ADVERTISEMENT

The NET is a Problem



I don't get it.

Iowa State has zero wins over NCAA Tournament teams (we might have 3 at this point depending on St. Joes and Penn)

Same goes with Houston, who is somehow number 1.

Houston has played NOBODY all year long (have played zero ranked teams, and only scheduled 1 team in the preseason top-25, a mediocre Texas A&M team and they almost lost).

Yet somehow they have FOUR Q1 wins (again, played a bunch of nobodies).

They really have the NET figured out like a cheat code.

Utah, Dayton, Texas A&M (7-4) and Xavier (6-5), all unranked.

Iowa State has beaten VCU, Depaul and Iowa (All 3 of them STINK), lost to Virginia Tech and that same mediocre Texas A&M team, and they are 6th in the NET.

MAKE IT MAKE SENSE.

Maybe Cal had it right. Pack in a bunch of games against stinky Power-5 opponents and beat them, and then blow out sub 300 teams and be top-5 in the NET.
It doesn’t make a lick of sense, but if those teams keep playing like they have they will fall hard by the end of the year.
 
Iowa State lost to 2 pretty mediocre, some might say BAD teams, and has zero good wins.

And I have watched them twice (both wins) and they barely even looked like a top-25 team IMO.

They are not the 13th best team in America.
I think the Big 12 will expose them.
 
NCAA in general are Clowns with with EVERYTHING

But Big 3 tournaments have always been a Joke
...Football: Had to be forced into a 12 Team football
...Baseball: Gives home field hosts based on $$ income
...Basketball: Garbage in Garbage out Computer ratings
 
  • Like
Reactions: Fox2monk
They got rid of RPI when that one mid-major hired a math genius to figure it out and tell their teams how to schedule.

Looks like NET has been figured out as well
 
people keep saying "UNCW" "UNCW" "UNCW".
And yes that was a bad loss. The worst part is it was at home.

But look at it this way, right now in the NET, UNCW is right there with Ariz St & Okl St, and ahead of Penn St, FSU, UCLA, Ga Tech, Stanford, LSU, and far ahead of WVU, G-town, Oregon St, Maryland, Cal, UL, ND, Vandy, & DePaul.

So UNCW, according to the NET, is as good or better than:
- 2 Big East team
- 2 SEC teams
- 2 Big 10 teams
- 2 Big 12 teams
- 4 ACC teams
- 5 PAC 10 teams
That is over 20% of the P5 schools.
 
Last edited:
For those saying "Just use Vegas!", remember that Vegas is using mathematical models to predict the opening line.
The "best" mathematical model isn't the NET, it is the Colley Matrix. It was developed by Dr. Wes Colley and uses objective mathematics to determine rankings based only on win and loses.
Not sure why the NCAA doesn't use this method over the NET.
 
For those saying "Just use Vegas!", remember that Vegas is using mathematical models to predict the opening line.
The "best" mathematical model isn't the NET, it is the Colley Matrix. It was developed by Dr. Wes Colley and uses objective mathematics to determine rankings based only on win and loses.
Not sure why the NCAA doesn't use this method over the NET.

What they should do is just use a combination of respected computer models. WIsdom of the crowds. The aggregate is usually better than any one specific system.

Also I can't see how a system that determines ranking based only on wins and losses would ever be considered the "best". It's been proven in many many sports that margin of victory is a better predictor of future success than actual won/loss record.
 
  • Like
Reactions: Aike
But I think the whole discussion is so overblown. Ultimately, the seeding is still being done by a bunch of humans sitting in a room. While NET figures are plastered all over the team sheets they are using, it's not as if they go through the NET ranks team by team and base the seedings exactly on that. It's not as if they go well this team is 16th so they obviously are the worse 4 seed. It doesn't really work like that.
 
No, no it won't. Enough with it.
It really won't. So long as UNC-W doesn't fall off a cliff.

UNC-W is currently a Quad 3 Loss. Houston had 1 quad 3 loss last year and was a 1 seed. #1 overall IIRC. There's still time to climb to a 1 seed.
 
  • Like
Reactions: TheDude73
The NET is definitely not the most accurate measure of which teams are best. The only positive thing I will say about it is the same thing I'd say about most of the math-based systems- they usually become more accurate as the season goes. Right now, they're not very accurate. After these teams start playing their conference opponents, it will become more accurate among the high major programs. Unfortunately, it then becomes less accurate among the mid-major and smaller programs. I hope (and expect) that the people choosing the NCAA-T field are aware that these systems are not the authority.
 
What they should do is just use a combination of respected computer models. WIsdom of the crowds. The aggregate is usually better than any one specific system.

Also I can't see how a system that determines ranking based only on wins and losses would ever be considered the "best". It's been proven in many many sports that margin of victory is a better predictor of future success than actual won/loss record.
Saying any one model is better than another, or is "the best", is very subjective. Especially when there is no agreement on what is more/most important wrt the various factors. I could say my model (which I ran for several years about 10-20 years ago) is "the best", as could any other model creator. So honestly, the "best" would be one that is a composite of several different models.



Just an FYI, my model included as factors:
- Who you beat or lost to, and where (Home, Away, Neutral, semi-Neutral like UK playing in Louisville). The home/away factor was based on data from prior year within conference games to determine how much to weight H vs A.
- Margin of victory, with an adjustment if it was an OT game, but also a cap where winning by 40 not much better than winning by 20.
- How long ago the game was (a game in late Feb was weighted more than one in early Nov).
- And I figured out a way to run the model so that it created probabilities between teams with common opponents, and each iteration extended that out further.
 
Yep, one bad loss like that can serve like an anchor weighing down your NET score all season long.

But, fortunately, UNCW is ranked in the 100s instead of the 300s like Evansville a few years ago. We’ve got plenty time and opportunity to fix that glitch.

Yeah if we go 15 and 3 in the sec and win out rest of ooc we are 26 and 5 going into sec tourney. That team record gets a 2 seed or better no matter what IMO.
 
  • Like
Reactions: Cats192
Saying any one model is better than another, or is "the best", is very subjective. Especially when there is no agreement on what is more/most important wrt the various factors. I could say my model (which I ran for several years about 10-20 years ago) is "the best", as could any other model creator. So honestly, the "best" would be one that is a composite of several different models.



Just an FYI, my model included as factors:
- Who you beat or lost to, and where (Home, Away, Neutral, semi-Neutral like UK playing in Louisville). The home/away factor was based on data from prior year within conference games to determine how much to weight H vs A.
- Margin of victory, with an adjustment if it was an OT game, but also a cap where winning by 40 not much better than winning by 20.
- How long ago the game was (a game in late Feb was weighted more than one in early Nov).
- And I figured out a way to run the model so that it created probabilities between teams with common opponents, and each iteration extended that out further.

And you ever notice that the majority of them don't ever go back and actually look at exactly how they did.

Kenpom has a model diagnostics page but for the most part many of them I don't see anything regarding just how accurate they measured things.

In general tho I think most of these computer rankings have the correct methodology. Factor in SOS, factor in margin of victory, where the game took place.

Yeah honestly I wouldn't mind if they just canned the committee and took an average of say 5-10 good models.
I do think tho the humans get a bit too much hate and they generally get the top teams (the ones that can actually compete for a title) correct.
 
  • Like
Reactions: fatguy87
All of these models should be considered but about 50% of the seeding factor should be placed on the eye test or common sense.
 
There is nothing to " get". Its a bunch of pinheads in a board room making shit up . Morons .
Nobody is in a board room “making up” the NET rankings. They are entirely the result of a computer objectively processing data through an algorithm.
 
I think there is a fair way to do it. Let Vegas do it. Who is better at handicapping games? Won't ever happen, but I think it would be the most accurate. I would say all conference champions get their automatic bids and let Vegas determine who else gets in and the seeding.
As strange to some people ( and to me, as well) as this idea may seem, I think you may have the best way to do it right there. Vegas puts money where the mouth is, and they are amazing at handicapping games. You make an awesome point with this post, imho.

well done, well said.

the more i think about it, the better I like it.
 
  • Like
Reactions: Blueheart32
And you ever notice that the majority of them don't ever go back and actually look at exactly how they did.

Kenpom has a model diagnostics page but for the most part many of them I don't see anything regarding just how accurate they measured things.

In general tho I think most of these computer rankings have the correct methodology. Factor in SOS, factor in margin of victory, where the game took place.

Yeah honestly I wouldn't mind if they just canned the committee and took an average of say 5-10 good models.
I do think tho the humans get a bit too much hate and they generally get the top teams (the ones that can actually compete for a title) correct.
Yes, but. I think most of them do not use SOS very well. I think a lot of them simply average the SOS of a team's opponents which is flawed. For example two top 25 teams 10 game schedules are:
Team A opps: 5, 10, 15, 20, 25, 300, 305, 310, 315, 320 has an average SOS = 162.5
Team B opps: 120, 125, 130, 135, 140, 145, 150, 155, 160, 165 has an average SOS = 142.5
So, by average SOS Team B has the tougher schedule, BUT if they are top 25 teams they should both almost always win the vs the 100+ Opps, thus Team A has 5 "tough" games and Team B has 0 "tough" games, therefore Team A really has the more difficult schedule. SOS should be adjusted relative to where you are. Because if in that example Team A & B were both in the 120-150 range, then it changes and Team B has the tougher schedule.

By using multiple models (like you suggest) a team would have a much tougher time "playing the system"
 
As strange to some people ( and to me, as well) as this idea may seem, I think you may have the best way to do it right there. Vegas puts money where the mouth is, and they are amazing at handicapping games. You make an awesome point with this post, imho.

well done, well said.

the more i think about it, the better I like it.
There is a problem with that. Vegas bases their odds not strictly on outcome, but moreso on WHERE the MONEY is going to be BET. For example, Vegas knows how many UK, Duke, UNC fans there are, and even casual fans know those Blue Bloods, but may not know Creighton or San Diego State. And so they know more will be bet on those Blue Bloods than probably should be, so they adjust because they want an approximate equal amount of betters on each side of the bet (assuming the games aren't rigged).
 
  • Like
Reactions: runt#69
There is a problem with that. Vegas bases their odds not strictly on outcome, but moreso on WHERE the MONEY is going to be BET. For example, Vegas knows how many UK, Duke, UNC fans there are, and even casual fans know those Blue Bloods, but may not know Creighton or San Diego State. And so they know more will be bet on those Blue Bloods than probably should be, so they adjust because they want an approximate equal amount of betters on each side of the bet (assuming the games aren't rigged).
great point
 
The NET, like the RPI, can be hacked.

Who is hacking it faster and better? Not Kentucky.

The reality is, you can hack the net by playing "the right" teams in neutral and road games. You can beat "not good" teams on the road/neutral and you can get Q1/Q2 wins, without really challenging yourself.

Instead, UK plays a bunch of games at Rupp against teams that can only hurt them.
 
There is a problem with that. Vegas bases their odds not strictly on outcome, but moreso on WHERE the MONEY is going to be BET. For example, Vegas knows how many UK, Duke, UNC fans there are, and even casual fans know those Blue Bloods, but may not know Creighton or San Diego State. And so they know more will be bet on those Blue Bloods than probably should be, so they adjust because they want an approximate equal amount of betters on each side of the bet (assuming the games aren't rigged).
This is true also, but I still think as far as picking who are the best teams after the auto bids are done, Vegas would do a better job than a committee of people who are associated with conferences and schools. Truth is there is probably no perfect way to do it.
 
  • Like
Reactions: The_Answer1313
I'd just let Vegas determine it.

Honestly it's all a bit silly when one team is 6 seeded and another team is 11 seeded and the 11 seed is the favorite when the game comes around.

So Vegas sees that the 11 seed is more talented.
Computer models that pretty much mirror the Vegas lines realizes this.

But a committee of humans somehow think that not only is the 6 seed better but the 6 seed is 5 seed lines better.
 
UNCW, that’s what is going to hurt Kentucky all year.
But Dook lost to a worse NET team in GA Tech 154, our lost was to 127 yet, both are Quad 3 losses but they are 17th in the NET we are 30th and they also have 3 losses.

I gave up on figuring out the NET awhile back.
 
UNCW is a bad loss and it will hurt us all season long. We can mitigate the damage but anyone who suggests that losing to an unranked, sub 100 team, at home doesn’t hurt is in pure straight denial.

These are likely the same posters that said Evansville wasn’t so bad after we lost that game. And yes, there were posters saying that at the time.
If we didnt have basically four of our top 10 players out for that one Id agree but we did so I dont. Its a throw away game and would be if it was any of the other blue blood but not UK as we are persecuted unmercifully all the time and its total BS.

Everyone on this board knows if Wagner, Ugonna, Z and a normal (not with 1st gm jitters and limited minutes after injury) Bradshaw played that Kentucky would not have lost that game. Thus, its a throw away, like it never happened. That's what the committee would do for Duke and UNC and Kansas. Why not Kentucky? Because we are the King, thats why, and they all hate us.
 
Last edited:
If we didnt have basically four of our top 10 players out for that one Id agree but we did so I dont. Its a throw away game and would be if it was any of the other blue blood but not UK as we are persecuted unmercifully all the time and its total BS.

Everyone on this board knows if Wagner, Ugonna, Z and a normal (not with 1st gm jitters and limited minutes after injury) Bradshaw played that Kentucky would not have lost that game. Thus, it’s a throw away, like it never happened. That's what the committee would do for Duke and UNC and Kansas. Why not Kentucky? Because we are the King, thats why, and they all hate us.
Maybe we win with all those guys. But we had PLENTY of firepower to take of business and we couldn’t.

I get your schtick but come on man. You shouldn’t need 10 first round picks to beat UNCW. Our second five should’ve beat them by 15-20 minimum.

I know Cal has lowered your standards but UNCW? On our home floor? Come on now. You’re being silly.
 
  • Like
Reactions: fatguy87
people keep saying "UNCW" "UNCW" "UNCW".
And yes that was a bad loss. The worst part is it was at home.

But look at it this way, right now in the NET, UNCW is right there with Ariz St & Okl St, and ahead of Penn St, FSU, UCLA, Ga Tech, Stanford, LSU, and far ahead of WVU, G-town, Oregon St, Maryland, Cal, UL, ND, Vandy, & DePaul.

So UNCW, according to the NET, is as good or better than:
- 2 Big East team
- 2 SEC teams
- 2 Big 10 teams
- 2 Big 12 teams
- 4 ACC teams
- 5 PAC 10 teams
That is over 20% of the P5 schools.
Exactly and they looked like a good team in that game. Yes we have had some head scratchers to teams that weren’t any good but they appeared to be a team that could be in the tournament. The numbers seems to show that they could in fact do that as well. We will have to follow them in conference.
 
We have some crappy teams in the conference eaccording to net rankings. Vandy is doing their usually sucking it up before being an en so if year tough team that can’t make the tournament. Lsu? Ark?
 
.....and yet in Kenpom, the SEC is currently the 3rd rated conference just behind the Big 12 and Big Ten. And not too far from the Big Ten.
 
  • Like
Reactions: fatguy87
The NET is a black box of sorts, but it isn't a total mystery with a light understanding of the inputs of other computer metrics.
 
ADVERTISEMENT
ADVERTISEMENT