Championship Profiles

I've been working on a lot of stuff (three of which were college basketball related) since the two updates on the January QC. Of the three college basketball related projects, one of those has been an attempt to identify a National Champion from among the many contenders/pretenders (seems like mostly the latter). Last year was a fairly easy task: A two-horse race between GONZ or BAY (apologies to third-wheeling HOU, your numbers were never on the same level as these two). In all honesty, I was on BAY the whole year up until their Covid-pause, and the before-and-after change was too scary to go all-in on them. I liked them because BAY reminded me of an even-more consistent 2019 UVA and they were playing with a similar composition in a much-weaker 2021 field. This year, I keep doing mental gymnastics and still cannot identify a gold-medal winner. However, I still like this type of project because it allows me to go backwards in time when basketball was actual basketball (not the low-skill variant played today) and it allows me to pay homage to my mentor as he was fond of these types of analysis. So let's have a look.

The Profiles of Past Champions

Let's start with the details, then we'll make some interpretations of the results, and then we'll apply them to this year's crop of hopefuls. My starting year was 1998, which was my very first year of bracket picking.

I started with several factors (qualities that define a team):

1. Did they win a conference title (season or tournament or both)?
2. Who was their Point Guard (Eligibility Class)?
   
3. How essential/productive was their Point Guard? (Points per game, Assists per game)
• Some teams have two or more listed. I tried differentiate a true floor general based on wide deviations in starts, minutes played, points per game and assists per game.
• Where deviations were too small to identify one true point guard, the average of their totals were used to compare to solo-led teams. (It's still pretty impressive that two players can average similar totals as the solo-PGs.)
  
4. What was the role of the front-court (Offensive, Defensive, Both)?
5. How productive was their front-court (Points per game, Rebounds per game and/or Blocks per game)?
• Like the process for point guards, I tried to differentiate the three front-court roles based on starts, minutes, points, rebounds, blocks, and personnel.
• Typically, the personnel was the deciding factor as defensive front-courts had multiple bodies to throw at the opposing team's front-court. For example, 2010 DUKE had four post players to rotate in/out of games (that's 20 fouls to expend stopping easy points at the rim).
  
6. Did they participate in the previous year's tournament (Play-in game losers do not qualify due to equal comparisons to past tournaments lacking the play-in format) and what was their seed?
7. If they participated, how many wins did they collect in the previous year's tournament?
8. Did their coach qualify as a tournament contender (reached the E8 in any previous tournament)?

Are these the only 'shared attributes' of title winners? Certainly not, and I would point you to this article if you would like to know a few more of them. It's time to see the results.

There is a lot to unpack here, and I'll try to be as concise as possible.

1. Of the 23 NCs, eight won both the regular-season and tournament titles for their conference (or at least tied for a share of the regular-season), nine won only the regular-season title (or at least tied for a share), four won only the tournament title, and two won neither. Regular-season winners have 17 out of 23 titles, and of the six years without a regular-season winner, four of those came in 2006, 2011, 2014, and 2018 (all insane years for tournaments). 2013 marks the last year a dual-title winner also won the NC, and it also marks the last year that college basketball wasn't shackled by the officiating philosophy known as "Freedom of Movement."
2. Of the 23 NCs, three were won by freshman PGs, 2.5 by sophomore PGs, twelve by junior PGs, and 5.5 by senior PGS. 17.5 NCs go through Senior or Junior PGs, so experience matters when leading a team to a title. Of the 5.5 underclassmen that won NCs, all but two came 2006 and earlier, which happens to be the pre-one-and-done era of college basketball (The 2007 tourney was the first year that the NBA mandate all high school seniors be one-year removed from high school before being draft-eligible).
3. Of the production of point-guards, the minimum assists-per-game is 3.7 (1999 and 2007). I also computed another value (not shown) which totals the points that the PG is responsible. Since an assist leads to a direct basket (minimum two points), the formula (PTS + 2xAST) totals the minimum points for which the PG is responsible. The minimum value was 17.7 in 1999 (the second-lowest was 18.1 in 1998). One notable factor is the uptick in PG-scoring in 2009 and later years, which coincides with the extension of the 3-pt arc starting in 2009. In the years 1998-2008, only one PG achieved greater than 13.3ppg. In the years with the extensions (2009-2021), only three were below the mark of 13.3ppg, with two of three being freshman.
   
4. Of the front-court's role, it's hard to find any pattern predictive of NCs other than the coupling (consecutive years of the same quality) and the intermittent rotation (usually every twelve years, 2000-2002 and 2012-2014, theoretically the next rotation wouldn't be due until 2024). Pure Speculation: If we start a new rotation in 2024, then 2022 and 2023 would have to be a couple and likely a different one than what we've recently seen (either B or O). Either way, basketball is won and lost in the paint. You either need to score exceptionally well in the paint or defend it exceptionally well (kudos if you can do both). Nonetheless, the O/D/B identity is the least reliable in predictive terms.
5. Of the front-court's production, it is really dependent on the front-court's identity. With the exception of one year, most defensive front-courts produce less than 12ppg. Although it is not shown, these teams make-up this lack of front-court production with heavy guard scoring (usually at least three scorers on the perimeter). For example, 2021 BAY had Butler/Teague/Mitchell with 14.0+ ppg each, 2019 UVA had Hunter/Guy/Jerome with 13.7+ ppg each, and 2010 DUKE had Scheyer/Singler/Smith with 17.4+ppg each.
6. Of the 23 NCs, 20 participated in the previous year's tournament (the second-most reliable attribute on the table). Even better, all were given a seeding of 1- thru 6-seed in the previous year's tournament, which narrows down our list of contenders to 24. Of the three teams that didn't participate, all three had a national All-American (my nod to Pete's contenders/pretenders rules): 2003 SYR had 2nd-team AA Carmelo Anthony, 2011 CONN had 1st-team AA Kemba Walker, and 2014 CONN had 1st-team AA Shabazz Napier (who was also a contributor to the 2011 NC team). Despite my sincerest efforts, I have yet to find information regarding All-American status before the start of the tournament. As for 2021 BAY, it is 100% likely that 2020 BAY would have made the NCAA tournament, either as a 1- or 2-seed (based on numerous 2020 bracketologies).
   
7. Of the 19 NCs that played in a tournament game in the previous year, one won 6 games, two won 5 games, five won 4 games, two won 3 games, three won 2 games, four won 4 games, and two lost their only game. Again, we have another relatively unreliable factor, but it tends to follow a grouping pattern: About five years per group with one year in the group being an outlier. From 2015-2019, it was either one or zero wins, with one year being a previous five-win tournament. From 2007-2013, it was either three to six wins, with one year being a two-win year (three-win tournaments is essentially half-way to a title or an E8-run which also has significance in our work). From 2001 to 2006, it was either one or two wins, with one year being a four-win year. From 1996-2000, it was 4, 3, 5, 2, and 3, so three to five wins, with one year being a two-win year. I would assume that 2022 starts a new grouping cycle, especially with the 2021 NC having an unknowable 2020 win count.
8. Of the 23 NCs, twenty-one of them were led by a head coach who had went to the E8 or further before that season's title run. This is arguably the most-reliable factor in terms of predictability. Of the two that did not, the similarities are few and far between. Both inherited talented teams from coaching legends. Both teams were ranked in the pre-season AP poll. One was in his first year with the team and one was in his second. One came from a 4-bid conference and one came from a 5-bid conference. Those last two are stretches, so you can see how hard it is to find common ties. It will have to happen a few more times to find common predictable factors.
9. Of the 23 NCs, only two of them have two disqualifiers: 1998 UK (PG Production and E8 Coach) and 2014 CONN (Prev Tourney and E8 Coach). If you wanted to stretch it a little, you could also include 2003 SYR in this list (PG Class & Prev Tourney).

2022's Crop of Hopefuls

To begin with, I had no idea how I wanted to compile a list of contenders. For starters, of the 23 NCs on the list above, only 2003 SYR and 2011 CONN did not start the season ranked in the AP Top 25. If I had used that criteria, I would be missing teams like ARI, WISC, TXTC, USC, and MIST, all of whom are expected to receive high-seeds in the 2022 tournament. I also wanted to use the Top 25 Advanced Metrics teams, but there would be no way historically to replicate this approach if I wanted to do a year-by-year approach. I simply settled for a composite of the AP polls for the week of 1/17 and 1/24. If they appeared on both, they received attention.Yes, this leaves out the likes of TEX, HALL, and ALA. However, this list effectively amounts to all teams that could claim at least an 8-seed in the 2022 tourney (as no NC has had a seed worse than 8-seed). Finally, all data is for games completed by 1/23. Thus, no team has won a conference title (reg-season or tournament). Let's see what we've got.

Let's follow the same structure as before:

1. Conference champions have yet to be crowned at this point in the season.
2. Disqualifier: FR or SO PG. If a FR or SO was paired with a JR or SR, I allowed it as the only one to do it was 1999 CONN. NOTE: If ILL Curbelo assumes total control of PG role returning from his injury, this would add a disqualifier to ILL.
3. Disqualifier: I went with either <3.9APG or <20.0 PRF (Pts Responsible For). I also included a half-point disqualifier if the PG had single-digit scoring (Only two have done this, and none in the 3-pt extended era).
4. Disqualifier: If the Offensive Post was <10.0ppg or the Defensive Post didn't have three heavy-scoring guards.
5. Disqualifier: If the team did not participate in the previous tourney, they received a full-point. Of these teams, DUKE and UK are likely to produce consensus All-Americans in Banchero and Tshiebwe (exempting them from the disqualifier). I gave half-point disqualifiers if they went to the previous tournament but received worse than a 6-seed.
6. Disqualifier: I only gave half-point disqualifiers for this category, and I gave it for a 0-, 1-, or 2-win run in the previous tourney (NOVA should also be yellow). If the cycle is supposed to change, it should be anything but this result. One team (BAY) had a 6-win run in the previous tourney. With all respect to this year's team, only one team has won back-to-back on this list, and they returned 99% of everything from the previous team (2022 BAY did not). Again, this isn't a reliable factor, which is why I only went with a half-point disqualifier.
7. Disqualifer: I gave a full-point disqualifier if the coach has not previously been to the E8.

Here are the results color-coded for simplicity in viewing.

As always, thanks for reading my work. I hope to get another one of my three projects done in time to produce a 2nd article before the February QC Analysis.