| View previous topic :: View next topic |
| Author |
Message |
calbearman76
Joined: 02 Nov 2009
Posts: 5167
Location: Carson City
Back to top |
 Posted: 02/24/21 6:14 pm ::: Evaluating the NET |
Reply  |
|
The NET is the new tool for ranking teams and while it is better than the RPI it still is far from perfect. Admittedly this is a particularly problematic year because of Covid causing uneven schedules, fewer non-conference games and home courts without fans. But with only a few weeks left it is time to look at the NET and see where the problems are.
There are several teams that are rated far differently in the NET than either the polls, Massey and even the first reveal by the Selection Committee.
#4 Baylor
#7 Oregon
#11 Rutgers
#13 Texas A&M
#18 Stephen F Austin
#30 Bucknell
#61 DePaul
Baylor is a top 10 team, but pushing them into the top 5 is questionable. The reason for the high rating is clear; Kim Mulkey likes to run up scores which improves their efficiency ratings. 93 point wins over McNeese St and Northwestern St would drop their RPI (which is 16) but it helps their NET.
Oregon is another team that is overrated because they like to win big. In their first 11 games they were 9-2, winning by an average of 33 pts but losing to UCLA and Stanford. The Ducks are 0-6 over top 10 teams and have not beaten a top 30 team. Oregon is 30 in the RPI.
Rutgers played 4 non-conference games against bad competition (Monmouth, Hampton, Sacred Heart and Manhattan) and won by an average of more than 50 points. They lost their only top 25 game and have losses to Iowa and Nebraska. They have played only 14 games which gives those non-conference wins more weight. Rutgers is 48 in the RPI.
These three teams show that, counter to what had always been the philosophy of the NCAA, the NET rewards teams for running up scores. Whether this flows through to actual tournament selections is unclear. Baylor was 4 in the NET but only got a 3 seed in the first reveal.
Stephen F Austin is also significantly overrated. They lost their only 2 games to top 100 competition but still rate #18. This is another team that wins big.
Austin is 61 in the RPI. The significance is that one of their losses was to Houston (the other was to Arizona St) and that gives Houston a high quality Tier 1 victory. The rest of Houston's resume is unremarkable, losses to all other top 50 teams and a split with #55 Tulane, but no bad losses. They are a bubble team that may get the nod because of the overrating of SF Austin.
Bucknell is an example of the difficulty of rating a team with limited data. The Bison are undefeated but they have only played 8 games against other Patriot League teams. I don't think this a real problem with the NET but rather a problem with rating teams with limited data. But it is fun to note that until Lafayette beat Lehigh last weekend Bucknell was #1 in the RPI. They are now #5.
Texas A&M is significantly underrated. The Aggies have lost only 1 game in OT and have 6 wins over top 25 teams. A win over South Carolina on Sunday or in the SEC should give Texas A&M a number 1 seed. But A&M has played close games, winning 5 games by 5 or less. Never mind that they beat Kentucky by 17, Georgia by 12 and Tennessee by 10; In the NET this is not a top 10 team. Texas A&M is 2 in the RPI.
DePaul is perhaps the most alarming team. They have 5 losses, 4 to top 10 teams and an upset to Creighton last weekend. They beat Kentucky and also have a top 40 win over Marquette. Why? Because in three of those losses they lost by 41 to Louisville and twice by 33 to Connecticut. Maybe DePaul shouldn't be a top 25 team but they are clearly better than 61 which would leave them on the sidelines for the NCAA. They should get in because of their victory over Kentucky, but for teams that played them this is bad news.
|
|
Marquette Fan
Joined: 06 Mar 2005
Posts: 3582
Back to top |
| Posted: 02/24/21 6:59 pm ::: |
Reply |
|
| I was surprised when the NET numbers first came out how much higher in the rankings Marquette was than DePaul actually. DePaul moved down to a projected 8 seed in Creme's latest bracketology and that wasn't surprising to me. I think they should still make the NCAA Tourney but may not get that great a seed.
|
|
myrtle
Joined: 02 May 2008
Posts: 32341
Back to top |
| Posted: 02/24/21 7:17 pm ::: Re: Evaluating the NET |
Reply |
|
| calbearman76 wrote: |
Texas A&M is significantly underrated. The Aggies have lost only 1 game in OT and have 6 wins over top 25 teams. A win over South Carolina on Sunday or in the SEC should give Texas A&M a number 1 seed. But A&M has played close games, winning 5 games by 5 or less. Never mind that they beat Kentucky by 17, Georgia by 12 and Tennessee by 10; In the NET this is not a top 10 team. Texas A&M is 2 in the RPI.
|
To me A&M doesn't meet the eye test for a top, top team. I expect SoCaro to beat them fairly easily but if they don't, then I could change my mind. Yeah they won six teams in the top 25 but none of them were/are in the top 15 - some not even in the top 25 any more. Credit to them for winning all those games that they should win, but it's really a bit misleading.
|
|
calbearman76
Joined: 02 Nov 2009
Posts: 5167
Location: Carson City
Back to top |
| Posted: 02/24/21 8:43 pm ::: Re: Evaluating the NET |
Reply |
|
| myrtle wrote: |
| calbearman76 wrote: |
Texas A&M is significantly underrated. The Aggies have lost only 1 game in OT and have 6 wins over top 25 teams. A win over South Carolina on Sunday or in the SEC should give Texas A&M a number 1 seed. But A&M has played close games, winning 5 games by 5 or less. Never mind that they beat Kentucky by 17, Georgia by 12 and Tennessee by 10; In the NET this is not a top 10 team. Texas A&M is 2 in the RPI.
|
To me A&M doesn't meet the eye test for a top, top team. I expect SoCaro to beat them fairly easily but if they don't, then I could change my mind. Yeah they won six teams in the top 25 but none of them were/are in the top 15 - some not even in the top 25 any more. Credit to them for winning all those games that they should win, but it's really a bit misleading. |
True enough, but South Carolina hasn't beaten anyone better and they are number 3. And they have actually won 8 games that were in the top 25 when they beat them; I didn't count the wins over Texas and Mississippi St.
|
|
pilight
Joined: 23 Sep 2004
Posts: 67164
Location: Where the action is
Back to top |
| Posted: 02/24/21 10:54 pm ::: Re: Evaluating the NET |
Reply |
|
| calbearman76 wrote: |
| But it is fun to note that until Lafayette beat Lehigh last weekend Bucknell was #1 in the RPI. They are now #5. |
Also amusingly, they would have still dropped from #1 if Lehigh had won.
I'm more or less willing to give the NET a pass on this season. It's calibrated for teams to play more games, mitigating the effect of a few big blowouts.
_________________
The truth is like poetry
Most people hate poetry
|
|
calbearman76
Joined: 02 Nov 2009
Posts: 5167
Location: Carson City
Back to top |
| Posted: 02/25/21 4:57 am ::: Re: Evaluating the NET |
Reply |
|
| pilight wrote: |
| calbearman76 wrote: |
| But it is fun to note that until Lafayette beat Lehigh last weekend Bucknell was #1 in the RPI. They are now #5. |
Also amusingly, they would have still dropped from #1 if Lehigh had won.
I'm more or less willing to give the NET a pass on this season. It's calibrated for teams to play more games, mitigating the effect of a few big blowouts. |
I agree that this is a bad year to evaluate a new system but I am concerned using efficiency stats can benefit teams that run up big scores against bad teams. There may see some mitigating factors in the calculation that will show up with more games (and particularly with more non-conference games) but I remain skeptical. The system is much better than the RPI but I believe it still needs work.
|
|
ClayK
Joined: 11 Oct 2005
Posts: 11232
Back to top |
| Posted: 02/25/21 10:46 am ::: |
Reply |
|
| Margin of victory is always problematic. One option, used in most club tournaments to break ties, is to cap the credit for margin of victory at 15 (or 20 or 30). That way you get credit for being significantly better but not for pressing an inferior team for 40 minutes.
_________________
Oṃ Tāre Tuttāre Ture Svāhā
|
|
myrtle
Joined: 02 May 2008
Posts: 32341
Back to top |
| Posted: 02/25/21 12:26 pm ::: |
Reply |
|
| ClayK wrote: |
| Margin of victory is always problematic. One option, used in most club tournaments to break ties, is to cap the credit for margin of victory at 15 (or 20 or 30). That way you get credit for being significantly better but not for pressing an inferior team for 40 minutes. |
yep. that's a good way to do it. I think 20 is a good number. Sometimes you can't help but win by 50 even playing half a game with the bottom of the bench. (Then you have to consider why that team was even scheduled) But I agree that shouldn't count for more than beating a decent team by 20. Some coaches really like running up the score and others tend to not do it and are content to win by 12 rather than leaving starters in for 38 minutes in a blowout.
|
|
willtalk
Joined: 13 Apr 2012
Posts: 1107
Location: NorCal
Back to top |
| Posted: 02/26/21 7:00 am ::: |
Reply |
|
| All these issues do not have to be a factor if the selection committee does not totally rely on the ratings but just uses them as one factor to consider. I think one thing that could eliminate a lot of problems is if wins over teams that are to much lower in the ratings from a team in question are entirely eliminated from counting for a team, except if they are upset losses. That way a team can not be hurt nor helped by playing cupcakes.
_________________
No one one is ever as good as their best game, nor as bad as their worst.
|
|
|
|
