Happy New Year! In the real world, it's the beginning of a new year, but in the college basketball world, almost half of the regular season is complete. This point in the season allows us to examine the current quality of college basketball, and it gives us the first prospective look into the quality of the 2023 tourney. Without further ado, let's take a look into what March could look like.
Current Quality (Vs NOV-Present Max/Min Curves)
There's a lot to unpack here. The blue line is the current quality curve for all games finished Sun Jan 1, 2023. The green and red lines are the maximum and minimum quality values across the range of dates from Nov 7, 2022 to Jan 1, 2023 for each position on the curve. Let's break this QC into groups.
- The #1 and #4 spots on the curve are the only two spots playing at or close to their maximum quality. These maximum values occurred on Dec 28 and Dec 21 (very recent which is very good). The date of the minimum values are Nov 15 for the #1 spot (very distant from today and from the max date, both very good signals) and Dec 12 for the #4 spot (very recent from today and from the min date, not very good as it suggests a temporary spike or a huge rotation of teams occupying that spot). Further investigation proves the former hypothesis, as PUR spiked up in quality points and TEX spiked down in quality points, and neither are occupying this spot.
- The #2 and #3 spots (along with a few other spots) are playing close to the mid-point of the max and min range (#2 below it and #3 above it). The maximum values occurred on Dec 21 and Dec 10 (a little less recent than #1 and #4, which is concerning). The date of their minimum values were Nov 26 and 27, respectively (earlier than their maximum values, which is a good sign, but still too close to their max dates to be comfortable).
- From the #5 to the #36 spots, the story is a lot more frightening. Of the 32 spots in this range of the QC, only five are at or above the 40th percentile of the max-min range: The #8, #13, and #24-26 spots. Last year, the majority of the teams in this range were playing at their maximum level, which at the time, I thought was a good sign. Then, Feb came along but the quality improvement didn't come with it, and it stayed that way through Mar. For this reason, I'm going to reserve judgement over this bottom-of-the-barrel status.
- From the #37-#50 spots, the story is better than the previous range. Of the fourteen spots in this range of the QC, only four are below the 40th percentile of the max-min range: The #39, #40, #45 and #46. However, none of the teams in this range are above the 55th percentile of the max-min range. In simple terms, most of these spots had their maximum values in early Nov (which is usually based on predictive comps) and are slowly regaining to that quality. It's actually not a bad thing that teams in this range are not at their maximum because when the back of the curve is strong, it implies weak tournament quality and a high number of upsets.
- I don't want to make excuses for the lack of quality, especially in the third group, but there have been a lot of notable injuries and precipitous drops in quality of teams with these notable injuries. IND, UNC, CREI, BAY, and ARK have seen important players miss games or minutes do to injury. Only ARK has been able to maintain equivalent quality ratings throughout the season, but the steep decline in quality points of the other four teams is unlike the rest of the QC, and maybe injuries had something to do with it.
Current Comparative Quality (Current QC vs Previous Jan QCs)
Of the last five years at this point in the season, 2023 tracks along with 2022 and 2018. In terms of tournament sanity, it's not what we want to see. In terms of tournament predictability, it means we have two reliable roadmaps to work with.
- Both years produced a F4 of 1,1,3,11 (2018) and 1,2,2,8 (2022), a M-o-M rating of 20.30% (2018) and 21.65% (2022), and round-by-round upset counts of 5-5-3-0-0-0 (2018) and 6-5-3-0-1-0 (2022). There's not too much difference between the two roadmaps.
- If you track the 2023 curve in comparison to the 2022 curve (dark blue vs pink), pink is stronger in the front of the curve and weaker in the back of the curve, which means 2023 would project to be slightly more crazier than 2022. This is also confirmed by the max-min comparison above, but it is opposed to other models like the pre-season ETM.
- In the Opening Week Thoughts article, I mentioned how bad shooting has been. It's improved slightly, but it's below last year's figures (not good for 2023 quality) and definitively below higher quality years like 2017 and 2019. In advanced metrics analysis, shooting is the most important of the four factors (I think Dean Oliver approximated it to be about 40% of a team's efficiency). Turnovers and Offensive Rebounds can get you more shots per possession, but these are easily controlled by ball security and defensive rebounding positioning. Shooting needs to get better!!!
- In the Mid-season Thoughts article, I mentioned that my title contender list expanded from three to seven. This expansion of a contender list in a short amount of time is not what I wanted to see (in my opinion, it should be shrinking in the likes of a shrink from 10 or 12 to 8 or 9 contenders), and the comparisons to 2018 and 2022 seem to suggest that 2023 is a wide open race.
Inflection Points
Now, we are going to look at the Kenpom Curve (the full range of quality values of all 363 teams).
In the graph above, there are a lot of spikes or gaps in the Kenpom Curve. These are areas of the graph where quality differentials are sharp and instantaneous. I label these spikes in quality differentials as inflection points. Let's examine them.
- The noteworthy inflection points on the KC are #1, #4, #8, #13, and #26. Looking at the same graph for 2022, we see a much different story. In 2022, there were only three 1+ pt spikes in the top 45 spots. In 2023, there are five 1+ pt spikes in the top 26 spots. Thus, quality is falling precipitously at the very top. If I were to translate this feature into practical bracket purposes, I would say more higher seeds losing in 2023 R64 than in 2022. Last year, there were six upsets in the R64: One 2-seed, two 5-seeds, and three 6-seeds. The rapid fall in quality could mean either more than six in the R64 or six again but better seeds losing (more 3- and 4-seeds taking the place of 5- and 6-seeds losing).
- The other inflection points of note are #65, #88, and #90 (the last two yellow arrows just before the orange clusters). In 2022, no inflection point from the #46-#95 spots breached the 0.5 differential. Like the 1+ pt spikes group, there are two more in 2023's range than last year. This range of the KC usually comprises the 11- thru 13-seed groups (The top 46 spots should be 1- thru 11-seeds assuming two additional 11-seeds as play-in game participants). In the final KC of 2022, this range had one 4-seed, two 9-seeds, one 10-seed, three (of five) 11-seeds, four (of five) 12-seeds, and three 13-seeds. Ironically, the 4-seed won two games against two teams in this range, the lone 10-seed went to the E8, and two of the 12-seeds pulled upsets over 5-seeds. The double-spike at #88 and #90 has my interest, and if I have to make a guess, I would say it could be a cut-off point for upsets (teams below this threshold are less likely to pull R64 upsets.
- Now, I want to discuss the two orange clusters. Here's why I find them interesting. The first cluster starts at #119 and ends at #146, covers a range of 28 spots, and falls 2.45 quality points. If I examine 28 spots in either direction, it falls 3.39 points from 28 spots before it and falls 2.09 points after it (this is normal as I expect declines to narrow towards the mean/median). The problem is the neighbor groups decline in a somewhat orderly fashion, whereas the orange cluster loses 48.57% of the 2.45-point decline in just four spikes (the middle spike is a double spike). The same phenomenon occurs at the other orange cluster. Even more noteworthy, these spikes were not as strong in 2023 as they were in 2022. This matters because the #96-#150 spots are where the 14- to 16-seed teams should be. For the record, I'm not exactly sure what it portends for the future, but to recap it, quality steadily/evenly declines from #91-119, then three-thrusts down from #119-146, then steadily/evenly declines from #146-#173.
- As a final note, there appears to be another similar spike cluster from the #38 spot to the #54 spot. When added to inflection points in the first bullet point, all of these are the same locations on the QC where current performance was at or above the mid-point of the max-min range for the month. It's almost a perfect match, which is why inflection points intrigue me.
Miscellaneous Information
Last year, I posted the following table and it was pivotal in identifying the potential for 2022 to be an insane tournament. These are counts for the number of Top-50 and Top-100 AdjO and AdjD teams according to KenPom ratings that are in the QC. As the table shows, there are 35 Top-50 AdjO teams in the Jan QC and 35 Top-50 AdjD teams in the Jan QC. If these counts fall by the start of BCW, it usually portends a crazy tournament, like 2022, 2021, and 2018. All three years saw the quantity of Top-50 AdjO and Top-50 AdjD fall from Jan to BCW. When I do the final write-up during BCW, this table will be filled out for the Fin23 (Final QC 2023) and presented as evidence in the article. For one final piece of information in this article, the mid-point of the KC (which I follow in my own work behind the scenes) is spot #182 and its quality-value is -1.00. Last year for comparison, the mid-point was -0.785 between the #179 and #180 spots. We have added five more teams to Division I CBB (and of course, their not going to be anywhere near the top of the curve). The two values are close enough to be inconsequential in meaning, but I also like to document and track the change in this value to understand how quality is dispersed throughout the country.
In closing, a lot of valuable information exists in the Jan QC, but its evolution as BCW approaches will tell the true secrets of the tourney. As always, thanks for reading my work, and I should have another article in about two weeks. However at this point, I cannot confirm if it will be what the Tentative Schedule says it will be, but I am working on it.
No comments:
Post a Comment