2022-23 NET/Bracketology Discussion

[Kedsy]

Even this feels misleading, as it doesn't address the second of my two points (the era is totally different than 1997-2014, when we had much more experience and continuity). While true that it would be only better than 3 of the past 13 seasons, it would also be better than 3 of the past 7 seasons. Which is probably more of an apples to applies comparison than the 1997-2014 seasons are, given the youth and lack of continuity that we've dealt with since the "one-and-done factory" really took off with the 2015 team. So a 6-loss (.700 win %) season would be right in the middle of the pack for Duke in the freshman-dominated rotation era.

Maybe. The fewer seasons you include, the data becomes a lot less worthwhile. First, you talk about 3 of the past 7 seasons, when I can't see any reason to exclude 2015 from your freshman-dominated rotation theory (which would make it 3 of 8). Second, in two of the three seasons under .700, Duke's seasons were derailed by injuries, and in the other we didn't have the quality of recruit that we've generally had in the OAD era (and that we have this season). Not only that, in one of the seasons under .700 (2017), we had three upperclassmen and a sophomore in the starting lineup, and thus that season didn't have a "lack of continuity" problem at all (making the "under" 2 of 7).

Finally, what's the evidence that having a lot of talented freshmen in your rotation leads to more losses (the justification for excluding the previous 18 seasons)? In the past 39 seasons, Duke freshmen played 40% or more of the team's minutes eight times. The only two teams not to exceed a 70% winning percentage were 2021 (lower quality recruit, i.e., no top 5 or top 10 recruits and only two top 20 recruits) and 2016 (season derailed by co-captain Amile Jefferson's season-ending injury).

Season	%frosh	Win%	Top 5	Top 10	Top 20
2018	67.5%	0.722	2	3	4
2019	61.0%	0.778	3	3	4
2015	50.0%	0.833	1	2	3
2021	49.4%	0.500	0	0	2
2016	46.8%	0.611	1	1	3
2000	45.9%	0.938	1	2	3
2020	44.5%	0.750	1	1	2
2022	43.6%	0.800	1	1	3

FWIW, so far this season, freshmen have played 55.2% of the team's minutes and has three top 5, top 10, and top 20 recruits.

A sample of eight is obviously not enough to draw any firm conclusions (either way), but I don't see any evidence supporting the hypothesis that Duke teams with less experience win fewer conference games.

[MarkD83]

My rule of thumb: square the win % of road games; square root the win % of home games. A game like @Wake had roughly a 70% chance of victory on Torvik that is probably more like 49%.

You have a pretty smart thumb to do that math.

[CDu]

Maybe. The fewer seasons you include, the data becomes a lot less worthwhile. First, you talk about 3 of the past 7 seasons, when I can't see any reason to exclude 2015 from your freshman-dominated rotation theory (which would make it 3 of 8). Second, in two of the three seasons under .700, Duke's seasons were derailed by injuries, and in the other we didn't have the quality of recruit that we've generally had in the OAD era (and that we have this season). Not only that, in one of the seasons under .700 (2017), we had three upperclassmen and a sophomore in the starting lineup, and thus that season didn't have a "lack of continuity" problem at all (making the "under" 2 of 7).

Firstly, that would make it "2 of 6", not "2 of 7." You don't get to delete an observation from the numerator and not the denominator. So again, 4th of 6 is right in the middle (along with 3rd of 6).

Secondly, I think that team still had a pretty high degree of continuity issues. There were still a bunch of freshmen minutes (~33% of total minutes), and the freshman availability wasn't ideal at that (Tatum and Giles with injuries early on). A veteran player and team can overcome those issues. But combining injury with inexperience is challenging both for the young player and the young team.

Thirdly, I don't agree that the data are less worthwhile. Yes there are fewer data points, but comparing to an era when freshmen played way fewer minutes just doesn't seem useful/meaningful to me. If you want to extract those seasons from yore when freshmen dominated the minutes? Feel free to include those. But comparing this year to a dataset which includes 1997, 1999, 2001, 2002, 2004, 2005, 2006, 2008, 2009, 2010, and 2011, and 2012 (which were veteran-dominated teams) feels way less worthwhile to me.

Smaller sample sizes aren't ideal, but they can be more useful than larger samples that introduce an entirely different population of data.

Finally, what's the evidence that having a lot of talented freshmen in your rotation leads to more losses (the justification for excluding the previous 18 seasons)? In the past 39 seasons, Duke freshmen played 40% or more of the team's minutes eight times. The only two teams not to exceed a 70% winning percentage were 2021 (lower quality recruit, i.e., no top 5 or top 10 recruits and only two top 20 recruits) and 2016 (season derailed by co-captain Amile Jefferson's season-ending injury).

Season	%frosh	Win%	Top 5	Top 10	Top 20
2018	67.5%	0.722	2	3	4
2019	61.0%	0.778	3	3	4
2015	50.0%	0.833	1	2	3
2021	49.4%	0.500	0	0	2
2016	46.8%	0.611	1	1	3
2000	45.9%	0.938	1	2	3
2020	44.5%	0.750	1	1	2
2022	43.6%	0.800	1	1	3

FWIW, so far this season, freshmen have played 55.2% of the team's minutes and has three top 5, top 10, and top 20 recruits.

A sample of eight is obviously not enough to draw any firm conclusions (either way), but I don't see any evidence supporting the hypothesis that Duke teams with less experience win fewer conference games.[/QUOTE]

The 40% seems a pretty arbitrary cutoff. It excludes 2017 (33%), which would still be well above the average for Duke over the last 25 or 39 years). I'd guess that 2007 (36%), 2003 (36%), and 1998 (35%) would add to the list of relatively freshman-heavy, lower-continuity teams. And perhaps not surprisingly, 2 of those 3 seasons would add to the "under .700 win% club. So 5 of the 12 freshmen-heavy teams (33% or more minutes by freshmen) had sub-.700 conference records, vs just 2 of the other 13 teams. Caveat being that there might be some other 30+% freshman teams in there, but I couldn't remember any obvious ones like 1998, 2003, and 2007.

I haven't done the math, but I'd suspect there is a noticeably negative correlation between % of minutes by freshmen and conference win %. Not a 100% correlation, but I'd expect it to be there.

[MChambers]

I love it when Kedsy and CDu talk stats!

[Kedsy]

Firstly, that would make it "2 of 6", not "2 of 7." You don't get to delete an observation from the numerator and not the denominator. So again, 4th of 6 is right in the middle (along with 3rd of 6).

I added the 2015 season, which you ignored for no apparent reason, and subtracted 2017. So, to me at least, that looks like "2 of 7."

Secondly, I think that team still had a pretty high degree of continuity issues. There were still a bunch of freshmen minutes (~33% of total minutes), and the freshman availability wasn't ideal at that (Tatum and Giles with injuries early on). A veteran player and team can overcome those issues. But combining injury with inexperience is challenging both for the young player and the young team.

That team was a veteran team, with (as I said above) two seniors, a junior, and a sophomore in the starting lineup. I'd note that the 2006 team also had approximately 33% of its minutes played by freshmen (32.5%, just one percent less than 2017), and that team went 14-2 in conference play.

Smaller sample sizes aren't ideal, but they can be more useful than larger samples that introduce an entirely different population of data.

I agree with your assertion in general, but since we haven't seen any evidence at all that less experienced teams win fewer ACC games, I don't agree that in this case the larger sample introduces an entirely different population of data. Believing something to be true (that less experienced teams win fewer games, for example) doesn't actually make it true.

A major problem with small samples, as you know, is that one or two flukes or outliers can skew the results. This particular small sample is a great example of that with one season with a very different freshman recruiting rank profile than the others, and a couple seasons with major injury issues. And those three seasons were the only teams in the sample with a conference winning percentage under 70%.

I haven't done the math, but I'd suspect there is a noticeably negative correlation between % of minutes by freshmen and conference win %. Not a 100% correlation, but I'd expect it to be there.

Here's the data for the past 35 seasons:

Year	%frosh	Cnf Ws	Cnf Ls	Pct
2018	67.5%	13	5	0.722
2019	61.0%	14	4	0.778
2015	50.0%	15	3	0.833
2021	49.4%	9	9	0.500
2016	46.8%	11	7	0.611
2000	45.9%	15	1	0.938
2020	44.5%	15	5	0.750
2022	43.6%	16	4	0.800
2003	37.3%	11	5	0.688
2007	36.5%	8	8	0.500
1998	35.2%	15	1	0.938
2017	33.5%	11	7	0.611
2006	32.5%	14	2	0.875
1995	31.5%	2	14	0.125
1990	28.9%	9	5	0.643
2008	26.4%	13	3	0.813
2012	23.8%	13	3	0.813
2013	23.6%	14	4	0.778
2014	19.8%	13	5	0.722
1991	17.4%	11	3	0.786
2001	16.6%	13	3	0.813
1994	16.5%	12	4	0.750
2004	15.8%	13	3	0.813
1997	15.5%	12	4	0.750
2010	14.8%	13	3	0.813
1989	13.5%	9	5	0.643
2005	12.3%	11	5	0.688
1996	11.8%	8	8	0.500
2011	10.9%	13	3	0.813
2009	10.5%	11	5	0.688
2002	9.7%	13	3	0.813
1999	9.2%	16	0	1.000
1988	8.7%	9	5	0.643
1992	8.0%	14	2	0.875
1993	7.0%	10	6	0.625

You're better at this sort of thing than I am, but I don't see a correlation as you describe. What I can see is that of the ten most experienced teams on this list (or at least the ten teams with the fewest freshmen minutes), only four of ten had a winning percentage greater than 70% (obviously 6 of 10 were worse).

[CDu]

A major problem with small samples, as you know, is that one or two flukes or outliers can skew the results. This particular small sample is a great example of that with one season with a very different freshman recruiting rank profile than the others, and a couple seasons with major injury issues. And those three seasons were the only teams in the sample with a conference winning percentage under 70%.

It's certainly true that injuries matter, but I think the data still suggests that having an inexperience/continuity problem leads to worse conference record.

The key to this sort of thing is having a clear hypothesis. Once one has that, then it is about identifying the appropriate dataset to test the hypothesis.

The hypothesis in this case is that teams that have an inexperience/continuity problem do worse. So you need to identify teams for which continuity/inexperience are a potential problem. So it's useless to include data from years/eras in which EVERYONE played all upperclassmen. Because in that sample, there is no distinction between teams in terms of experience; just talent differences. So including them introduces noise, because experienced teams aren't really experienced relative to their environment. Conversely, if every team played mostly freshmen, then including data points from that era would be somewhat meaningless as well.

It is no surprise to me that there is a lot of noise in the results from the most experienced Duke teams. Because most of those data points happened back when EVERYBODY was experienced. And 5 of those 6 under-.700-but-veteran teams played in an era where being veteran was not an advantage - it was the standard. In that era, it wasn't about being veteran; it was about being good veterans.

Here's the data for the past 35 seasons:

Year	%frosh	Cnf Ws	Cnf Ls	Pct
2018	67.5%	13	5	0.722
2019	61.0%	14	4	0.778
2015	50.0%	15	3	0.833
2021	49.4%	9	9	0.500
2016	46.8%	11	7	0.611
2000	45.9%	15	1	0.938
2020	44.5%	15	5	0.750
2022	43.6%	16	4	0.800
2003	37.3%	11	5	0.688
2007	36.5%	8	8	0.500
1998	35.2%	15	1	0.938
2017	33.5%	11	7	0.611
2006	32.5%	14	2	0.875
1995	31.5%	2	14	0.125
1990	28.9%	9	5	0.643
2008	26.4%	13	3	0.813
2012	23.8%	13	3	0.813
2013	23.6%	14	4	0.778
2014	19.8%	13	5	0.722
1991	17.4%	11	3	0.786
2001	16.6%	13	3	0.813
1994	16.5%	12	4	0.750
2004	15.8%	13	3	0.813
1997	15.5%	12	4	0.750
2010	14.8%	13	3	0.813
1989	13.5%	9	5	0.643
2005	12.3%	11	5	0.688
1996	11.8%	8	8	0.500
2011	10.9%	13	3	0.813
2009	10.5%	11	5	0.688
2002	9.7%	13	3	0.813
1999	9.2%	16	0	1.000
1988	8.7%	9	5	0.643
1992	8.0%	14	2	0.875
1993	7.0%	10	6	0.625

You're better at this sort of thing than I am, but I don't see a correlation as you describe. What I can see is that of the ten most experienced teams on this list (or at least the ten teams with the fewest freshmen minutes), only four of ten had a winning percentage greater than 70% (obviously 6 of 10 were worse).

First, thanks for collecting the data. That's a lot of work that I didn't want to do, so I appreciate it. And I apologize in advance for now immediately suggesting that 10-20 years worth of those data are not useful. But to my point above... I unfortunately think the data from 1988-2006 are mostly irrelevant. I could see an argument for maybe including 1997 or 1998 to 2006 (because by then early entry was becoming common during that era), but I really think the one-and-done era is the most appropriate dataset for this.

When you limit the data to the one-and-done era (which is probably the most appropriate window), we get a correlation of -0.24, which is a pretty substantial correlation for real-world data over a small sample size. For a 100% nonfreshman team, we would have an expected win % of 78.1%; with a 50% freshman team, we would expect it to be 71.7%.

If we extend the limit to 1997, the correlation remains -0.24. In that scenario, a team with 100% non-freshmen minutes would be expected to win 81.5% of their ACC games; a team with 50% freshman minutes would be expected to win 73% of their games.

Again, it's not perfect, as there are always a ton more variables. Honestly, I'm not thrilled with the data point of % of minutes freshman as it omits two key components: opponents' experience/continuity, and team game experience (not just years of college). But we simply aren't going to have the data for the rest of college bball, and it gets way to complicated to consider other nonfreshman inexperience. But I suspect that in the aggregate the data we have is slightly underestimating the impact of inexperience/continuity deficits. And certainly it is underestimating the impact of experience deficits when data from the 80s and 90s are included.

[CasualObserver]

While I do understand the experience component is a strongly logical element in foretelling outcomes (e.g. ranking UNC #1 to start the season) there’s always the “emotional” element of that particular day it seems.

Last year was heavily freshmen and it was often in evidence. But, then there was do or die time against some very experienced tourney teams and the freshmen came out on top.

While we had no ability to beat Edey at his game this year, I don’t think we would definitely lose again to Purdue without Edey…..and I think we are good enough to beat Kansas on another night.

So, to me, Duke’s ACC season depends on the “minds” (for lack of a better word) that show up that given day. More experience does logically suggest less chance of a ‘wobbly mind’ day in and day out, but I don’t think any ACC opponent this year is an “Edey”….and not any better than Kansas overall. Depends on ‘what night/what mind’ we play them.

A ‘good mind’ Duke 22/23 might end up as good as 3 close game losses. A 19-turnovers Duke might not win any of the bolded font games shown above.I just don’t think we’ll get the answers with prior stats.

[dlmzzz]

For what it is worth, here are the chances Torvik's model currently gives each ACC team of making the tourney.

TEAM	Current Rank	Projected Record	Chance of Making Tourney	Chance of Autobid	Proj. Seed
Duke	12	24-7 (15-5)	99.8%	27.9%	3.4
Virginia	20	23-6 (15-5)	99.7%	19.7%	3.6
Virginia Tech	18	25-6 (15-5)	95.4%	18.8%	7
North Carolina	30	20-11 (13-7)	85.2%	12.6%	7.3
Miami FL	49	22-9 (12-8)	76.0%	6.4%	8.7
NC St.	41	20-11 (11-9)	50.8%	7.2%	9.5
Wake Forest	69	18-13 (10-10)	9.7%	2.6%	10.6
Pittsburgh	89	17-14 (11-9)	3.3%	1.4%	10.9
Clemson	88	17-14 (10-10)	1.6%	1.3%	11.6
Notre Dame	93	17-14 (9-11)	0.9%	0.9%	12.4
Syracuse	100	16-15 (9-11)	0.7%	0.7%	12.1
Georgia Tech	106	16-15 (8-12)	0.7%	0.7%	13.1
Florida St.	167	8-23 (6-14)	0.1%	0.1%	15.8
Boston College	198	10-21 (4-16)	0.0%	0.0%	0
Louisville	282	4-27 (2-18)	0.0%	0.0%	0

A couple things stand out to me.

- Torvik's model agrees with those in this thread who suggest that 5 ACC teams are currently "on track" to make the tourney, with a 6th team (NC State) on track for the bubble.

- Torvik's model doesn't give the ACC teams projected to win 17-18 games much chance of making the tourney. Therefore, the middle of the pack ACC teams (Wake, Pitt, Clemson, ND, Syracuse, Ga Tech) are probably going to have to exceed expectations by 2-3 wins to have a realistic chance of making the tourney. I haven't run the probabilities, but my gut tells me that at least one of these teams is likely to end up with 19-20 wins, putting themselves in the conversation.

- The projected seed for Va Tech seems odd to me. Torvik projects Va Tech to finish the regular season at 25-6, yet he projects them with a 7 seed. It is hard to imagine a 25 win ACC team (with wins vs the Big 10 and Big 12) ending up with a 7 seed.

Interesting. I was looking at KenPom's info, which is fairly similar :
UVA 9
Duke 13
UNC 25
VA Tech 28
Miami 43
NC State 56
Clemson 72
Syracuse 77
Pittsburgh 78
Wake 79
ND 90

I was focused more on resume up to this point in the season.
Bad loss = anything not Tier 1 or 2
Non-con opportunities = a potential tier 1 or tier 2 win
Away and Neutral court matter in all of these

Team	Tier 1 Record	Tier 2 Record	Bad Losses	Non-con opportunities
Virginia	3-0	0-0	<none>	Houston (2)
Duke	3-2	0-0	<none>	<none>
uNC	0-4	1-0	<none>	Ohio State(20), Michigan(44)
VA Tech	3-0	0-1	<none>	<none>
Miami	1-1	3-0	<none>	<none>
NC State	0-2	2-0	<none>	Vanderbilt (99)
Clemson	0-1	1-0	South Carolina(168), Loyola Chicago(124)	<none>
Syracuse	0-2	2-0	Colgate(113), Bryant(170)	<none>
Pittsburgh	2-2	0-1	VCU (125)	<none>
Wake Forest	1-1	0-1	Loyola Marymount(121)	Rutgers (32)
Notre Dame	0-0	1-1	St Bonaventure(127)	Georgia (109)

Looking at it this way, Syracuse and Clemson would have the most ground to make up in conference play to get onto the bubble. But really, all of these look like typical bubble resumes if they don't hurt themselves in conference.
Miami and NC State are actually in decent shape if they can get things done in conference.
As for UNC, it doesn't help to play a strong schedule if you don't win any of them. We are all BIG fans for the rest of this month, Scott!

[Kedsy]

The hypothesis in this case is that teams that have an inexperience/continuity problem do worse. So you need to identify teams for which continuity/inexperience are a potential problem. So it's useless to include data from years/eras in which EVERYONE played all upperclassmen. Because in that sample, there is no distinction between teams in terms of experience; just talent differences. So including them introduces noise, because experienced teams aren't really experienced relative to their environment. Conversely, if every team played mostly freshmen, then including data points from that era would be somewhat meaningless as well.

I understand what you're saying, except even in the "OAD era," most of Duke's ACC opponents didn't lose many OADs:

ACC OADs Since 2007:

Duke: 24
UNC: 5
FSU: 4
GaTech: 3
Syracuse: 3 (since 2014)
NCSU: 2
N Dame: 1 (since 2014)
Wake: 1
Miami: 1
Pitt: none (since 2014)
L'Ville: none (since 2015)
Md: none (2007 to 2014)
Virginia: none
BC: none
VaTech: none
Clemson: none

So, I guess it's a little different from the '80s and '90s, but not all that much. Duke had less available experience in most years after 2010, but their opponents were pretty close to the same. Which could mean that including the larger sample might not be as useless as you surmised.

Also, regarding your bolded comment above, I would argue that talent differences are still the biggest deciding factor. Experience differences might be a factor, but not nearly as much as talent. Of the Duke teams that played a lot of freshmen, the more talented teams were much more likely to have good winning percentages. I know that sounds obvious, but you're lumping more talented and less talented freshman classes into the same sample when you calculate your correlation, which I think skews your conclusion.

[CDu]

I understand what you're saying, except even in the "OAD era," most of Duke's ACC opponents didn't lose many OADs:

ACC OADs Since 2007:

Duke: 24
UNC: 5
FSU: 4
GaTech: 3
Syracuse: 3 (since 2014)
NCSU: 2
N Dame: 1 (since 2014)
Wake: 1
Miami: 1
Pitt: none (since 2014)
L'Ville: none (since 2015)
Md: none (2007 to 2014)
Virginia: none
BC: none
VaTech: none
Clemson: none

So, I guess it's a little different from the '80s and '90s, but not all that much. Duke had less available experience in most years after 2010, but their opponents were pretty close to the same. Which could mean that including the larger sample might not be as useless as you surmised.

Also, regarding your bolded comment above, I would argue that talent differences are still the biggest deciding factor. Experience differences might be a factor, but not nearly as much as talent. Of the Duke teams that played a lot of freshmen, the more talented teams were much more likely to have good winning percentages. I know that sounds obvious, but you're lumping more talented and less talented freshman classes into the same sample when you calculate your correlation, which I think skews your conclusion.

It isn’t JUST one-and-dones. The one-and-done era has led to a lot more early entry in general. Because one-and-dones are happening, players’ draft stock decreases the longer they stay, whereas 20-30 years ago nobody batted an eye about drafting a senior. So teams simply don’t have nearly as many minutes being played by seniors in the 2010s and 2020s (although we are getting a bit of inflation in the 2020s countering this due to COVID). And generally speaking, those seniors are inherently not as good as the seniors of yore (because the really good players have already gone pro).

That said, I am quite sure that talent plays a huge part. I never intended to suggest otherwise. I am just saying that experience/continuity matters too. In the olden days, it was only talent difference that mattered, because every team was similarly old and continuous. But now, experience and continuity matters too.

[Southgate0809]

Latest bracketology has us as a 3-seed, which is fair when you look at all the teams ahead of us with 0-1 losses. UVA is at a 1, which is also fair given their wins and lack of losses. Purdue is the overall 1. Kansas got a 2.

UNC is at a 9. Ouch. Still 2 tough games coming for them before the end of the calendar year.

What do you think is the lowest seed a team can be given without winning their conference tournament? I assume a lot of those 14's, 15's, and 16's go to automatic bids. Once you're around 12 or 13, is that pretty much being on the bubble?

Edit: Teams currently listed as the Last 4 In are all 11 seeds.

[Kedsy]

Latest bracketology has us as a 3-seed, which is fair when you look at all the teams ahead of us with 0-1 losses. UVA is at a 1, which is also fair given their wins and lack of losses. Purdue is the overall 1. Kansas got a 2.

UNC is at a 9. Ouch. Still 2 tough games coming for them before the end of the calendar year.

What do you think is the lowest seed a team can be given without winning their conference tournament? I assume a lot of those 14's, 15's, and 16's go to automatic bids. Once you're around 12 or 13, is that pretty much being on the bubble?

Edit: Teams currently listed as the Last 4 In are all 11 seeds.

The last four are 11-seeds because the "First Four" play-in games generally involve 11-seeds. Last year, one of the play-in games was among 12-seeds, so Indiana was a 12-seed. In 2021, even though the play-in games only involved 11-seeds (and 16-seeds, obviously), both Georgetown and Oregon State were 12-seeds. Going back beyond that, it doesn't happen every year, but the worst Power 6 at-large teams as well as non-P6 at-large teams seem to be regularly seeded #12. I didn't go further back than 1996, but back in 1999, at-large team Oklahoma was seeded #13. A few non-P6 at-large teams have also been seeded #13 or lower: BYU and Iona were both #14 in 2012 (so I guess #14 is the actual answer to your question); Air Force and Bradley were both #13 in 2006; UTEP was #13 in 2004.

[scottdude8]

Latest bracketology has us as a 3-seed, which is fair when you look at all the teams ahead of us with 0-1 losses. UVA is at a 1, which is also fair given their wins and lack of losses. Purdue is the overall 1. Kansas got a 2.

UNC is at a 9. Ouch. Still 2 tough games coming for them before the end of the calendar year.

What do you think is the lowest seed a team can be given without winning their conference tournament? I assume a lot of those 14's, 15's, and 16's go to automatic bids. Once you're around 12 or 13, is that pretty much being on the bubble?

Edit: Teams currently listed as the Last 4 In are all 11 seeds.

That's about on par with the "consensus" at BracketMatrix, which has us as the #1. That said, the "consensus" is less so since fewer people even bother with Bracketology at this point, haha. FWIW, The consensus is higher on UNC than Lunardi's latest (a 7 seed), but that's likely because these early-season brackets probably have some major pre-season skew left in them.

As for the Last 4 In, as Kedsy mentioned that can vary a bit from year to year depending on how the auto-qualifiers are viewed by the committee. I don't believe there's any "rule" that says the Last 4 in have to be at any seed line... so if the committee believed that the champ from conferences like the Colonial, WAC, Missouri Valley, etc. had a better resume than two of the Last 4 In, the play in game could move to the 12 line. I don't think there's even anything saying that it couldn't be the 13, but historically that has a near-zero chance of happening. A lot of that also depends on "bid stealers" too.

EDIT: I'm not sure why I bothered with this since Kedsy said everything I said above, but better. Serves me right for skimming, haha

[Southgate0809]

I don't understand why anyone would have to play a play-in game except for the 16 seeds. That's never made sense to me. Whether they're automatic bids or not, they are the lowest ranked teams. They should have to play the extra game. Further reward the 1 seeds by letting them play someone who just played 2 days before.

[scottdude8]

I don't understand why anyone would have to play a play-in game except for the 16 seeds. That's never made sense to me. Whether they're automatic bids or not, they are the lowest ranked teams. They should have to play the extra game. Further reward the 1 seeds by letting them play someone who just played 2 days before.

Many people certainly agree with you, but proponents of smaller schools and some "traditionalists" argued that following that path would further the divide between the power, multi-bid leagues and smaller conferences (while also taking away some of the "magic" of potential Cinderellas). Splitting the four play-in games was seen as the compromise solution.

That said, with continued rumblings that further expansion might be on the horizon, I imagine that this topic will be hotly debated once again in the coming years.

[Nugget]

The last four are 11-seeds because the "First Four" play-in games generally involve 11-seeds. Last year, one of the play-in games was among 12-seeds, so Indiana was a 12-seed. In 2021, even though the play-in games only involved 11-seeds (and 16-seeds, obviously), both Georgetown and Oregon State were 12-seeds. Going back beyond that, it doesn't happen every year, but the worst Power 6 at-large teams as well as non-P6 at-large teams seem to be regularly seeded #12. I didn't go further back than 1996, but back in 1999, at-large team Oklahoma was seeded #13. A few non-P6 at-large teams have also been seeded #13 or lower: BYU and Iona were both #14 in 2012 (so I guess #14 is the actual answer to your question); Air Force and Bradley were both #13 in 2006; UTEP was #13 in 2004.

I think one reason why it's becoming rarer to see at-large teems seeded below 12 is that conference realignment has moved so many of the former non-P6, non Mountain West/A-10/WCC, teams that might have earned higher seeds out of the smaller leagues. For example, in 2012, Murray St. (then in OVC) got a 6 seed, Memphis (then in C-USA) and Creighton (then in MVC) 8 seeds, VCU (then in CAA) a 12 and Davidson (then in So. Con a 13). With all the realignment, it seems like it's increasingly unlikely to see more than maybe 1 or 2 teams (max) out of "low-major" leagues like the OVC, CUSA, CAA, etc. be viewed by the Committee as warranting seeds above 12-13.

[Southgate0809]

Many people certainly agree with you, but proponents of smaller schools and some "traditionalists" argued that following that path would further the divide between the power, multi-bid leagues and smaller conferences (while also taking away some of the "magic" of potential Cinderellas). Splitting the four play-in games was seen as the compromise solution.

That said, with continued rumblings that further expansion might be on the horizon, I imagine that this topic will be hotly debated once again in the coming years.

It seems like that's double-doing it. I love giving automatic bids to conference winners, and I agree that allows for the magic of the tournament. But I don't think you need to allow for that twice and punish a team that fairly earned its 11 or 12 seed. It's like calculating how much someone owes in taxes based on a percentage of their income, so that people who make more money pay more in taxes. Then, creating a tax bracket system where even that percentage increases based on how much the income is.

Please, I am not inviting a discussion on how taxes should work. It was just the first comparison that came to mind. I'm sure many of you can think of good reasons our taxes work that way. I do not think those reasons apply to the NCAA tournament.

[Bluedog]

It seems like that's double-doing it. I love giving automatic bids to conference winners, and I agree that allows for the magic of the tournament. But I don't think you need to allow for that twice and punish a team that fairly earned its 11 or 12 seed. It's like calculating how much someone owes in taxes based on a percentage of their income, so that people who make more money pay more in taxes. Then, creating a tax bracket system where even that percentage increases based on how much the income is.

Please, I am not inviting a discussion on how taxes should work. It was just the first comparison that came to mind. I'm sure many of you can think of good reasons our taxes work that way. I do not think those reasons apply to the NCAA tournament.

Many people certainly agree with you, but proponents of smaller schools and some "traditionalists" argued that following that path would further the divide between the power, multi-bid leagues and smaller conferences (while also taking away some of the "magic" of potential Cinderellas). Splitting the four play-in games was seen as the compromise solution.

That said, with continued rumblings that further expansion might be on the horizon, I imagine that this topic will be hotly debated once again in the coming years.

I don't understand why anyone would have to play a play-in game except for the 16 seeds. That's never made sense to me. Whether they're automatic bids or not, they are the lowest ranked teams. They should have to play the extra game. Further reward the 1 seeds by letting them play someone who just played 2 days before.

Two reasons: 1. better ratings with big name schools for those early games. 2. Rewarding conference champions by not requiring even MORE of the 'small name' conference teams to have to do "play-in" games and not get the full NCAA experience. A middling team from a big conference still gets afforded the opportunity to be in the tournament (unlike before the expansion), but has to earn to make it to the First Round.

Personally, I like it.

[Southgate0809]

I think it's funny that gambling pools pretty much make the play-in games irrelevant. Everyone's deadline to have your bracket in is on Thursday morning. Making the deadline on Tuesday doesn't give everyone enough time.

[pcal6vb]

It seems like that's double-doing it. I love giving automatic bids to conference winners, and I agree that allows for the magic of the tournament. But I don't think you need to allow for that twice and punish a team that fairly earned its 11 or 12 seed. It's like calculating how much someone owes in taxes based on a percentage of their income, so that people who make more money pay more in taxes. Then, creating a tax bracket system where even that percentage increases based on how much the income is.

Please, I am not inviting a discussion on how taxes should work. It was just the first comparison that came to mind. I'm sure many of you can think of good reasons our taxes work that way. I do not think those reasons apply to the NCAA tournament.

I'd advocate that the play-ins should include NO conference champions. Make the 16s all automatically in, and have the play-ins be all 11- and/or 12-seeds that were on the bubble. Winning one's conference, even in a low-major, should be rewarded more than finishing 7th thru 9th in a major conference.

Of course, I also think that a 4-team college football playoff should include ONLY conference champions, even if consensus is that the second strongest (second most talented? second most likely to win?) team finished second in its own conference. If you don't win your conference, you don't get in. A 6-team playoff (or 12-team, as we're now headed toward), on the other hand, opens the door for at-large teams, after the conference champions of the strongest 5 or 6 conferences are all granted slots. I think 8 teams is the ideal: champions of the Power 5 conferences, plus 3 at large.