Your points are all valid… no disagreement. Although, technically, that scene was a case of misinterpreting the data, not faulty data itself.
Bracketology
- Status
DeonThomas
- South Carolina
2024 Pomeroy College Basketball Ratings
Data through games of Friday, March 15 (6095 games)| Strength of Schedule | NCSOS | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Rk | Team | Conf | W-L | AdjEM | AdjO | AdjD | AdjT | Luck | AdjEM | OppO | OppD | AdjEM | ||||||||
| 1 | Houston | B12 | 30-3 | +33.83 | 120.3 | 12 | 86.5 | 1 | 63.4 | 348 | +.032 | 105 | +11.03 | 21 | 111.6 | 36 | 100.6 | 11 | -0.86 | 223 |
| 2 | Connecticut | BE | 30-3 | +31.71 | 126.9 | 1 | 95.2 | 16 | 64.5 | 328 | +.048 | 68 | +9.90 | 40 | 111.3 | 45 | 101.4 | 38 | -3.46 | 284 |
| 3 | Purdue | B10 | 29-3 | +29.50 | 125.6 | 2 | 96.1 | 21 | 67.5 | 169 | +.064 | 46 | +13.33 | 5 | 113.7 | 5 | 100.3 | 8 | +10.33 | 13 |
| 4 | Auburn | SEC | 25-7 | +28.40 | 120.8 | 10 | 92.4 | 5 | 69.8 | 57 | -.076 | 332 | +8.81 | 67 | 111.4 | 43 | 102.6 | 74 | +1.36 | 150 |
Resumes frequently don’t line up with efficiency metrics, and the committee’s consideration skews toward the resume when that happens. Tennessee + Creighton + St. Mary’s last year, Loyola in 2021, etc. I’m not arguing whether that’s right or wrong—different discussion—but Auburn will not sniff the 3 line.
Punesguy
- Ft. Collins, CO
Thank goodness for the BU era! We no longer wonder if we’re going to dance or not. We’re debating over a 3 versus 4 seed. Let’s go out today a beat a challenging Nebraska team and then surprise PU in the finals! Then we can debate on a 2 versus 3 seed!
With BU’s recent performance in the transfer portal and U of I’s NIL funds we should always attract strong talent.
Enjoy the ride my Illini brothers and sisters … I feel it’s going to be an exciting several weeks!
With BU’s recent performance in the transfer portal and U of I’s NIL funds we should always attract strong talent.
Enjoy the ride my Illini brothers and sisters … I feel it’s going to be an exciting several weeks!
Yes, but the selection committee is not likely to have much skill in knowing when (and how much) to go against the data.
When they go against it and betting markets don't, I'll side with betting markets & data (like in the case of Auburn).
While betting markets like Houston over Purdue, they like UConn even more. So I'd have no problem with the committee putting UConn first overall despite computers liking Houston, but I don't see a good argument for Purdue.
Because they were wrong
I'm sure some amount of recency adds skill to a ranking algorithm.
With NCAA BB, though, I wonder if the lack of late season inter-conference play is an obstacle. You'd have to make sure the last non-conference games don't skew the relative weighting of conferences.
I'm not sure what your point is?
Houston was a 5 seed 2 years ago and entered the NCAA tournament as #4 on KenPom as well.
Auburn has ONE Q1 win. Their win yesterday just got them to the 4 line. Beating Florida/Miss St./Texas A&M is not going to get them to jump another 3 or 4 teams.
Last edited:
Much of the difference in the computer rankings vs committee decisions is due to the committee having different goals than the predictive systems. They want to encourage early season road games between the better teams for example. Or they want an entertaining tournament so CBS keeps writing the big checks.
It's your opinion that the committee doesn't have skill in knowing when and how much to go against the data. But you're splitting hairs about the #1 overall seed when it doesn't matter whatsoever. Each of the top 3 teams are going to get the region they want.
But you need to look at the actual resumes instead of computers and Vegas (Betting patterns do influence lines to some degree too. Even opening lines... it's not just pure data.) Houston and UConn's non-conference strength of schedules are in the 200s. Purdue's is 13th.
Houston's non-conference wins over NCAA tournament teams: Dayton and Texas A&M
UConn's non-conference wins over NCAA tournament teams: Texas, UNC, and Gonzaga
Purdue's non-conference wins over NCAA tournament teams: Gonzaga, Tennessee, Marquette, Alabama, Arizona, and even Samford (by 53 points)
Last edited:
Looked at the bracket updates today and here's what I found:
-Lunardi: Has us a 3 seed in the South bracket facing 14 seed Akron in the first round in Pittsburgh. 6/11 matchup in that portion of the bracket is Florida vs. play-in game winner between St. John's and New Mexico.
-Palm: Has us a 3 seed in the East bracket, also facing 14 seed Akron in the first round in Omaha. 6/11 matchup in that portion of the bracket is Nevada/Colorado.
Looks like Bracket Matrix still has us as the #2 4 seed but it doesn't appear that it has been updated for today.
-Lunardi: Has us a 3 seed in the South bracket facing 14 seed Akron in the first round in Pittsburgh. 6/11 matchup in that portion of the bracket is Florida vs. play-in game winner between St. John's and New Mexico.
-Palm: Has us a 3 seed in the East bracket, also facing 14 seed Akron in the first round in Omaha. 6/11 matchup in that portion of the bracket is Nevada/Colorado.
Looks like Bracket Matrix still has us as the #2 4 seed but it doesn't appear that it has been updated for today.
I agree we're splitting hairs in the top overall seed case, but not in general. Auburn is a more dramatic example where the anticipated committee decision deviates significantly from computers and Vegas. If it were that easy to out-smart computers and Vegas, more people would get rich doing it.
As another post said, the committee has different goals (i.e. revenue during the tournament and influencing non-conference scheduling), so I wouldn't expect them to exhibit such skill even if they possessed it.
Houston beat a 1 seed that year, so the committee's decision to overrule the computers didn't work out well.
ChiefGritty
- Chicago, IL
The truth of it at this point is that we may already have a 3 seed or we might be a 4 even if we won out. Really fine margins between a bunch of teams in that part of the S-curve, and no real earth-shattering wins left to get with the BTT final too late for consideration.
I do think we've pretty much eliminated the possibility of a 5, which was a bigger threat than people were letting on a week ago, imho.
1 seeds lose every year.
Also in 2022, San Francisco was #21 on KenPom heading into the NCAA tournament and a 10 seed. They lost in the 1st round to 7 seed Murray State who was #27 on KenPom before the tournament.
Pointing to a single game is a very poor argument.
Also in 2022, San Francisco was #21 on KenPom heading into the NCAA tournament and a 10 seed. They lost in the 1st round to 7 seed Murray State who was #27 on KenPom before the tournament.
Pointing to a single game is a very poor argument.
The computers are still making educated guesses. Auburn has beaten 1 team in the top 29 of KenPom. They have lost to 5 such teams.
I'm a stats major of sorts and even I would have a hard time relying on data that told me a team like that was the 4th best team in the country. I'd want more data.
I'm a stats major of sorts and even I would have a hard time relying on data that told me a team like that was the 4th best team in the country. I'd want more data.
201154JC
- Rockford, IL
We can produce statistics until the cows come home and for me, the one that matters most is, we have been
in every game we've played this year right up until the end in most cases. Not a single double digit loss and no bad
beats. How many teams ahead of us can make that claim? This team is special in that respect. I admire them for their
fight.
in every game we've played this year right up until the end in most cases. Not a single double digit loss and no bad
beats. How many teams ahead of us can make that claim? This team is special in that respect. I admire them for their
fight.
Illini2010-11
- Sugar Grove
That is why there is always the need for data analysts to detect outliers, etc. I never want a system solely beholden to computers alone, and I am a person who loves as much data as possible.
I think teams, like Purdue, should be rewarded for playing such difficult non-conference schedules compared to teams that fill their non-conference schedule with many cupcakes.
Can the NET system formula be tweaked...maybe, we simply do not have enough info available. However, I think it is great that the committee uses a blended approach to selection and seeding. It is best practice, IMO.
OrangeBlue98
- Des Moines, IA
The bottom line is that, once again, humans need to provide context behind the data to make the best decisions. If we are not going to bother with context, then just develop an algorithm to seed teams and don't even bother with a selection committee.
Data is critical. Data without context is noise.
DeonThomas
- South Carolina
NCAA Selection Committee Rankings
(Through games of Feb 16, 2024)
(Through games of Feb 16, 2024)
| RANK | SCHOOL |
|---|---|
| 1 | Purdue (No. 1 MW) |
| 2 | UConn (No. 1 EAST) |
| 3 | Houston (No. 1 SOUTH) |
| 4 | Arizona (No. 1 WEST) ....moving down to a #2 seed |
| 5 | North Carolina ..............moving up to a #1 seed |
| 6 | Tennessee ......................6-2 since |
| 7 | Marquette ........still a #2 seed |
| 8 | Kansas ..............3-4 since (HOU & BAY & BYU & CIN) |
| 9 | Alabama ...........4-4 since (KY, TN, FLA and FLA) |
| 10 | Baylor ..................5-4 since (HOU, BYU, IOWA ST & TTU) |
| 11 | Iowa State ...........moving up to a #2 seed |
| 12 | Duke ...............5-3 since (WF & NCState & UNC) |
| 13 | Auburn ...............5-2 since (TN & KY) |
| 14 | San Diego State .............dropping out. bad losses. 5-3 since (Boise, UNLV and Utah St) |
| 15 | Illinois ................6-2 since (PUR & PSU) PASSING KANSAS, ALABAMA AND BAYLOR) |
| 16 | Wisconsin ............dropping out. |
No one said this. We always needed to make the BTT Final and get some help. Since we got a lot of help, we could still sneak in as a 3 with a loss today, but that would be pushing the envelope.
Auburn at #4 in KenPom is an outlier even among computers, so I'm not looking at that either. Avg of various computer ranks is more like 8/9, and Vegas has them around #7. To go against that consensus (i e. not even a 3 seed) to me would require some very unusual "contextual" information or a different motivation, otherwise it sounds like cherry picking to make a narrative.
Last edited:
- Status