qualify for an at large berth.

The tournament is about winners, the conference tourney winners and the best of the rest. You can't be the best of the rest and not be in the best of your conference. Do this and you will see a lot of the complaints about ratings and the committee die down. It sets a floor.

This is a terrible point that people repeatedly make because they focus on your actual NET number (which matters very very little) instead of the actual evaluation of your resume the committee does (which is heavily focused on your wins/losses... i.e. playing the games).

Predictive metrics are better for doing that evaluation because they provide a more accurate view of the quality of your wins and losses. I love Strength of Record metrics and those should be used as well, but not to group the teams you played in comparing quad wins.

.
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

.

Overall Question:

Does the NET ranking system do an accurate, un-biased job of identifying quality teams from non-P5 and non-Big East conferences?


Q: How many current NET top 50 teams have lost 10+ games to date?

A: 16 top 50 teams have 10+ losses.

Q: How many of those 10+ loss teams are P5 or Big East teams?

A: All 16 (100%) teams with 10+ losses are P5 or Big East teams.


Q: How many teams from the P5 and Big East conferences are currently ranked in the NET top 50?

A: 40 (80% of top 50 NET teams)

Q: What percentage of the 75 P5 and Big East teams are in the NET top 50?

A: 53% (40/75 = 53.3%)

Q: How many of the teams with current top 50 NET rankings come from the other 26 (non P5/Big East) conferences?

A: 10 (only 20% of all top 50 teams)

Q: What percentage of the 272 non-P5/Big East teams are currently ranked in the NET top 50?

A: 3.7% (10/272 = 3.68%)

Q: What was the average likelihood that any P5 or Big East team, selected at random before the season started, would currently have a top 50 NET ranking?

A: Approximately 50/50 chance (.533)

Q: What was the average likelihood that any non-P5/Big East team, selected at random before the season started, would currently have a top 50 NET ranking?

A: Approximately 4/100 chance (0.0368)

Q: How much more likely was it that any randomly-selected P5 or Big East team (selected before the season began) would be in the current NET top 50 than it was for any randomly-selected non-P5/BE team to be ranked in the current NET top 50?

A: On average, any randomly-selected P5 or Big East team (selected before the season began) was 14.5 times more likely than any randomly-selected non-P5/BE team to be in the current NET top 50 (0.533/0.0368 = 14.48).


============================================

On average, only 5.25 (16.4%) of the 32 NCAA at-large bids have gone to non P5/BigEast teams between 2016 and 2019.
[/quote]

============================================

.
NOTE:

From 1939 through 1974 (the first completely D1 NCAA Tournament with 25 teams), only one team per conference was invited into the NCAA tournament field.

From 1975 through 1978, the field was expanded to 32 teams. All conference champions were invited, and there were only 11 at-large teams, several of which went to independent teams. At most, only 2 teams per conference participated.

This, and the prior arrangement, was the practice in the spirit of fairness. The NIT tournament provided an opportunity for non-conference champions to compete in the post-season.

For example, in 1978, there were only two ACC and two Big 10 teams in the 1978 NCAA tournament: Duke, N.C., MSU & Indiana.

The number of at-large bids increased from 11 to 36 when the tournament field was expanded from 32 to 68 teams.


Q: How have the 25 additional at-large bids been allocated?

A: All 25 additional at-large bids have gone to P5 and Big East teams.

Q: What about the 11 at-large bids that have existed since 1975?

A: Some, but not all have gone to non-P5/Big East teams.

Examples:

In 2019, only 7 at-large bids went to non P5/Big East teams.

In 2018, only 5 at-large bids went to non P5/Big East teams.

In 2017, only 4 at-large bids went to non P5/Big East teams.

In 2016, only 5 at-large bids went to non P5/Big East teams.
[size=medium]

.

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

(03-03-2020 02:28 PM)bearcatmark Wrote:  

(03-03-2020 01:38 PM)Foreverandever Wrote:  Again the solution to this issue is quite easy and didn't require a change from a results based rating to a predictive one, which is ridiculous to begin with since we play the games for a reason.

One simple rule and the tournament would be incredible.

You must be .500 or better in conference including your tournament to qualify for an at large berth.

The tournament is about winners, the conference tourney winners and the best of the rest. You can't be the best of the rest and not be in the best of your conference. Do this and you will see a lot of the complaints about ratings and the committee die down. It sets a floor.

This is a terrible point that people repeatedly make because they focus on your actual NET number (which matters very very little) instead of the actual evaluation of your resume the committee does (which is heavily focused on your wins/losses... i.e. playing the games).

Predictive metrics are better for doing that evaluation because they provide a more accurate view of the quality of your wins and losses. I love Strength of Record metrics and those should be used as well, but not to group the teams you played in comparing quad wins.

People in favor of metrics always make this worthless point. Losing by 10 or by 1 or by 100 is losing. I care what their win loss record is, not how "well" they played losing. Stat junkies many of whom rarely played competitive sports love that stuff because it's a comparison for them not a competition.

I don't care how close the Nazi's came to breaking the line at the battle of the bulge. Patton's brought home the win.


Of course my point remains, use any stat system you want. If your under .500 in your conference you don't deserve an at large. That's result based.

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

(03-04-2020 12:47 AM)Foreverandever Wrote:  

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

You must love the RPI then.

(03-04-2020 12:47 AM)Foreverandever Wrote:  

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

I never said anything about computers deciding the whole thing, or not playing the games. He's talking about the NET being biased in favor of 6 conferences, I cited other analytics that show it's not biased. If he made the argument that all predictive analytics were biased, what you're saying would be a relevant topic to debate. But he's making the case that NET specifically is, and that it was created by the "cartel" to favor teams from those conferences. Which is asinine when you consider other formulae that have been shown to be useful for predicting the outcome of games all pretty much line up with the NET, or actually rate some of these teams higher than the NET does.

Obviously, playing in a conference with a lot of highly ranked teams is going to boost your ranking in any predictive system, which will in turn inflate the ranking of bad teams. A good team playing in a conference with a lot of bad teams will obviously lower the ranking of said team. No system is going to be perfect in a league with over 300 teams, the talent disparity is too wide. But these predictive systems do a solid job overall of predicting outcomes, and are thus useful tools when seeding teams for a tournament.

(03-04-2020 09:15 AM)Nameless Wrote:  

(03-04-2020 12:47 AM)Foreverandever Wrote:  

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

I never said anything about computers deciding the whole thing, or not playing the games. He's talking about the NET being biased in favor of 6 conferences, I cited other analytics that show it's not biased. If he made the argument that all predictive analytics were biased, what you're saying would be a relevant topic to debate. But he's making the case that NET specifically is, and that it was created by the "cartel" to favor teams from those conferences. Which is asinine when you consider other formulae that have been shown to be useful for predicting the outcome of games all pretty much line up with the NET, or actually rate some of these teams higher than the NET does.

Obviously, playing in a conference with a lot of highly ranked teams is going to boost your ranking in any predictive system, which will in turn inflate the ranking of bad teams. A good team playing in a conference with a lot of bad teams will obviously lower the ranking of said team. No system is going to be perfect in a league with over 300 teams, the talent disparity is too wide. But these predictive systems do a solid job overall of predicting outcomes, and are thus useful tools when seeding teams for a tournament.

Ok so here is how T-Rank, a near clone of KenPom, is calculated.
Game Adj. OE = PPPo / (Opponent's Adj. DE / Average PPP)
Game Adj. DE = PPPd / (Opponent's Adj. OE / Average PPP)
This is adjusted for location.

He adds recency bias and devalues blowouts in mismatches. He also locks the game when the game is analytically final (when the losing team has no chance to win).

How is this in any way biased towards* the P5?

If you want to read for yourself: http://adamcwisports.blogspot.com/p/ever...ounts.html

Some of you are clowns.

(03-04-2020 10:23 AM)FadedCrown Wrote:  

(03-04-2020 09:15 AM)Nameless Wrote:  

(03-04-2020 12:47 AM)Foreverandever Wrote:  

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

I never said anything about computers deciding the whole thing, or not playing the games. He's talking about the NET being biased in favor of 6 conferences, I cited other analytics that show it's not biased. If he made the argument that all predictive analytics were biased, what you're saying would be a relevant topic to debate. But he's making the case that NET specifically is, and that it was created by the "cartel" to favor teams from those conferences. Which is asinine when you consider other formulae that have been shown to be useful for predicting the outcome of games all pretty much line up with the NET, or actually rate some of these teams higher than the NET does.

Obviously, playing in a conference with a lot of highly ranked teams is going to boost your ranking in any predictive system, which will in turn inflate the ranking of bad teams. A good team playing in a conference with a lot of bad teams will obviously lower the ranking of said team. No system is going to be perfect in a league with over 300 teams, the talent disparity is too wide. But these predictive systems do a solid job overall of predicting outcomes, and are thus useful tools when seeding teams for a tournament.

Ok so here is how T-Rank, a near clone of KenPom, is calculated.
Game Adj. OE = PPPo / (Opponent's Adj. DE / Average PPP)
Game Adj. DE = PPPd / (Opponent's Adj. OE / Average PPP)
This is adjusted for location.

He adds recency bias and devalues blowouts in mismatches. He also locks the game when the game is analytically final (when the losing team has no chance to win).

How is this in any way biased against the P5?

If you want to read for yourself: http://adamcwisports.blogspot.com/p/ever...ounts.html

Some of you are clowns.

Did you mean to ask how is it biased in favor of the P5? No one is saying it's biased against them, lol.

I like the recency adjustment used in T-Rank. Teams are constantly improving (UConn) and getting worse (DePaul), so accounting for fluctuations in performance is important when looking for accurate results.

(03-04-2020 10:23 AM)FadedCrown Wrote:  

(03-04-2020 09:15 AM)Nameless Wrote:  

(03-04-2020 12:47 AM)Foreverandever Wrote:  

(03-03-2020 11:59 PM)Nameless Wrote:  

(03-03-2020 05:52 PM)jedclampett Wrote:  .
Minnesota (13-15) has a current NET ranking of #46.


Q: Mathematically, how can a team with a losing record be ranked among the top 50 teams in the nation?

A: All but one of Minnesota's 15 losses have been to teams that currently have winning records.

Q: How about their wins?

A: Seven (54%) of their wins were vs. teams that currently have losing records. Two were vs. teams with current records of 15-14. Only four (31%) of their wins were vs. teams with more than 17 wins, and only three (23%) of their wins were on the road/away.

Q: What was their Q1/Q2 record?

A: Q1: 5-12 Q2: 2-2 Q3 1-1 Q4: 5-0 (Q1/Q2: 7-14) (Q3/Q4: 6-1)

NOTE: 21 (75%) of their 28 games have been Q1 or Q2 contests.

============================================

Q: What does this tell us about how the NET formula works?

A: A team can get a top 48 NET ranking with a losing record by:

#1 Being in a conference with a schedule of ≥ 80% Q1 or Q2 games

#2 Scheduling 5 or 6 probable Q1 and Q2 OOC games.

#3 Winning ~ 50% of Q1 home games, and ≥ 33% of all Q1/Q2 games.

#4 Winning ≥ 85% of all Q3 and Q4 games.

============================================

As others have pointed out, your points would be a lot more valid if NET overrated these teams relative to other analytical systems. Minnesota is 46 in NET, 32 in Kenpom, and 48 in Sagarin at the time of me writing this. So why bring up the Gophers as if they prove your point about fairness? Seems like their NET is in line with other respected formulae.

Also, in regards to your post below the one I'm quoting, you realize how disingenuous it is to compare 6 conferences to 26 and act as if the percentages you're coming up with hold any weight... Right? Of course there is a lower percentage chance that a team from a pool of 272, when selected at random, will be in the top 50 than a team from a pool of 75, that's simple math. Throw in the talent disparity between conferences like the B1G, ACC and Big East versus conferences like the America East, SWAC, MEAC, etc and of course that gap is going to be astronomical.

If you really wanted to be fair, look at teams that have historically been competitive against the 6 conferences you listed (Gonzaga, St Mary's, Houston, Memphis, Cincy, Dayton, VCU, etc) and then examine their chance of being in the top 50 relative to teams from the other 6 conferences. Don't dilute your results by putting Hartford in the same pool as Gonzaga lol.

Well geez aren't all those systems predictive in nature? Wouldn't that be the reason they are all similar?

Again we play games for a reason and it's not to build stats so we can argue about who is better. Hell if that's what we are doing let's quit going to games, have players shoot in empty gyms, and then just let the computers decide the whole thing.

I mean predictive measures, even in games as they happen, are often wrong whether it is computer nerds or Vegas. Let's just play the games.

I never said anything about computers deciding the whole thing, or not playing the games. He's talking about the NET being biased in favor of 6 conferences, I cited other analytics that show it's not biased. If he made the argument that all predictive analytics were biased, what you're saying would be a relevant topic to debate. But he's making the case that NET specifically is, and that it was created by the "cartel" to favor teams from those conferences. Which is asinine when you consider other formulae that have been shown to be useful for predicting the outcome of games all pretty much line up with the NET, or actually rate some of these teams higher than the NET does.

Obviously, playing in a conference with a lot of highly ranked teams is going to boost your ranking in any predictive system, which will in turn inflate the ranking of bad teams. A good team playing in a conference with a lot of bad teams will obviously lower the ranking of said team. No system is going to be perfect in a league with over 300 teams, the talent disparity is too wide. But these predictive systems do a solid job overall of predicting outcomes, and are thus useful tools when seeding teams for a tournament.

Ok so here is how T-Rank, a near clone of KenPom, is calculated.
Game Adj. OE = PPPo / (Opponent's Adj. DE / Average PPP)
Game Adj. DE = PPPd / (Opponent's Adj. OE / Average PPP)
This is adjusted for location.

He adds recency bias and devalues blowouts in mismatches. He also locks the game when the game is analytically final (when the losing team has no chance to win).

How is this in any way biased towards* the P5?

If you want to read for yourself: http://adamcwisports.blogspot.com/p/ever...ounts.html

Some of you are clowns.

Hmmmm, does everyone play the same number of home games? Or are certain conferences running up their offensive and defensive efficiency numbers at home against bad teams who are on the road for a month straight? Then hold on versus other conference teams who played similar schedules and boosted their ratings.

That's how its biased. The system is reinforcing a series of inequalities to say that the team who started off with the benefits are better than ones who did not. Yet when these games are actually played it doesn't hold to the model. Of course recently the committee has gone out of it's way to force non p5 teams into playing each other early and often in the bracket to knock down some of those frequent little school upsets.

Now can you make a strong arguement for why a sub .500 team in a conference deserves an at large?