Warming Up the Crystal Ball: 2018-19 Edition

If you are an avid reader of my blog -- and if you're not, then now is a good time to start -- you may remember I did a post in the 2016-17 season with a similar title. No, I won't be providing a link to that article like I normally do, but the article appeared in December of 2016 if you want to look it up. This article will be another attempt at preseason prognostication in December, but in a different fashion than the previous article (which is why I didn't link it). Let's get to it.

The Current Era of College Basketball

In the previous article (The Upset Zone), I described the current era of college basketball (2010-2018 tournaments) as an era of insanity. Every tournament during that stretch except for 2015 and 2017 (7 out of 9 years) recorded a M-o-M rating of 17% or higher. From 1992 to 2009 (18 tournaments total), only three recorded a M-o-M rating of 17% or higher (1999 - 17.74%, 2001 - 17.44%, and 2006 - 17.44%). 3 out of 18 versus 7 out of 9 is definitely a sea-change in tournament stability. Throughout my blog, I've theorized many reasons why this sea-change occurred, but the three most impactful are the three-point arc, the freedom of movement philosophy, and the NBA arms race (and I despise all of them). The three-point arc turns the game into a low-percentage/high-reward crap-shoot. The freedom of movement philosophy of officiating favors certain styles of play over others and puts box scores more in the hands of subjectivity. The NBA arms race takes potential impact players out of the college game before they can accumulate experience and understanding. Together, they add more variance and unpredictability for tournament results. If you want to catch-up on or even re-read this content, I would direct you to the "Timeline Series" (Part 1, 2 and 3), the "Upsets in the Making" series (Part 1, 2, and 3), and the "Experienced Talent Model" article, which are just a few of many articles. To make a baseline for projecting the 2019 tournament, these three factors are more likely to produce a high quantity of upsets in the tournament. Thus, any projection of 2019 should begin with a M-o-M rating baseline of 17% or higher and modify it (higher or lower) as the data suggests.

The Experienced Talent Model - 2019 Preliminary Readings

The Experienced Talent Model is a descriptive model measuring the experience accumulated by the most talented players in each recruiting class. In a way, it measures the impact of the NBA arms race on college basketball. If more players from a range of recruiting classes return for sophomore, junior and senior seasons, then the experienced talent model will produce a high reading (suggesting a sane tournament with a low M-o-M rating). If less players from a range of recruiting classes leave early for professional basketball, then the experienced talent model will produce a low reading (suggesting an insane tournament with a high M-o-M rating). The table below shows the preliminary reading for 2019.

First, the 2019 Tourney, Base and Parity values cannot be known until the bracket is revealed on Selection Sunday. The only information available is the overall 2019 total, which is closer to its 2018 counterpart than its 2017 or 2016 counterpart (not that 2016 and 2017 are similar years). Second, if the values for the 2016-2018 variables look different, it is because they are different. There are always eligibility issues that appear later in the season, and these issues affect the experience multiplier in the model. There were also a few typographical errors in the spreadsheet made by, detected by, and corrected by yours truly. Third, I expect the 2018 total to be different (most likely to be slightly higher) in March than it is in this article for the former reason (but I wouldn't be surprised if the latter reason also happened once or twice). As it currently stands, I don't see any reason to modify our baseline range for the 2019 M-o-M rating. If I had an educated guess, I would project lower than 2018's 20.3% M-o-M rating, but still above the "insanity era" minimum of 17%.

The New NET Ratings

I love them, but it's not for the reason you would think. Everyone in the sports media says the new NET Ratings, which is supposed to replace the RPI, completely misses the mark. As a bracket scientist, it makes my job easier. When teams receive seeds well above or well below their true quality, it makes upset-victims and Cinderellas much more identifiable, and when they can be easily identified, they can be picked correctly. Last year, the Seed Curve-Quality Curve overlay suggested the committee may have got it right on the whole (in my honest opinion, not every team was seeded truly), and this model helped me avoid calling for a large number of upsets (reviewed in this article). If the Selection Committee decides to go all-in on an untested and potentially flawed ratings system, then the high-probability upsets should stick out like a sore thumb. If I had an educated guess, I would project this factor to impact upset counts in the R64 games and potentially 1 or 2 in the R32 games. However, I won't know the full impact until a season-long study of the 2019 QC and SC.

The Eye Test

Even after a month of games especially those in pre-season non-conference tournaments, there is not enough data for the advanced metrics to predict the course for 2019. However, there have been enough games played to reveal flaws in all of the teams, even those currently without a loss. Nonetheless, if I had to make a prediction on which tournament years to which 2019 feels most similar, I think I would say 2008 and 2015. WTF!?!?!?!?! I can't be serious in picking those two years. They each have a M-o-M rating of 12.33% and 12.03% (well below our current projected range of 17-20.3%). The each have a round-by-round upset count of 5-2-1-0-0-0 and 4-3-1-0-0-0, both of which total eight upsets for the whole tournament. Most striking of all, they each feature a Final Four with at least three 1-seeds. For the record (and I hope it doesn't play out this way), if the right four teams each received 1-seeds (in no particular order, GONZ, MICH, DUKE, MIST), 2019 could produce the same Final Four seeds that 2008 produced. Right now, it feels like those four teams are on a different level than everyone else (with apologies to KU, NEV, and UVA). I do worry that two of those teams may not maintain this separation over the course of the season (MICH and DUKE), but a 1-seed is the path of least resistance, which makes a four 1-seed F4 most possible with this group of teams.

Conclusion

As usual, thanks for reading my work. I hope to produce another article before the week of Christmas, and then have the January Quality Curve Analysis completed and posted around Jan 3-6.