2019 Quality Curve Analysis - February Edition

Yes, it is the start of a new month, and at PPB, it usually means a new update to the Quality Curve Analysis. If you want to read the January Edition, here is the link. Before I jump into the article, a few quick thoughts are in order. First, the QC made a huge shift, one I didn't see coming and one for which I most likely don't have a full explanation. Second, I said in the previous edition for a high-magnitude shift to occur, shooting would have drastically improve. Well, shooting did slightly improve, but not enough to explain the magnitude in shift, so other unexplained factors exist. Third, the shift did not affect every team across the board, and for those that experienced the shift, it doesn't appear to be at the expense of others in the QC. Fourth and final, I should definitely point out the advanced metrics data being used includes all games played on January 31 and before. With that said, let's dive right into the analysis.

The Shift in the QC

As usual, I'll start with a chart of the KenPom QC and go from there.

As I stated in the introduction, the QC made a huge shift. In fact, the only part of the QC that did not experience the shift was the #23-28 QC-ranked teams with a few of those teams shifting downwards instead of upwards. To put the shift in perspective, the next chart shows the shift in the QC in the context of the maximum and minimum values each Xth-ranked position held across the dates from January 4th to January 31st.

By all games completed on and before January 31, the majority of teams in the Kenpom QC were playing near the peak of the curve (the lone exceptions being the #24-26 and the #50 QC-ranked teams). Furthermore, of the top 50 teams used for the January QC, 44 remained in the Top 50 teams used for the February 50. In the Kenpom QC, there big shift isn't coming from a changing of the guard. For a more specific breakdown, here are the statistics:

• Of the top 7 teams in the Feb QC, only one was not in the Top 7 of the Jan QC. The one team outside of the Jan QC Top 7 moved up 11 spots, the six originals moved less than one spot in either direction.
• Of the #8-16 teams in the Feb QC, seven of the nine were in the top 8-16 teams of the Jan QC. Of the two new members, one obviously fell from the Top 7 of the Jan QC and the other rose nine spots to move into the #8-16 teams of the February QC. Of the seven members of both ranges, the change was three to seven spots in either direction. In other words, those near the 8-12 spots in Jan moved down and those near the 11-16 spots in Jan moved up.
• Of the #17-27 teams in the Feb QC, only one made a 1-spot movement, seven made a movement of four to nine spots in either direction, four made a movement of eleven to thirteen spots in either direction. Of these 11 teams, one fell from the #8-16 range, four stayed in the range, the remaining six rose from the #28-50 teams.
• Of the #28-39 teams in the Feb QC, two made movements of 6- and 9-spots upward respectively, nine made movements of 10- to 19-spots in either direction, and one came from outside the Jan Top 50. Five of these twelve teams fell from the #17-27 range, one stayed in the range, and five came up from the #40-#50 range.
• Of the #40-50 teams in the Feb QC, two made no change in their spot, one fell down three spots, one fell down six spots, one fell down nine spots, one fell down fifteen spots, and one plummeted a whopping twenty-seven spots. The other four came from outside the Jan top 50.

As always, we want to validate our results by using a different advanced metrics data. Theoretically, if the shift is due to methodology of the ratings system being used, then a different ratings system will show a different phenomenon. If the phenomenon has validity to it, a different ratings system will show the same thing. The chart below shows the monthly shift in the QC using the Sagarin QC.

For the most part, the same shift is occurring, just not as large of a magnitude as the KenPom QC. The locations along the QC where improvement and stagnation occur are roughly the same in both QCs. Let's look at the shift of the Sagarin QC in the context of maximum and minimum values like we did for the KenPom QC.

For the Sagarin QC, it appears as though the #1-20 QC-ranked teams and the #35-#42 QC-ranked teams are playing near the peak of the curve, and the rest are near the trough of the curve (with a few intermittent spikes along the curve, such as the #28, #48, and #49 QC-ranked teams). Thus, we are seeing the same shift in the curve, with the difference in magnitude, in my opinion, being due to a difference in scale for the two ratings systems. Now, let's move to the important part of the analysis.

What is the February QC Telling Us?

The easiest way to answer this question is to overlay the 2019 Feb QC upon the 2017 and 2018 Final QC, which I have done in the chart below.

Well, if there is one truth about the QC that I have reiterated a thousand times, it is this: Flat QCs mean wild tournaments and steep QCs mean calm tournaments. My eyes see steepness. Let's take a closer look at the contentious ranges.

• The #1-6 QC-ranked teams are way above both curves. 
• The #7-12 QC-ranked teams hover closer to 2017, whereas the #14-26 QC-ranked teams hover closer to 2018. 
• The #28-43 QC-ranked teams over-perform both curves, whereas the #44-50 QC-ranked teams under-perform both years.

Before we make any interpretations, let's see what the same comparison looks like using the Sagarin QC.

I see the same pattern of steepness in the Sagarin QC that I see in the KenPom QC. Let's point out the contentious ranges first before diving into the interpretation.

• The #1-4 QC-ranked teams over-perform both years (1-6 in the KQC).
• The #5-8 QC-ranked teams fluctuate within a bottle-neck pattern from the 2017 and 2018 curves (this was not present in the KQC, but there was two points that split the difference between the two curves in the KQC: #13 and #27 KQC-ranked teams)
• The #9-16 QC-ranked teams hug the 2018 curve (14-26 in the KQC)
• The #17-27 QC-ranked teams tend to split the difference between 2017 and 2018 (this was not present in the KQC)
• The #28-37 QC-ranked teams over-perform both curves (28-43 in the KQC)
• The #38-42 QC-ranked teams split the difference (not present in the KQC)
• The #43-50 QC-ranked teams under-perform both curves (44-50 in the KQC)

Conclusion

So, what does it all mean? Though both curves show steepness, which is predictive of sane, better-seed advancement tournaments, I don't believe this is a curve similar to the 2007 tournament, which produced a 3-upset, 4% M-o-M chalky dance. Likewise, I don't believe this a curve similar to the 2014 or 2018 tournaments, which produced 13+ upsets and 20%+ M-o-M ratings.

• The weakness in the 13-28 range (which equates to 4- thru 7-seeds) suggests our upsets could come from this range. 
• The absolute strength near the 1-8 range (which equates to 1- and 2-seeds) suggests longer tournament runs for these teams. In 2017, one 1-seed and two 2-seeds missed the S16. In 2018, two 1-seeds and two 2-seeds missed the S16. The front of both the 2017 and 2018 curves are roughly in the same area, so the similar results in S16 runs could be correlated to this. 2019's 1- and 2-seeds are far stronger, meaning less pre-S16 exits could be a high possibility.
• The out-performance from 29-40 (which equates to 8- thru 10-seeds) is interesting. The would make historically strong challengers to 1- and 2-seeds. One thing to keep in perspective, 8- and 9-seeds were historically above average through 2007-2009, yet not a single 8- or 9-seed knocked off a 1-seed in those years (probably due to the historical strength of those 1-seeds).
• I'm not too concerned over the fall-off on the curve beyond the #40 QC-ranked teams. Most of the teams in this range either fail to make the tournament or receive over-seeds and lose in the R64. However, I will keep an eye on this part of the curve over the next month because any significant rise in the curve's tail means less steepness (or more flattening) and more craziness.

As of right now, I would make a projection of a M-o-M rating between 12 and 17% for the 2019. Seeing how 2017 had one 8-seed and one 11-seed in the S16 and 2018 had two 9-seeds and two 11-seeds in the S16, I think the structural differences between the 2019 curve and the other two suggests we could get a different mix of S16 contenders: My best guess would be probably one 10-seed (no 8- or 9-seeds) and one 12- or 13-seed (due to the weakness in the 2019 QC at the 4- and 5-seed lines). As I always say, the most important clue will be the shape of the seed curve, which we cannot know until the Selection Committee releases the bracket. Once the bracket is revealed, my final predictions will be unveiled in the Final Edition of the QC Analysis. Until then, thank you for reading my work and I'll be back on Feb 17th with Part 2 of the Seed-Group Loss Table study.