2023 Experienced Talent Model (Pre-Season Update)

If you saw the article schedule on the right-hand side of the blog, you may have noticed I jokingly labeled this model as the PPB Top 25. I personally think rankings are meaningless and wouldn't care if they were removed from all sports. However, if I had to do a pre-season ranking system, I think the ET Model would be a great substitute for a Top 25 poll, especially since it factors in two of the attributes I consider important to the NCAA tournament. In a way, it anticipates not only who is good at the start of the season but also who has the chances of being the last one standing at the end. Granted, the 2021 ETM Top 25 didn't project the eventual National Champion, but it wasn't far off and it produced some other gems along the way. So let's take a look at the 2022-23 ETM.

GRADING LAST YEAR

If you remember in last year's article (published at the end of Nov), I predicted teams that I thought would make the tournament based upon historical probabilities derived from previous results. Let's see how I did.

First, you will notice that the rankings have changed. This is because grades for certain players changed. When I tabulated the initial rankings in Nov, I gave full-season credit for players even though they may not have been playing full-season minutes. As these individual ET values were revised (based on their actual experience attained through the season and not their projected experience attained), the team scores and ETM ranks fell accordingly. Likewise, players that were injured or left teams during the season were removed from their team's ETM score, as they were no longer contributing on the floor for that team.

As for the predictions, I was 2-2 in the ETM Lottery (Top 14) and 4-0 in the bottom 11. Last year, there were 18 out of 25 ETM teams in the NCAA Field of 64 (not counting the Play-in games), whereas 16 was the highest in all previous years. I honestly hope this is due to the new methodology and not just random luck, but currently there is no way to know.

THE 2023 ETM RANKINGS

Below is the 2023 ETM Top 25, and nine of them have question marks, meaning these are the teams I believe that are most likely to miss the NCAA tournament from the Top 25.

First we have MIST. They are loaded with ET players, and their schedule is loaded with high-ET teams (GONZ, UK, NOVA, ALA, ORE/CONN depending on W/L, @DAME, INDx2, MICHx2, OHSTx2, PURx2, IOWAx2, @ILL, and @WISC). Essentially, one or two wins against the non-conf opponents and a .500% Win% in the B10 should get them there. Of the nine teams on this list, MIST has the most obvious path to the Big Dance. The B10 is also down this year, as I doubt they will get nine teams like last year (two squeaked into the First Four) and I don't see the B10 vying for many of the 1-4 seed-lines. I think MIST is more likely to make it than not simply because of the high number of chances at resume wins.

MICH for the essentially the same reasons as MIST. However, MICH has a softer schedule (only UK, UVA and UNC in the non-conf with a potential game against AZST/VCU), which means if they don't acquire quality wins in the non-conf, then they will have to prove they are one of the better B10 teams with a strong conference performance. Also, MICH lost more personnel in the off-season from a team that went further than MIST in the tourney.

CONN seems like the TEX years with Shaka Smart: Consecutive years of Top 25 ETM appearances yet failing to appear in the tourney or win a game when they do. From 2016-2019, they had consecutive ETM Lottery (Top 14) appearances, but only one tourney berth. They've been absent from the Top 25 ETM for the last two years, yet they've made the field of 64 without winning a game. Looking at the bigger picture in the BEC, I think CREI may be the conference's only team not to take a step backward in quality (Translation: lower seed-line) from last year, so CONN can't finish 6th or worse in the BEC and expect a berth. This is probably good for CONN's chances of making the tourney in 2023, albeit at a seed line lower than last year's 5-seed.

AZST is on this list because it feels like sink or swim time for Bobby Hurley. After eight years at the helm for AZST, he has less success to show for it than Herb Sendek (the coach he was hired to replace). According to ETM, this is his best team-to-date, and the P12 is less top-heavy than last year's three musketeers (ARI, UCLA, USC). In a more balanced P12 this year, AZST has to show something or else Bobby Hurley will be on the hot seat with his current contract scheduled to expire at the end of 2024. Going with my gut, AZST seems to be in the same position as 2022 MIA: A high ET rating in a down year for their conference, overlooked by the polls and the Selection Committee, only to surprise with an E8 run. Let's see how this dynamic plays out this year (the other option for this dynamic is MIST).

ORE was on and off the bubble last year like a light switch. They lost four of the Top 8 producers from last year's squad, but they've also added some transfers and a key freshman. The first task is to fix the porous defense that gave up four 20+ point losses last season, which should be possible with four seniors in the starting lineup. Some preseason matches against HOU, CONN, and ALA/MIST should tell us where they stand. They will also need a Top 4 conference standing to be in consideration, as I don't see more than five P12 teams with viable chances at a berth.

WKU and DET for the same reason: 1-bid conference teams have a harder path to the tourney than power conference teams. If they don't win their conference tournament, it will probably mean a high-ETM team will be watching the tourney instead of playing it. I would bet more on DET making the tourney than WKU because CUSA is just so unpredictable in Nov.

LSU is in transition mode this season as they move on from the Will Wade era (that's probably the nicest way I can say that without treading into legal territory). New coach Matt McMahon has a lot of rebuilding to do, but gets to do it with three of his former players from MURR, two Top 100 freshman recruits, and a lot of transfers from the portal. I doubt these new faces will be ready for SEC-level play, especially considering the teams currently in their way (UK, TENN, ALA, ARK, AUB, and TXAM).

MISS is the last team in question, Last year, they were decimated by injuries in a strong SEC conference. The SEC isn't as strong this year top-to-bottom, and IF THEY STAY HEALTHY, they are returning four of their top six in production plus three role players from last year who saw increased minutes due to all of the injuries. They draw the likes of STAN, FLST, and MEM in non-conference play, so losses in those games would be more devastating than wins would be helpful. There is a lot for them to overcome, but at least ETM says they have the experience and talent to achieve it. I'm not so sure, but I do think their chances are better than LSU's chances.

THE BIG PICTURE

Instead of looking at individual teams, let's turn our attention to the entire crop of teams for the current season and make some year-to-year comparisons.

On the whole, the situation for the 2023 tournament looks much more promising than 2022 or 2021 (both periods with the same scoring methodology). Here are my takeaways.

1. At every class level, the ET value is higher. Our theory follows that tournament sanity comes from higher levels of experienced talent. Thus, 2023 looks to be a far more saner tourney than either 2022 or 2021 (in fairness, I couldn't imagine it getting any worse). The real question we have to answer over the course of the season is the level of predictability. Of the twelve tourney years from 2010 and forward, three had 10-12% M-o-M rating, three had 17-18% M-o-M rating, and the other six had approximately 20%+ M-o-M rating. If the ETM is projecting a saner tournament than 2022 or 2021, then we only have two possible ranges of outcomes from a historical view: 10-12% or 17-18%. At this point in the season and from a practical point of view, I would expect these two ranges to be minimum and maximum values for our target range. I think the 19.85-23.46% range of the six insane tournaments may already be off the table.
   
2. As for the 1st Year class, I expect that value to fall because the values for first year players are projections based on maximum experience. They will be revised lower as the season progresses and it is discovered they didn't achieve the projected levels of experience. The only thing I cannot predict is how far it will fall. Even if it falls into the same range as the two previous years, it still keeps the 2023 total ET around the 28,500-29,000 level, which is well above the 24,282-26,180 range of the previous two years.
3. The final detail that I find interesting has to do with class progression. The 4th Year class in 2021 was 6,742 ET points, but that same group (5th Year in 2022) fell to 5,914 ET points when theoretically it should have increased (another year of experience gained by talented players). This little detail may provide the clue as to why my season-long expectation of a calmer tournament never came to fruition by Selection Sunday. Perhaps in my own intuition/bias, I was over-estimating the impact of the 5th year rule, and this metric calculates that there was less ET in the year than I was assuming to be. For the most part, the class progressions from 2022 to 2023 look conducive to an improvement in tournament quality.

ET VALUES AND THE TOURNAMENT

I always want my models to have some kind of value for the purposes of bracket-picking. Through my work on this model in the off-season, I may have uncovered some potential nuggets of value.

Above is a table of the 1- and 2-seeds over the last six years, their ET rank (N=No ETScore/Unranked), and their performance in the tourney. While there is no direct correlation to ET rank and exact performance, the odds of under-performance vs seed expectations is much higher with a lower ET rank. Of the nine 1-seeds ranked higher than 25, only one (2021 BAY) has reached the F4, which is an 89% failure rate to achieve seed-expectations. Of the fifteen 1-seeds who ranked in the ETM Top 25, eight (>50%) reached at least the F4 (seed expectations) and another four reached the E8 (one win short of seed-expectations). These are probabilities that get my attention. Of the ten 2-seeds ranked higher than 25, only one (2016 OU) has reached the E8, which is also a 90% failure rate to achieve seed-expectations. Of the fourteen 2-seeds who ranked in the ETM Top 25, eight (>57%) reached at least the E8 and two more reached the S16 (one win short of seed expectations).

While these probabilities hold their weight in analysis, what I found even more convincing is how many of the "failures to achieve seed-expectations" can be explained by pathing. Let's consider 2022 1-seed BAY and their pathing in the bracket. First, BAY ranked 45th in ETM, which gives hints of pretender status. In the 8/9 pod, they faced UNC with an ETM rank of 7. In the 4/5/12/13 pod, they faced 4-seed UCLA at 4th or 12-seed IND at 22nd. In the lower octet, they had the likes of 2-seed UK at 6th or 6-seed TEX at 5th (though neither made it to the 2nd weekend). In essence, four of the top seven ETM teams were obstacles in BAY's path to the F4. While pathing doesn't explain every short-coming, it does explain a lot and here are a few examples.

• 2021 1-sd MICH (17th): 2-sd ALA (13th), 3-sd TEX (1st), 4-sd FLST (15th), 8-sd LSU (6th), 10-seed UMD (22nd) and eventual opponent 11-seed UCLA (11th). With six Top 25 ETM teams in their bracket and five ranked better than them, I'm surprised MICH reached the E8 against this path.
• 2019 1-sd DUKE (4th): 2-sd MIST (5th), 3-sd LSU (19th), 4-sd VT (25th), 5-sd MSST (20th), 6-sd UMD (22nd), 7-sd LOU (12th), 10-sd MINN (32nd), and 13-sd STL (37th). With this loaded of a bracket, it's a wonder both DUKE and MIST made it to the E8 game, where MIST ultimately won on a final in-bounds play.
• 2019 UNC (14th): 2-sd UK (2nd), 4-sd KU (1st), and 5-sd AUB (21st). Of the six years under observation, 14th was the lowest rank in any year for UNC. Not only facing high-ranking opposition but with their historically worst team, another seed under-performance is explained.
• 2017 KU (14th): 2-sd LOU (8th), 3-sd ORE (39th), 5-sd IAST (30th), 7-sd MICH (17th), 9-sd MIST (25th), 11-sd UR (21st). Ironically, the worst-ranked team on this list made the F4 (especially with a key injury), but it may be the case of every else wearing each other down.
  

Again, this is not the full list of examples of under-performance explained by pathing. I just don't want to bore you with every single example when I think you get my point. Likewise, here's a quick list of other victims in the tournament.

• 2022: 3-sd WISC (N) and 5-sd IOWA (80th)
• 2021: 3-sd WVU (50th), 4-sd PUR (86th) 6-sd BYU (81st), and 6-sd SDST (N)
• 2019: 4-sd KNST (N), 5-sd MARQ (41st), 5-sd WISC (57th), and 6-sd IAST (49th)
• 2018: 4-sd WICH (37th), 6-sd TCU (74th), and 7-sd ARK (81st)
• 2017: 5-sd MINN (78th), 6-sd SMU (43rd), 6-sd CREI (70th), 7-sd DAY (N)
• 2016: 3-sd WVU (42nd), 3-sd UTAH (71st), 7-sd DAY (N)

For reference, a low ETM-rank doesn't automatically signal an upset. It merely signals the potential for one. When I looked at some of the historic runs by Cinderella-teams, ETM explains them too.

• 2022: NR 8-sd UNC (7th), E8 10-sd MIA (10th) and 11-sd MICH (13th) who defeated 3-sd TENN (16th) to reach the S16.
• 2021: F4 11-sd UCLA (11th), E8 6-sd USC (27th), and E8 12-sd ORST (35th)
• 2019: F4 5-sd AUB (21st) and S16 12-sd ORE (10th)
• 2018: S16 7-sd NEV (26th) defeated R32 2-sd CIN (76th), S16 7-sd TXAM (7th) defeated R32 2-sd UNC (6th)
• 2017: F4 7-sd SCAR (26th), E8 4-sd FLA (3rd), and aforementioned S16 7-sd MICH (17th). Here are the low-seed/high-ETM teams that defeated victims listed in the previous section: 11-sd USC (10th), 11-sd URI (21st) and 10-sd WICH (55th), in order respectively.
• 2016: S16 11-sd GONZ (30th) over R64 6-sd HALL (31st) and R32 3-sd UTAH (71st), E8 6-sd DAME (38th) who had to beat R64 11-sd MICH (19th) and then the rest of the octet was smooth sailing against 2-sd XAV (65th), 3-sd WVU (42nd), 7-sd WISC (76th) and 10-sd PITT (51st), and finally F4 10-sd SYR (4th) defeated 15-sd MTSU (who beat pretender 2-sd MIST ranked 41st in ETM), then defeated aforementioned S16 11-sd GONZ, and lastly defeated pretender 1-sd UVA (56th). With SYR as 4th-ranked in ETM, their next closest three in rank were GONZ, HALL, and 5-sd PUR (32nd). That is favorable pathing for SYR or any team if I've ever seen it.

As warned about the victim probability, the same holds true for upset probability. Just because a team has a low-seed and a high-ETM does not guarantee a Cinderella run in the tourney. As demonstrated, pathing matters, and a combination of other metrics should also confirm the probability of an upset. I also want to point out again that the years 2016-2019 use a different methodology. If I update their methodology, it's very likely that the ranks change, but from what I have seen, at worst the margin of change is +/- 6 rank-spots, and the lower the rank, the higher the change as those teams are closer in ET values than higher ranks.

A SNEAK PEAK INTO THIS YEAR

I wouldn't make you read all of this and not tease how it applies to the current season. Since I focused on 1- and 2-seeds as measured against seed-expectations, let's look at this year's potential crop. Looking at the projections made by The Bracket Matrix, here are the current ETM ranks of all sixteen teams in the top four seed lines (these ranks are subject to revision once "experience gained" is determined later in the season and as stated before the range of change is +/- 6 rank-spots).

GONZ (2nd), UK (3rd), UNC (1st), HOU (52nd), BAY (51st), KU (10th), UCLA (24th), DUKE (8th), TENN (14th), TEX (13th), ARK (25th), CREI (39th), IND (7th), ARI (33rd), NOVA (11th), and TCU (16th).

Though all of these teams will not be 1- or 2-seeds in this year's tourney, I labelled the pretenders (potential failure to meet seed-expectations as a 1- or 2-seed) in red (typical PPB fashion). I can't wait to revisit this model in March with updated values and tournament seeds to see how it plays out. Until then, thanks for reading my work and I'll have the next article -- opening week review -- ready to go.