Experienced Talent Model for 2021-2022

Back in 2018, I created a conceptual model to evaluate tournament quality. Throughout the blog, I've loosely used the phrase "experienced talent" to conceptualize a rare quality in college basketball that identifies potential national champions if a team possesses it and identifies chalkier tournaments (top-seed advancement) when many tournament teams possess it. This model, purposefully labelled the "Experience Talent Model," aims to identify teams and seasons with this quality. (NOTE: If you would like to fully understand this model, click here to read the primary article, and the sections labelled "Data & Methodology" and "Assumptions" should bring you up to speed.) This article will bring the E.T.M. up to the 2022 season with some interesting new insights.

Note on Methodology

For the years 2021 & 2022, I used an experience value instead of an eligibility value. The reason for the change was the similarity of the model's results for 2018 and 2019, even though those two years were drastically different in quality. Players get credit for either 0, 0.5 or 1 year of experience based on games played, usage (avg minutes), or production per minute. In the older years (2016-2019), the experience value was based on the eligibility status (Fresh=1, Soph=2, Junior=3, Senior=4). At the current moment, there are about 40 players across the freshman and sophomore classes that qualify at the 0.5 experience value. This amounts to a reduction of about 800 total ETM points for the year 2022 (the Total Value for 2022 would be 800 pts higher if these players were qualifying for full experience). For the time being, this 800 points is probably insignificant in a total of 26,000, but it is worth noting.

Overall Results

Let's see where 2022 ranks with past years.

As I've thought, 2022 has more experienced talent (3,700 ET more) points than 2021, which forecasts a more chalkier tournament than 2021 (I'd probably quit PPB if tournaments become more insane than 2021, especially that level of insanity with a year-after-year consistency). For perspective, if every Top 100 player stayed all four years with zero red-shirting, the ETM would give a reading of 50,500. The current reading is a little better than half of the maximum ETM. The rest of the information cannot be completed until the bracket is revealed.

Predictive Results

One of the more interesting results produced by the ETM is its pre-season predictive power. This is the Top 25 ETM teams for each of the last five seasons, and their tournament performance for that season (NOTE: PIL stands for Play-In-game loss).

Here are the quick facts: 

1. The National Champion is among the Top 25 ETM teams except for 2021 (Baylor, more on this in the appendix). 
2. Every year except 2017 had a Top 25 ETM team lose in the play-in game (maybe having to do with the sheer quality of 2017).
3. The Top 14 ETM teams (aka - the ETM lottery teams) typically have 3-4 teams not make the tournament that year. 
4. The bottom 11 ETM teams typically have more varying results, anywhere from 3-6 teams not making the tournament (and this more correlates with tournament quality as more ET teams missing meaning more insanity in the tournament).
   
5. Less predictive reliability, but I thought I would point it out anyways: The highest ETM team has never won the title, and three of the five have failed to make the S16. (NOTE: I would avoid using this model for predictive functions period.)

So how does 2022 match-up?

If we apply the trends of the last five years, this would be a safe assessment of the 2022 tournament participants. Of the Lottery ETM teams, these look most likely to miss the tournament based on early season bracketology in combination with advanced metrics rankings. North Carolina has looked wide-variance in its performance, but most important is their inability to protect the paint on defense. As bad as they have looked at it, Oregon has looked even worse at it suffering two 20+ defeats already this season. Miami and Louisville have bad losses early in the non-conference, and even worse, they have to compete for few tournament berths coming from another down year in the ACC. Last year, the ACC sent 7 teams with the highest seed earned being a 4-seed. I don't see eight from this conference, so there will be twelve teams fighting over seven (possibly six due to depth of other conferences) berths.

Of the 15-25 ETM teams, I've identified the four teams with the shakiest prospects of a tournament berth. Arizona State is already 2-5 while Washington is 4-4, and they go up against a top-heavy Pac-12 (UCLA, ARI, USC) and berth-competitors in ORE (each x2 vs their fellow Top 25 ETM team), Utah, Colorado, and Stanford. Mississippi State has no quality wins so far, and their one loss was a blow-out to LOU (also on this list competing for a berth). Georgia Tech has no quality wins, a bad home loss opener to Miami-Ohio, and competes in the ACC. Their one saving grace may be a string of home games in December against WISC, UNC, LSU, and USC (only WISC is not a Top 25 ETM team). This will not matter if they go sub-.500 in ACC play, which plays a 20-game conference schedule.

Last but not least, given that the ETM for 2022 shows an ET Total close to 27,000 with another potential 800 to be added to the total, it's another piece of the puzzle that suggests a calmer year in 2022. Assuming this holds to the end of the season, I would estimate around 8-9 upsets and an M-o-M rating around 12-15%. We'll see as the season progresses, and I'll probably put the final counts in a later article instead of coming back to this one and updating it. As always, thanks for reading my work and I'll be back in a few weeks with my next article.

Appendix

I've spent a lot of time looking at the rankings and the values assigned by the ETM after BAY won the 2021 title. If you read the previous article, which details the methodology of the ETM, the one aspect of the data that troubles me is the Interval Scale. Essentially, the 100th ranked player is only worth 4 total ETM points as a senior, which is essentially negligible. Baylor's Jared Butler was ranked T-95th in his class (which was actually ten spots lower than his teammate and bench-player Matthew Mayer), so by his Junior year (last year's title run), he was only worth 18 ETM points (and thirty points lower than Mayer). This was also the case of 2016 F4-team OKLA, who spent the entire season in the Top 8, but didn't have a single player in the Top 100 of any high school class ranking. These results made me rethink the value scale for Talent. Instead of the highest-ranked player being worth 100 points and the 100th-ranked player being worth 1 point, the talent scale should be condensed by a factor of 0.5 so that the highest is worth 75points and the 100th-ranked is worth 25 points. The scale would still be interval data, but ratio analysis would have a little more significance since the talent of the best would only be worth 3x as much as the 100th-best instead of 100x as much.