Quality Curve Analysis: Final Edition

It is that time of the year. The brackets have been released, and it's time to see what this year has in store for us. If you are not familiar with Quality Curve Analysis, here are some articles you should read first (Jan Edition, Feb Edition, Mar Edition). Let's start with a consolidation of all four QCs from each analysis this season.

The results are pretty straight-forward.

• The top teams (1s and 2s) showed some separation since the March analysis and they are playing as efficiently as they were before conference play started in January.
• The 9th- thru 28th-ranked teams (3s thru 7s) are almost are roughly in the same position as they were in the March analysis. There was a slight rotation in the curve around the 13th- thru 20th-ranked (4s and 5s) teams as the Final Curve dipped below the March curve around this area.
• The 29th- thru 45th-ranked teams (8s thrus 11s) have deteriorated significantly since March.
• The 46th- thru 50th-ranked teams (12s) have slightly improved.

How does the 2016 QC compare to other years?

It is the all important question. Well, a picture is worth a thousand words, so let's take a look.

Well, it seems to be a mixed bag, so let's break it down into groups.

• For the top 20 teams (1s thrus 5s), the 2016 QC hugs the 2014 QC. The 2014 tournament was the craziest tournament on record. It resulted in a record 15 upsets (for upsets, read Upset City), it produced an E8AV value of 36 (for AVs, read this), and it featured a 7vs8 in the National Championship game.
• For the 21st thru 28th ranked teams (6s and 7s), the 2016 QC splits the difference between the 2014 QC and the 2015 QC. This is very interesting. These teams are weaker than the craziest tournament (2014) but they are stronger than average-upset tournament (2015). This could be a sign as to origin of upsets for later rounds (i.e. - a 7-seed over a 2-seed in the round of 64 or a 6-seed over a 2-seed in the round of 16). It is something to keep an eye on for sure.
• For the 29th thru 45th ranked teams (8s thru 11s), the 2016 QC mirrors the 2015 QC. The 2015 tournament, as described in the previous bullet mark, was an average-upset tournament. It produced 8 upsets (average for all tourneys since 2003, but the lowest since 2010), it produced an E8AV value of 21 (also average), and it featured three 1-seeds (along with a 7-seed) in the Final Four (but that may have more to do with the strength of the 1-seeds in 2015 as shown in the 2015 line).
• For the 46th thru 50th ranked teams (12s), they are significantly worse than those of the 2015 QC. Weaker lower teams mean a lower probability of upsets, but it all depends on where these teams get seeded (if they get seeded at all).

The 2016 Seed Curve

When we introduce seeding into the QC, it becomes the Seed Curve. The seed curve shows us where teams may be over-seeded (received a higher seed when they deserved a lower seed) or under-seeded (received a lower seed when they deserved a higher seed). Let's take a look at the 2016 QC compared to the 2016 SC.

This chart is weird to say the least. Like the I did in the January Analysis, I'll go seed by seed with brief explanations.

• 1-seeds are below QC, but only because they are dragged down ORE, who ranks 9th in the KenPom rankings (KPR) suggesting they should be a 3-seed.
• 2-seed are in line with QC, but they could be stronger as they are dragged down by XAV, who ranks 14th in the KPR (taking out LOU since they are ineligible) suggesting they should be a 14-seed.
• 3-seeds are below QC, being dragged down 16th rank TXAM and 27th ranked UTAH (factoring out LOU and SMU in their rankings). These teams could easily be a 4-seed and a 7-seed.
• 4-seeds are slightly below QC, boosted by UK and dragged down by CAL and DUKE.
• 5-seeds are slightly above QC, boosted by 10th ranked PUR (suggested 3-seed) and 13th ranked IND (suggested 4-seed).
• 6-seeds begin a stark plummet in the SC. They are boosted by ARI (suggested 4-seed) but dragged down by TEX (suggested 7-seed) and DAME (suggested 10-seed).
• 7-seeds are way below quality. IOWA checks in at 18 (a 5-seed), Wisconsin at 30 (an 8-seed), DAYT at 52 (a 13-seed) and ORST at 58 (a 15-seed).
• 8-seeds are also way below quality. They feature 35th JOES (a 9-seed), 40th TXTC (a 10-seed), 47th USC (a 12-seed), and 53rd COLO (a 14-seed).
• 9-seeds start a three-seed zig-zag pattern, which also showed up in the March Edition of the QC Analysis. These teams are stronger than the QC, as they feature 23rd CONN (a 6-seed), 29th CIN (an 8-seed), 38th BUT (a 10-seed), and 44th PROV (an 11-seed).
• 10-seeds are actually on par with their rankings, as they feature a potential 9-seed, 10-seed, and 11-seed. They are heavily weighed down by 84th ranked TEMP (which would be a 21-seed if the tourney went that far. That is such a drastic over-seed that I'm leaning heavily into picking against them, especially since their 7-seed opponent (IOWA) is a 5-seed in disguise.
• 11-seeds are ridiculously strong this year. They are boosted greatly by 11th KPR WICH (a potential 3-seed), 25th KPR VAND and 26th KPR GONZ (potential 7-seeds). They are dragged down by 54th MICH and 56th TULS (potential 13-seeds) and 74th UNI (a 19-seed).
• 12-seeds are ridiculously weak this year. 36th KRP YALE is a 9-seed disguised as a 12-seed and 45th KPR AR-LR is on-par with its seeding. 78th KPR SDKS and 105th KPR CHAT wouldn't even qualify for the NIT with those terrible efficiency ratings.

Comparisons to Previous Tournaments

Well, let's start with the obvious two comparisons. First, let's look at the 2014 SC.

In the comparison of QCs above, we saw the 2016 QC hug the 2014 QC from the 1- thru 5-seeds. We see no such resemblance in the 1- thru 5-seed area in the SC comparison. In fact, the 2016 SC mirrors the 2014 SC in the 7- thru 12-seed area, but the 2016 SC is weaker at each seed than its 2014 counterpart. From the 1- thru 6-seed area, the two SCs look like inverted variants of each other.

Now, let's try the 2016 SC vs the 2015 SC, and see what shows up.

If you see what I see, there are far less similarities between these two than the 2016vs2014 chart. The only pattern I see is the declining pattern from the 1- to 3-seeds, with the 2016 members being weaker than the 2015 members. In the QC analysis at the beginning, the 2015 QC and the 2016 QC were almost similar from the 8-seeds to the 12-seeds. You don't see any of that with the two seed curves. However, this could be a product of the rampant over-seeding in these areas. For example, 10-seed TEMP checks out as a 21-seed according to the efficiency rankings. If TEMP wasn't included in the 10-seed averages, the 2016 10-seeds could possibly match or even exceed the 2015 10-seeds.

One final point I would like make about the SC comparisons. The 2016 7-seeds do not match the quality of those in 2014 or 2015, and both of those years saw a 7-seed in the Final Four. I think it is safe to say that the 2016 SC is not predicting a third straight year for a Final Four 7-seed.

If I had to make comparisons of 2016 to older tournaments for the upper seeds, the two tournaments that come to mind are 2010 and 2012. Here are the images below.

In both of these years, the SC bows upward from the 1-seeds to the 3-seeds, and then the SC flattens out from the 3-seeds to the 5-seeds. From there, the similarities end, as 2010 continues to flatten all the way to the 11-seeds and 2012 cups from 5 to 8, bows downward from 8 to 10, and then zig-zags from 10 to 13. 2016 exhibits a zig-zag pattern from 9 to 12, but again, the 10s would be much stronger if TEMP was not a part of that group. Just for reference,

• Upsets: 2010 saw 11 and 2012 saw 9.
• E8AV: 2010 produced a 25 and 2012 produced a 21.
• Final Four: 2010 saw 1,2,5,5 and 2012 saw 1,2,2,4.

One final curve that shows some resemblances to 2016 is 2005. The image is below.

If I had to see one tournament that looks like a consolidation of 2010/2012 from the 1 to 5 seeds, 2014 from the 8 to 12 seeds, with 2016 from 5 to 8, it would be the 2005 curve. It is not going to be exact because I am meshing together different elements from different years, but all in all, 2005 could proxy for 2016.

• From 1 to 3, it declines uniformly, but 2016 bows upward.
• From 3 to 5, it rises steadily whereas 2016 flattens.
• From 5 to 8, it bows downward whereas 2016 snakes downward.
• From 8 to 10, it slightly rises whereas 2016 begins zigging.
• From 10 to 13, it zig-zags just like 2016.

2005 had 8 upsets, an E8AV of 29, and the Final Four was 1,1,4,5.

As always, I hope you found this informative. If you have any questions or opinions related to the article, feel free to ask them in the comments below.