2021 NCAA Women's Preview

The 2021 NCAA women's volleyball tournament is underway, with a few matches having taken place last night (the seeded teams who played advancing) and the bulk of the matches on tap for late this afternoon and tonight. 

After many seasons of seeing the same teams atop the seeding, year after year, the last couple of years have seen a changing of the guard. Penn State and Stanford are in the field but unseeded, something that would have been unimaginable a few years ago. USC is not even in the field. 

Last year, Kentucky captured the NCAA title as a relative newcomer on the elite scene and this year, another team from the Bluegrass State, Louisville, is the top national seed. Given that the Cardinals went 28-0 on the season (including wins over fellow national seeds Pittsburgh, Purdue, Kentucky, Georgia Tech, and Nebraska), it would be hard to seed them anywhere else.

However, according to my Conference-Adjusted Combined Offensive-Defensive (CACOD) rating (explained here), which I introduced in 2011, Louisville comes out only as third best. The following chart lists the 16 seeded teams in order of their CACOD.

No team with a CACOD lower than 1.91 has ever won the championship, so I've inserted a dividing line to separate the likely contenders from the rest (although just because something hasn't happened before, that doesn't mean it can't happen).

The top two teams in the CACOD are BYU (28-1 with the only loss occurring at Pitt) and Texas (last year's national runner-up and owners of a 25-1 record this season, the only loss coming in a split of a two-match series at Baylor).

Another team drawing a lot of attention is Wisconsin, with its group of "super-seniors," who are playing in their fifth season as a COVID-19 allowance. This group includes 6-foot-8 middle Dana Rettke and setter Sydney Hilley. The Badgers, who went undefeated last season until falling to Texas in the national semifinals, have not been as dominant this year, losing twice to Purdue and once to Maryland, and barely escaping at Minnesota.

Belated Analysis of U.S. Olympic Gold in Women's (Indoor) Volleyball

Back in August, the U.S. took Olympic gold in women's (indoor) volleyball for the first time. It was quite a transformation for the American squad, whose sometimes shaky performance in pool play gave way to dominance in the medal round. I like to take some time to reflect on a competition, make some graphics, and then write up an analysis. Three months is a long time, but it's just been a busy year. However, I'm finally ready to present my analysis!

To illustrate how the U.S. women upgraded their performance between the pool and medal rounds, I've created the following graphic of their game-by-game point-differentials throughout the tournament, on which you can click to enlarge.

Each game of each match is shown from left to right. In its opening match against Argentina, the U.S. won in a sweep, with margins of 5, 6, and 5 points in each respective game. Hence, victorious games appear as blue bars, greater height reflecting more decisive wins. Skipping to the third match, vs. Turkey, the U.S. took the first two games, lost the third and fourth (shown as red bars, the greater the drop-down, the more decisive the loss), but rebounded for a 15-12 win in the fifth. Next came a blowout loss to the Russian Olympic Committee, and then a two games to one deficit vs. Italy, which the U.S. overcame to win in five. During one stretch, the U.S. dropped seven out of nine games, some by large margins (where all the red is).

This hardly looked like a team poised to win the gold, especially in as dominant fashion as it did! Once the medal round began, however, the U.S. recorded three straight sweeps, by an average of 7 points per game. Readers of this blog will know that the first place I always look is hitting percentage, both that amassed on offense and that allowed defensively. Here are the hitting percentages from the U.S. matches...

In its pool-play matches (Argentina, China, Turkey, ROC, and Italy), the U.S. didn't dominate anyone in hitting percentage. The only domination was by the ROC over the U.S. That all changed in the medal round. After hitting for the most part between .200 and .250 in pool play, the U.S. ranged between roughly .275 and .350 in the medal round. In addition, the Americans started holding their opponents below .200.

The Olympic scoresheets (which no longer seem to be available online) contained a lot of potentially valuable information. As shown below, I gravitated toward three statistics, which I charted...

The first row ("Non-Scoring Reception") pertains to serve-receipt, the second row ("Non-Scoring Dig") pertains to digging, and the third ("Scoring Block") pertains to blocking. Given the importance of launching an "in-system" attack, I thought quality of serve-receipt (defined as serve-receipts deemed "excellent" by some observer, divided by the sum of [opponent aces plus serve-receipts kept in play]) might be key to the U.S. team's improved hitting percentage and overall play in the medal round. But no, U.S. serve-receipt was actually worse in the medal round than in pool play.

What about digging of opponents' spike attempts (defined as "excellent" digs over total attempts)? Not much there either, other than a really good digging match vs. the Dominican Republic in the quarter-finals.

Finally, we have scoring blocks (defined as [points scored via the block - blocking errors]/total blocking attempts). Now, in each of its matches, the U.S. committed more blocking errors (primarily net violations) than it scored points directly via the block, leading to consistent negative values on this statistic. However, the statistics became less negative as the U.S. entered the medal round. It seems anticlimactic to conclude that the key to the American' historic gold medal was blocking for points and staying out of the net while trying to do so, but that is what these data seemed to suggest.

Perhaps, I thought, this was merely a fluke correlation between the blocking statistic and the U.S. team's success. I initially planned to examine the statistics of every women's volleyball match of these past Olympics (not just the U.S. matches) to see if teams with blocking statistics not far below zero tended to do well. However, as I was falling behind in writing up this analysis, I decided to examine only a subset of matches.

In doing so, I found a few instances in which a team scored a large number of points blocking, relative to their blocking errors. Italy scored 12 points directly via the block, while committing only 12 blocking errors (making their statistic .000) vs. the ROC, with Italy winning the match three games to none. Also, Turkey scored 9 points via the block while committing 12 blocking errors vs. China and, what do you know, Turkey recorded a three-game sweep. Finally, the ROC scored 13 points on blocks while committing 13 blocking errors in sweeping Argentina.

As we academics like to say, further research is required to establish the reliability of these findings. Still, there really may be something going on here.

U.S. Men (Indoor) Eliminated from 2020(21) Olympics

The U.S. men's indoor squad has missed the eight-team medal tournament, losing its final pool match to Argentina. Vinnie Lopes at Off the Block has a statistically laden summary (link).

Kentucky Wins 2020 (held in 2021) NCAA Women's Title; Hawai'i Takes 2021 Men's

It was an unusual spring for U.S. collegiate volleyball, thanks to COVID-19-related adjustments, with the women's championship (typically decided in December) and the men's championship (typically being decided five months later) being held only two weeks apart. The women's final was played on Saturday, April 24, 2021, with the men's on May 8. Having nearly three months to digest these matches, I present my analyses of the championships below.

WOMEN'S

For starters, here's a sentence I never expected to write in my lifetime (I'm 58): The Kentucky Wildcats are the new NCAA women's volleyball champions.* UK stopped Texas in four games, 20-25, 25-18, 25-23, 25-22, to claim what really is the 2020 championship (the 2021 title will be determined this coming November).

True, Kentucky was the No. 2 seed. However, previous high seeds from outside the power conferences (Big 10, Big 12, Pac 12) have rarely lived up to their placements**, so I remained skeptical. In the end, however, Kentucky proved to be a more reliable pick than even No. 1 seed Wisconsin, which entered the tournament undefeated, but bowed out in three to No. 4 Texas in the national semifinals.

Kentucky led the nation in team hitting percentage during the regular season (.361) and, to a large extent, rode that hitting to the national championship. Wisconsin came out better than Kentucky on my Conference-Adjusted Combined Offensive-Defensive (CACOD) measure, which is based on each team's own regular-season hitting percentage (HP) divided by the aggregate hitting percentage a team allowed its opponents (with this ratio multiplied by a strength-of-conference adjustment factor). Wisconsin hit .342 in the regular season (a bit below Kentucky's .361), but the Badgers allowed their opponents to hit a combined .128 (compared to .144 for the Wildcats' opponents). The Big 10's higher conference strength (1.25) than the SEC's (1.00) also made Wisconsin look better:

Wisconsin: .342 own HP/.128 opponents' HP = 2.67, which when multiplied by the Big 10's 1.25 conference adjustment factor = 3.34. 

Kentucky: .361 own HP/.144 opponents' HP = 2.51, which when multiplied by the SEC's 1.00 conference adjustment factor = 2.51.

At no time was Kentucky's kill-production more effective than when the Wildcats had the opportunity to close out the championship match against Texas in Game 4. Looking at a portion of the Game-4 play-by-play, once the Wildcats had overcome the Longhorns' early 6-1 edge to take a 15-13 lead, Kentucky successfully kept Texas at arm's length. A trailing team cannot afford merely to trade side-outs, but must put together scoring runs on its own serve to catch up. Yet, as shown in the following screenshots, five of Texas's seven final serving stints consisted of one serve only (the other two consisting of two serves). And, nearly always, it was Kentucky kills (highlighted in yellow) that kept Texas from gaining any ground. (You can click on all graphics to enlarge them.)

Another way to examine Kentucky's hitting attack is through usage/success graphs. In the two graphs shown below (the top one for the championship match vs. Texas and the bottom one for the Wildcats' four-game semifinal victory over Washington), each of Kentucky's five most active hitters (Madi Skinner, Avery Skinner, Elise Goetzinger, Azhani Tealer, and Alli Stumler) is depicted by a rectangle. The width of each rectangle represents the percentage of the team's hitting attempts taken by a player. For example, against Texas, Stumler took 34% (.34) of Kentucky's total number of swings (51/152), the most of any Wildcat, so her rectangle is widest. The height of each rectangle denotes the player's hitting percentage in the match, which was .471 for Stumler in the title match. Rectangles with the largest area convey great productivity by the hitter: a large number of kills on a large number of hitting attempts.

Looking at Kentucky's graphs against Texas and Washington one atop the other allows us to discern at a glance any changes in the Wildcats' allocation strategy between the two matches. For example, Kentucky set Stumler considerably more often (34% of the team's hitting attempts) in the final match than in the semifinal (23%). Changes in players' hitting percentages, such as Madi Skinner's improvement from .360 in the semifinal to .455 in the final, are also evident.

Before leaving the women's tournament, I wanted to revisit Wisconsin's concluding matches. Due to COVID, Big 10 teams played only conference matches during the regular season. The Badgers went 15-0 in this perennially strong conference, sweeping every match with the exception of three that went 3-1 (vs. Illinois, Michigan State, and Minnesota). However, both matches with Nebraska (ranked No. 5 nationally at the time), both matches with Penn State (No. 9 at the time), and one match with Minnesota (No. 5 at the time) were cancelled. It seems, in retrospect at least, that the missed opportunity to play these matches likely cost Wisconsin in terms of tournament sharpness.

The Badgers breezed through their first two NCAA tourney matches against Weber State and BYU, before hanging on 15-12 in the fifth game vs. No. 8-seed Florida in the round of eight.*** Now, Florida is an excellent team, in fact the only team all year to beat Kentucky (the Gators and Wildcats split their two SEC matches). Still, the decline in hitting percentage vs. Florida (relative to Big 10 play) among three leading Wisconsin hitters (6-foot-8 middle Dana Rettke, 6-2 middle/right Devyn Robinson, and 6-4 middle Danielle Hart) was considerable.

Interestingly, Robinson and Hart recovered nicely in their hitting vs. Texas, but it was not enough. As a team, the Longhorns outhit the Badgers, .301-.220.

MEN'S

Due to the COVID-related cancellation of the 2020 NCAA men's season, 2019 runner-up Hawai'i had to wait two years for another shot at the title. This time, the Rainbow Warriors made good on the opportunity, sweeping BYU, 25-21, 25-19, 25-16.

As this match wore on, Hawai'i looked more and more like it could attack the ball unchallenged. BYU recorded a healthy 5.5 blocks in Game 1 (really 5, as the extra half-block comes from awarding .5 credit to three blockers who went up together). However, the Cougars had zero blocks in Game 2 and two in Game 3 (7.5 total). Hawai'i hit .381 for the match (.400, .333, and .400, respectively, in Games 1, 2, and 3). Also, the Rainbow Warriors had only nine hitting errors on the night; we know seven of these are from BYU's "7.5" blocks, so that means UH spiked only two balls out of bounds. 

The Bows' Rado Parapunov, who started off hot in the 2019 final before cooling off, hit .357 this time vs. BYU on 13 kills (with three errors) on 28 swings (one-third of Hawai's overall 84 spike attempts). Several of Parapunov's teammates had even higher hitting percentages, albeit in far fewer attempts (Patrick Gasman, .545, 7-1-11; and Chaz Galloway and Guilherme Voss, each .667, 6-0-9).

Before I go, I wanted to mention that the Hawai'i athletic website's box score from this match is a very elaborate one, containing far more than the usual statistics. As shown in the following screenshot, the extended box score breaks down hitting attempts into first-ball attacks (immediately upon serve-receipt), transition attacks (once a rally has started), and first transition attacks. Hawai'i hit better on first-ball attacks (.524) than it did on transition attacks (.238).

---
*Any more than I expected to write about a post-1908 World Series win by the Chicago Cubs or an NBA title by the Toronto Raptors (2019) or Milwaukee Bucks (2021).

**Examples include the 2019 Pitt squad that was seeded No. 6 and lost in the second round, and Kentucky itself, which as the No. 4 seed in 2017, lost in the Elite Eight to Nebraska.

***In the usual 64-team field, a team must win four matches to reach the Final Four. However, due to COVID, this spring's NCAA tourney consisted of only 48 teams.

2020 (held in 2021) NCAA Women's Tourney Reaches Round of 16

The NCAA women's tourney is now down to 16 teams, who will play on Sunday (except for Wisconsin vs. BYU on Saturday). Fourteen of the 16 national seeds have advanced this far. The two exceptions are Western Kentucky, which upset No. 15-seed Washington State in five games and will now face No. 2-seed Kentucky; and Pitt, which swept No. 14-seed Utah and will now take on No. 3-seed Minnesota.

In my tournament preview (previous posting below), I identified Western Kentucky as a leading upset candidate, based on the Hilltoppers' ratio of nearly 3 (2.91) between their own season-long offensive hitting percentage (.355) to the hitting percentage they defensively allowed their opponents (.122). (WKU's conference-difficulty adjustment in Conference USA was 1.00, so the Hilltoppers' ratio statistics is not changed by multiplying by 1.00.) 

 Another team on my radar was High Point, whose ratio was an even more gaudy 3.29 (=.296/.090). Multiplying by the Big South adjustment factor of .75 yields an adjusted ratio of 2.47 for the Panthers. High Point scored a first-round win over Central Florida, but was then swept by No. 7-seed Purdue in the second round.

I just checked Pitt's regular-season statistics for own and opponents' hitting percentages. They were  .262 and .146, respectively. Dividing .262/.146 = 1.79 and multiplying by 1.10 for the ACC conference adjustment, yields an adjusted ratio of 1.97.

Belated 2020 NCAA Women's Brackets Announced for COVID-Disrupted Season

Typically, the NCAA women's tournament is held every December. Against the backdrop of COVID-19, however, this has not been a typical season. Back in September 2020, the NCAA moved forward on a plan to shift several fall championships to spring, including women's volleyball. Some conferences (such as the Big 10 and Pac 12) opted to delay their entire women's volleyball seasons from fall 2020 to winter/spring 2021. Others (such as the Big 12) stuck to the usual framework and played the bulk of their schedule in fall 2020, but added some 2021 matches to stay sharp for the national tourney. The SEC played a little less than half of its conference schedule in the fall and a little more than half of it in the spring. Throughout this makeshift season, of course, numerous matches were postponed or cancelled.

The regular season has now been played and here we are, ready for the NCAA tournament. Only 48 teams (down from the usual 64) will participate and the tournament will take place entirely in Omaha, Nebraska. The brackets are available here. 

Wisconsin, last year's national runner-up, has remained true to form, going 15-0 in a conference-only season and sweeping all but three matches (which the Badgers won 3-1). Beyond the Badgers, however, the field features several oddities.

Stanford, winner of the last two national titles (and three of the last four), is absent. The Cardinal was able to play only 10 matches this season and went 2-8.

Kentucky is seeded No. 2, higher than I can recall the Wildcats ever being seeded (a little research shows UK was the No. 4 seed in 2017). UK went 19-1 in a conference-only schedule, its only loss coming to Florida (the No. 8 national seed, and with whom the Wildcats split two matches). Kentucky hit .361 as a team for the season, the highest in the land.

Perennial national contender (and seven-time champion) Penn State is seeded No. 13. Many of the other high seeds are familiar faces (No. 3 Minnesota, No. 4 Texas, No. 5 Nebraska, No. 6 Washington).

Readers of this blog will know that I developed a statistic in 2011, the Conference-Adjusted Combined Offensive-Defensive (CACOD), to gauge teams' prospects for doing well in the NCAA tournament based on their regular-season hitting percentages and opposition hitting percentages. It is explained here. For context, no team with a CACOD below 1.91 (which was recorded by 2016 champion Stanford) has won the NCAA women's tournament. CACOD values for the top four national seeds are as follows.

Team	Hitting%	Opp Hit%	Ratio	Conf Adj	Final CACOD
Wisconsin	.342	.128	2.67	x 1.25	3.34
Kentucky	.361	.144	2.51	x 1.00	2.51
Minnesota	.262	.191	1.37	x 1.25	1.71
Texas	.333	.168	1.98	x 1.20	2.38

Wisconsin's CACOD of 3.34 is the highest ever recorded, surpassing the 3.09 recorded by Penn State in the 2014 regular season. That Nittany Lion squad went on to win the NCAA tournament. Hence, if all goes according to form, the victorious fans (however many of them are admitted for live attendance) will be singing "On Wisconsin."

If you're looking for possible upsets, some other teams with high CACOD values are:

• Western Kentucky .355/.122, ratio = 2.91 (x 1.00 for Conference USA), CACOD = 2.91
• High Point .296/.090, ratio = 3.29 (x .75 for Big South), CACOD = 2.47

Action gets underway a week from Wednesday, on April 14.

Stanford Sweeps Wisconsin for NCAA Women's Title, Aided by Multifaceted Blocking Game

With Stanford's win over Wisconsin for the NCAA national women's championship about a month ago coming so easily, 25-16, 25-17, 25-20, it was hard at first to come up with a statistical angle. The Cardinal, led by 6-foot-6 senior outside-hitter Kathryn Plummer's torrid spiking (.459 on 22 kills and 5 errors on 37 attempts), outhit the Badgers, .358-.152 (box score). Madeleine Gates, a Stanford graduate transfer who finished her degree at UCLA, also came up big (.529, 10-1-17). I've already written a lot on Plummer's hitting, however, so I wanted to focus on something else. Then, an idea from five years ago popped into my mind.

As I wrote in February 2015 then-Penn State men's assistant coach Jay Hosack (now head men's coach at George Mason) noted on the Internet-radio show The Net Live that, "blocking should be evaluated more broadly than via direct stuff-blocks for points." For example, blockers could slow the ball down from a spike attempt, making it easier for the players behind the blockers to dig the ball. Guests on that episode of The Net Live referred to slowing the ball as "control blocking." I prefer the term "dampening" the spike attempt.* Here is a schematic I created to illustrate dampening:

Blockers can also ricochet the ball back to the attacking team's side of the net, with the original attacking team keeping the ball in play. A "block back in play" probably would not be as advantageous to the blocking team as would a dampening, as the original attacking team can set up again for a spike. However, a block back in play would buy the blocking team some time to set up to receive the next spike attempt.

I had never gotten around to examining Hosack's idea, as one could not do so simply from box scores and would have to engage in careful observation and charting while watching matches. Perhaps some teams and volleyball statistical services compile dampenings and block-backs, but I'm not aware of how to access these numbers. I had forgotten about Hosack's idea until, for whatever reason, it re-emerged in my mind after the Stanford-Wisconsin match. I decided now was the time to follow through on it using video of the match on YouTube.

With pencil in hand, I began recording Stanford and Wisconsin's dampenings and block-backs, as shown in the following graph. I also plotted stuff blocks (listed in the box score simply as "blocks). As can be seen, Stanford outperformed Wisconsin in all three areas -- stuff blocks, dampenings, and block-backs -- on the night.

If one has the time and interest in charting dampenings and block-backs, I would recommend doing so. The three measures together yield a richer measure of blocking success beyond just stuff blocks. Conceivably, a team could compile very few stuff blocks, but a large number of dampenings, for example. Looking at the box score alone, one would not appreciate the team's blocking success.

***

Every year since 2011, I have computed my Conference-Adjusted Combined Offensive-Defensive (CACOD) measure to predict success in the NCAA women's tourney. I noted before this year's tournament that the lowest CACOD score for a team that went on to win the national title was 1.91 (Stanford, 2016). Stanford's was 2.05 this season, whereas Wisconsin's was 1.85. Hence, had the Badgers defeated the Cardinal, Wisconsin would have been become the team with the lowest CACOD to win the championship.

---

*According to a scenario from the NCAA manual, a defender fielding a ball after a dampening block still receives a dig in the statistics, even though the ball has been slowed down:

Team White player No. 1 attacks the ball. The ball goes off Team Blue player No. 1 and ... (b) goes to Team Blue player No. 2 who keeps the ball in play. RULING: ... In (b), Team Blue player No. 1 is not awarded a block but Team Blue player No. 2 is awarded a dig.