Wednesday, December 25, 2013

NCAA Women's Final: Penn State Dominates Early, Hangs on Rest of the Way vs. Wisconsin for Title

Penn State won its sixth NCAA women's volleyball title in school history last Saturday night, stopping an amazingly tenacious Wisconsin Badger squad in four games, 25-19, 26-24, 20-25, 25-23.  The Badgers had lost 3-0 to the Nittany Lions both times in Big 10 play this season (although there were some tight "deuce games" in there). Other Wisconsin losses during conference play (e.g., 0-3 to Michigan and 1-3 to Illinois, both in Madison) made it seem even less likely the Badgers would make it to -- and seriously contest -- the national championship match. But, two nights after shocking defending champion Texas, Wisconsin most certainly did make a serious run for the national title!

Penn State dominated for a stretch, spanning roughly the middle of Game 1 to the middle of Game 2 (discussed below). Once Wisconsin woke up in Game 2, however, the Badgers outplayed the Nittany Lions for most of the rest of the evening. PSU seemed to struggle to play on even terms with Wisconsin from the middle of Game 2 onward. Still, the Nittany Lions came through with brief flourishes at the end of Games 2 and 4, which, coupled with Badger miscues at the same time, proved enough for a Penn State championship.

Based on the play-by-play sheet from the championship match, I created the following graphic of the scoring sequences in the four games (Wisconsin in red, Penn State in blue). Unless you have some amazing eyesight, you'll want to click on the graphic to enlarge it.

The Nittany Lions' dominant stretch, to which I alluded above, began right after Wisconsin had cut PSU's Game-1 lead to 16-15. The Nittany Lions proceeded to close out the opening game 25-19 and then dart off to an 11-5 lead in Game 2. During this run, Penn State hit a sizzling .538 according to my unofficial tally (14 kills with no attack errors, on 26 swings).

At this point, it looked as though the Nittany Lions might waltz off with the national title as easily as they had dispatched the hometown Washington Huskies in Thursday night's semifinal romp (25-14, 25-13, 25-16). Seemingly out of nowhere, however, Wisconsin rebounded from its 11-5 Game-2 deficit to outscore Penn State 14-6 and take a 19-17 lead. The Badgers maintained the upper hand to hold a game-point at 24-23, but a missed serve (coming on the heels of a missed Wisconsin serve while holding a 23-22 lead), kept the Lions alive at 24-all. Back-to-back kills by PSU's Ariel Scott then put Game 2 in the Lions' column, 26-24. Scott, a senior right-side hitter, was Penn State's most effective attacker from the "pins" (outside, by the antennae that define the width of the court), with a .294 evening (21 kills and 6 errors, on 51 attempts).

One likely factor in Wisconsin's turnaround -- and near-win -- in Game 2 was the disappearance of the Penn State block. The Nittany Lions took a 2-1 lead in Game 2 on a stuff by senior middle-blocker Katie Slay and then did not score again via the block until well into Game 3, when Nia Grant and Megan Courtney teamed up to give the Lions a 9-8 lead. (Wisconsin outblocked Penn State on the evening, 14 to 9.) The Badgers pulled off a 6-0 run in Game 3 to take a 17-12 lead and closed things out 25-20.

Jumping ahead to Game 4, Wisconsin used a pair of 4-0 runs to help take a 23-20 lead, and a decisive fifth game seemed all but certain. However, reprising their late comeback against Stanford in the regional finals, the Nittany Lions scored the final five points against the Badgers to take Game 4 (25-23), the match, and the championship. PSU setter Micha Hancock put her signature on the win with two service aces down the stretch.

Two other trends defined the match in my mind. First was the "Battle of the Middles," featuring Penn State's Slay and Wisconsin junior Dominique Thompson. When in the front row, Slay typically remains in the center to hit quick sets. Thompson, in contrast, frequently runs the slide play, in which a middle hitter darts out to the right-hand side in an attempt to hit against one blocker instead of two. Thompson, who ended up hitting .314 on the evening (16-5-35), compiled four kills in rapid succession on slide plays in Game 3, as the Badgers rallied from a 5-3 deficit to even the score at 7-7. Here's a graph of Slay and Thompson's kills per game.

Slay's hitting line for the match was an excellent .481 (14-1-27), helped by an almost complete absence of hitting errors. As the graph shows, however, 10 of her 14 kills occurred in the first two games (6 in Game 1 and 4 in Game 2). Thompson, in contrast, appeared to peak in Games 2 and 3, with only a single kill in Game 4.

The second trend that struck me was how, relative to the Nittany Lions, the Badgers didn't even obtain that many of their points from kills. Instead, using what I call a "grind-it-out" style, Wisconsin kept a lot of balls alive and capitalized on Penn State errors of various kinds (e.g., service errors; ball-handling errors; and hitting errors, which include balls spiked out of bounds and those blocked back to the floor on the hitter's side of the court for an immediate defensive point). In pie-chart form, the following graphs (again, which you can click to enlarge) show how each team got its points (the Lions scored a total of 96 points on the evening, the Badgers, 91).

The Badgers amassed 45 kills*, accounting for just under half of their 91 total points. The Nittany Lions, in contrast, pounded out 61 kills, accounting for nearly two-thirds of their 96 points. However, Wisconsin received larger shares of its respective points than did Penn State from blocking, opponent spikes hit out of bounds (calculated from the box score as opponent hitting errors minus a team's own blocks), opponent serving errors, and opponent ball-handling errors.

Most observers probably would not have expected this relatively high rate of errors from a senior- and junior-laden Penn State squad. Nevertheless, the Nittany Lions' moxie and offensive firepower were enough to earn them yet another crown.


*The box score shows 46 kills for Wisconsin and 62 for Penn State. However, it also shows 1 blocking error for each team, which typically involves a blocker touching the net. According to AVCA scorekeeping guidelines, “if there is a block error, there MUST be a kill for the other team.” In other words, once a net-violation or other blocking error is called, the play must also have a kill recorded, as a technical requirement. To prevent the disposition of any play from being counted twice, therefore, I counted block errors without their accompanying kills. The AVCA guidelines ("Making Volleyball Statistics Simple") are linked in the right-hand column of this page.

Wednesday, December 18, 2013

Statistical Notes Heading into Women's Final Four (2013)

With this year's NCAA women's Final Four getting underway Thursday night in Seattle, today's posting offers some statistical observations. The two semifinal match-ups feature defending champion Texas vs. upstart Wisconsin, and Penn State vs. hometown favorite Washington.

Wisconsin, a one-time power that had missed the NCAA tourney from 2008 through 2012, is now back in an ascendant mode under new coach Kelly Sheffield. Seeded 12th nationally, the Badgers benefited in their part of the bracket from the fact that SEC teams Missouri (No. 4 seed) and Florida (No. 5 seed) were Paper Tigers and Gators, respectively. Having said that, Wisconsin may be the kind of team that can give Texas a tough match (like Michigan in last year's semifinal).

A year ago, I developed a statistic that attempts to measure teams' "grind-it-out" tendencies. To me a grind-it-out team is one that lacks spikers with pulverizing power, but digs opponents' attacks well and avoids hitting errors of its own. A grind-it-out team may need two or more spike attempts within the same rally to finally win a point. The calculation of a team's grind-it-out statistic is quite simple: the number of spike attempts it takes in a match, divided by total points in the match (to control for match length). A power team, which puts away a lot of spikes immediately, will (usually) have few total swings in a match and thus a low grind-it-out score. Last year, Michigan scored highly on the grind-it-out measure.

This year's Badgers, who ranked sixth in the Big 10 in hitting percentage, but first in digs, seem to me to be a grind-it-out team. Below are my calculations of every team's match-specific grind-it-out statistics within the Sweet Sixteen and Elite Eight. Indeed, Wisconsin is the only team to exceed .90 in both of its matches (vs. Florida State; and Purdue). It must be noted that even a power team, such as Penn State, which led the Big 10 in hitting percentage, can find itself in grind-it-out matches, as the Nittany Lions did in their very tight regional final against Stanford. Similarly, Washington (vs. Kansas) and USC (vs. BYU) didn't need many swings in their respective regional semifinals, but sure did in their epic regional final. Wisconsin is the only team shown below to record high grind-it-out numbers while winning its matches relatively easily (each 3-1).


Neb (vs. Tex): 139 total points, Neb 142 TA (1.02)
PennSt (vs. Stan): 217 total points, PSU 219 TA (1.01)
Wash (vs. USC): 224 total points, Wash 217 TA (.97)
Min (vs. Stan): 149 total points, Min 142 TA (.95)
Wisc (vs. FlaSt): 175 total points, Wisc 165 TA (.94)
Wisc (vs. Pur): 184 total points, Wisc 172 TA (.93)
MichSt (vs. PennSt): 179 total points, MSU 167 TA (.93)
USC (vs. Wash): 224 total points, USC 209 TA (.93)
Amer (vs. Tex): 175 total points, Amer 159 TA (.91)

USD (vs. Neb):  140 total points, USD 125 TA (.89)
Tex (vs. Neb): 139 total points, Tex 124 TA (.89)
Stan (vs. Min): 149 total points, Stan 132 TA (.89)
BYU (vs. USC): 187 total points, BYU 166 TA (.89)
Pur (vs. Wis): 184 total points, Pur 162 TA (.88)
Tex (vs. Amer): 175 total points, Tex 148 TA (.85)
FlaSt (vs. Wisc): 175 total points, FSU 149 TA (.85)
PennSt (vs. MichSt): 179 total points, PSU 152 TA (.85)
Kan (vs. Wash): 131 total points, Kan 110 TA (.84)
USC (vs. BYU): 187 total points,  USC 155 TA (.83)
Stan (vs. PennSt): 217 total points, Stan 181 TA (.83)
Ill (vs. Pur): 143 total points, Ill 117 TA (.82)
Pur (vs. Ill): 143 total points, Pur 116 TA (.81)
Neb (vs. USD):  140 total points, Neb 114 TA (.81)

Wash (vs. Kan): 131 total points, Wash 96 TA (.73)


Serving may also play a large role in determining the next national champion. On Monday's broadcast of "The Net Live," a volleyball talk show on Internet radio, the coaches of the Final Four teams were each interviewed (archived broadcasts can be accessed via the link to Volleyball Magazine in the right-hand column of this page). Penn State's Russ Rose, a noted statistical guru, pointed out that Washington, his team's Thursday night opponent, achieved an unusually strong ratio this year of aces to service errors. Serving aggressively will usually drive up both a team's ace and error numbers, but the Huskies managed to keep their miscues relatively in check. Here are the ratios, as I calculated them for each Final Four team.

Team................Aces..........Service Errors...............Ratio

Washington.......201........................221....................... .91

Penn State........152........................243....................... .63

Texas...............117........................226....................... .52

Wisconsin.........162........................355....................... .46

Volleyblog Seattle has a more elaborate look at the Final Four teams' serving prospects.


Finally, I wanted to acknowledge the amazing performance of USC frosh right-side hitter Ebony Nwanebu against Washington. Nwanebu recorded 30 kills and no errors on 53 spike attempts, for a .566 hitting percentage (box score). She took 53 swings and never once hit the ball out of bounds or into the net, or had the Huskies block the ball back onto the Trojan side of the floor for a point. Each time Nwanebu was set, she either got a kill or the ball was kept in play by U-Dub. Such a high hitting percentage on that many swings is quite unusual.

Thursday, December 5, 2013

My Simple Prediction Equation for the NCAA Women's Tourney

Two years ago, I created a very simple prediction equation for the NCAA women's tournament. Each team gets its own value on the predictive measure. To calculate it, you take a team's overall hitting percentage at the end of the regular season and divide it by the hitting percentage the team allowed its opponents (in the aggregate). The result is then multiplied by an adjustment factor for conference strength, as shown here. For any match in the NCAA tourney, the team with the higher value on my measure would be expected to win.

In both 2012 and 2011, my formula did about as well as other, more complicated ranking formulas. I'm not going to do a full-scale analysis for this year's bracket, but I wanted to mention the formula and provide some sample calculations, in case anyone wanted to compute a score this week for his or her favorite team. The necessary information should be available from the volleyball page of a given school's athletics website. Here are 2013 values for my equation for this year's top eight seeded teams...

1. Texas........... (.295/.174) (1.20) = 2.03
2. Penn State.... (.312/.134) (1.25) = 2.91
3. Washington... (.282/.192) (1.25) = 1.84
4. Missouri........ (.362/.170) (1.00) = 2.13
5. Florida...........(.331/.166) (1.00) = 1.99
6. USC............. (.281/.178) (1.25) = 1.97
7. Stanford........ (.313/.170) (1.25) = 2.30
8. Nebraska...... (.271/.185) (1.25) = 1.83

The median value on my measure (i.e., the value that half the teams score above and half the teams score below) tends to be around 1.40. Most teams making the Final Four in the previous two years have scored above 1.80 on my measure, so that is a benchmark for gauging a team's chances this year of making the Final Four. Last year, NCAA champion Texas had a value of 2.19, runner-up Oregon was at 1.82, and semifinal losers Michigan and Penn State had values of 1.47 and 2.85, respectively. In 2011, national champion UCLA clocked in at 1.94, runner-up Illinois scored 1.81, and semifinal losers Florida State and USC had values of 1.47 and 2.04, respectively. So there is hope for teams scoring just above the median!

Saturday, November 30, 2013

Women's College Liberos: Part II (Digging)

We began our analysis of the nation's top women's college liberos with Part I on serve receipt (plus an addendum). Today, we unveil Part II on digging. As I've noted before, the ideal way to evaluate liberos would be with quality ratings of each contact, such as difficulty level of each dig and whether the dig leads to a team running its offense in system. Lacking such fine-grained ratings, however, I must rely on publicly available, online box scores.

When looking at serve-receipt, I initially evaluated USC's Natalie Hagglund, Iowa State's Kristen Hahn, and Michigan State's Kori Moster. Then, based on suggestions from readers at VolleyTalk, I added three more liberos: Nebraska's Justine Wong-Orantes, Hawai'i's Ali Longo, and the University of San Diego's C'era Oliveira. I also have a local interest in Texas Tech's Rachel Brummitt, an excellent libero whose performances may go "under the radar" nationally. I concluded from the analysis of serve-receipt that, whereas Hagglund may consistently be the most flawless at the task, she handles relatively few of the Trojans' serve-receipt opportunities. By my formula, which accounts for both error-free receipt and share of the team's serve-receiving load taken on, Hahn and Wong-Orantes scored best.

Today, we examine the digging statistics for these same seven players and their teams. Whereas the conventional statistic in this area is digs per set (or game), we want to look more specifically at digs per opportunity. To remind everyone, according to the NCAA volleyball manual, "Digs are given only when players receive an attacked ball and it is kept in play..."      

When an opposing team attempts a spike, five types of results can occur, as shown in the following table. (Unless you have amazing eyesight, you'll want to click on the graphic to enlarge it.) The opposing team may score a kill. Or it may commit one of two kinds of hitting errors: getting the ball blocked right back onto the hitting team's floor for a defensive point, or hitting the ball out-of-bounds or into the net (for simplicity, I refer to such outcomes only as "out-of-bounds"). An attempted attack can also be dug; below, I list the digs both by the focal libero and the rest of the team (i.e., the libero's teammates). Finally, we have the "miscellaneous" category. This would include scenarios in which the defense blocked the spike attempt back to the offensive team, but the offensive team kept the ball alive.

To walk everyone through an example, USC's opponents collectively attempted 3773 spike attempts (as of last Sunday). Of these, 1267 or 33.6% resulted in kills for the opponents. A combined 15.7% of opposing spike attempts resulted in errors: 281 (7.4%) were blocked by the Trojans and 313 (8.3%) were hit out-of-bounds or into the net. A total of 43.9% of opposing attack attempts were dug: 552 (14.6%) by Natalie Hagglund and 1107 (29.3%) by Trojans other than Hagglund. Finally, 253 (6.7%) opposing spike attempts fell into the miscellaneous category.

Using Texas Tech as a second example, here is how the above information was gleaned from the team's online statistics page.

Note that the metric in the above chart is percentage of all opposing spike attempts dug. Iowa State's Kristen Hahn leads among the liberos studied at 16.6%. One out of every six spikes attempted by Cyclone opponents ends up being dug by Hahn! The chart also yields other interesting tidbits of information, such as which of the studied teams gives up the highest and lowest percentages of kills to their opponents, which teams block the highest and lowest proportions of opposing spike attempts, etc.

However, we can get even more specific. A player cannot dig an opposing spike attempt that is blocked at the net or hit out-of-bounds. Therefore, we use opponents' non-error attacks as the denominator representing possible dig opportunities (modified to also exclude balls blocked back to the attacking team and kept in play). The following chart includes the relevant equation (in red), which is applied both to teams and individuals.

Again, let's walk through some examples. Iowa State, as a team, dug 1508 balls. The number of opposing spike attempts the Cyclones conceivably could have dug is 2521 (the 1508 they actually dug, plus the 1013 opponent kills Iowa State theoretically could have dug, but didn't). Actual digs divided by conceivable digs (1508/2521) thus yields .598 or 59.8%, as shown above.

What about individual liberos? If we had spatial data on regions of the court to which spike attempts were directed and an idea of how much ground the libero (and other players) was expected to cover, we could compute an extremely precise ratio of a given libero's digs over how many spiked balls she could have dug. Lacking the spatial information, however, I made the less realistic assumption that a libero would be expected to dig all of the opposing teams' spikes that went either for kills or digs. In Kristen Hahn's case, she had 524 digs. Dividing 524 by the 2521 attacks the entire Cyclone team conceivably could have dug (calculated above) yields .208 or 20.8%. All of the liberos studied were subject to the same assumption, so I think the resulting percentages (ranging from Hahn's 20.8 to Justine Wong-Orantes's 12.7) are comparable to each other.

These results should be interpreted with caution, for a few reasons. First, teams do not play equally strong opponents. Comparing teams and players within the same conference is probably reasonable, although I used teams' overall records, not just conference ones. Second, volleyball involves a lot of interlinked, moving parts. For example, when a team blocks a large share of its opponents' spike attempts (e.g., Nebraska and Michigan State, which each blocked slightly more than 8% of opposing attacks), it could mean that many "dig-able" balls are getting stopped at the blocking stage, with more challenging spikes getting past the block and making things harder for the backcourt players to dig.

One final note: Texas Tech coach Don Flora made digging a key priority this season. As shown in the last table, the Red Raiders dug 55.3% of the conceivably "dig-able" balls hit their way. A year ago, the comparable figure for Texas Tech was 50.9%: 1684 TTU digs/(1684 TTU digs + 1624 opponents' kills).

Friday, November 1, 2013

Addendum to "Women's College Liberos: Part I (Serve Receipt)"

Last Friday, I introduced my statistical examination of the top women's college liberos with an analysis of serve-receipt statistics (link). Using methods described in that earlier posting, I created a measure called S2R (for success-squared times receipt activity) and reported scores for Natalie Hagglund (USC), Kristen Hahn (Iowa State), and Kori Moster (Michigan State). I also invited readers to suggest additional liberos for statistical analysis. In response to a message I posted on VolleyTalk about my original analysis, an online discussion ensued in which commentators listed other liberos they thought were among the nation's best. These additional liberos are Justine Wong-Orantes (Nebraska), Ali Longo (Hawai'i), and C'era Oliveira (University of San Diego).

As I noted in my previous posting, one needs a "detailed" box score (which goes beyond the statistics in an ordinary box score) to compute the S2R score. In scouring the Internet for matches played by these latter three liberos, I found five detailed box scores for Wong-Orantes, and two each for Longo and Oliveira. Here are the numbers for the three players (the "^2" expression refers to squaring a number or raising it to the second power).

Justine Wong-Orantes
Serve Receipt
% Success 
% of Team's
Receipts (R)
.333 (16/48)
.479 (45/94)
.256 (11/43)
.326 (14/43)
Ohio State     
.211 (15/71)
Average of values in final column = .313

Ali Longo
Serve Receipt
% Success 
% of Team's
Receipts (R)
1.000.359 (14/39).359
.938.213 (16/75).187
Average of values in final column = .273

C'era Oliveira
Serve Receipt
% Success 
% of Team's
Receipts (R)
.921.380 (38/100).322
1.000.265 (22/83).265
Average of values in final column = .294

As can be seen, all three of these liberos come out at around .300 on the S2R measure, comparable to the leaders in my previous analysis, Kristen Hahn and Kori Moster. A .300 average corresponds to error-free serve-receipt, while taking on 30% of the load in receiving serve for one's team. Taking on a larger share of the team's receiving load can, of course, compensate for making a few errors on the serves one receives.

I'm not sure how much confidence we can have in the results for Longo and Oliveira, as the statistics for each come from only two matches. Oliveira's, however, do come from matches against two extremely tough opponents, Texas and USC, with the Trojans featuring server extraordinaire Samantha Bricio (currently second in the nation in aces per set).

Wong-Orantes, a Nebraska frosh who earned AAA status (the highest possible) from the California Beach Volleyball Association at age 12, obviously is adjusting well to the indoor collegiate game!

Friday, October 25, 2013

Women's College Liberos: Part I (Serve Receipt)

Over at the VolleyTalk discussion boards, there has been periodic debate as to who the top liberos (backcourt defensive specialists) are in the women's college game. Two measures -- serve-receipt and digs -- seem particularly relevant to judging the merits of different liberos. I will look at serve-receipt in the present posting and then examine digs in Part II. Three players I will focus on, in particular, are USC's Natalie Hagglund, Iowa State's Kristen Hahn, and Michigan State's Kori Moster.

For studying these matters, the ideal would be to have observer ratings of the quality of each pass and dig made by a player, such as many teams compile from their study of game video. Lacking such ratings, we can only look at box scores published online. For serve-receipt, moreover, one really needs to find extended or detailed box scores to get good information (here's an example of both a regular and detailed box score being available).

Regular box scores contain each player's serve-reception errors (RE), but not player-specific information on number of serve-receipt opportunities. For the latter, one needs a detailed box score. A confusing aspect of detailed box scores is that, for some matches, a column in the serve-reception section with "O" at the top presents each player's total number of opportunities, whereas for other matches, the "O" column contains each player's number of successful receipts, which need to be added to reception errors to arrive at reception opportunities. The meaning of "O" for any given match can be determined by looking at the opposing team's number of non-error serves.

What formulas can be devised to combine serve-receipt data into a measure of skill at the task? Certainly, receiving a lot serves without error is an important quality. Also, arguably, receiving a large share of the serves directed to one's team is a marker of receipt skill. Someone who receives, say, 40 percent of balls served to her team probably has a wider range of movement and is willing to take more initiative than someone else who receives, say, 10 percent of serves. If a player can receive a lot of balls and do so with very few errors, that's probably someone you want on your team. (Ideally, that player would exhibit not only non-error passing, but also high-quality passing that lets the team's offensive attack run "in system." However, quality ratings are not available online.*)

My first idea for a formula to measure serve-receipt prowess would multiply, for a given match, a player's success rate at receiving serve (non-error receipts over total receipt opportunities) by the proportion of a team's serve receipts the focal player took (total number of serve-receipts is equal to the opponent's number of non-error serves). So, for example, if a player successfully executes .90 of her serve-receipts and takes .30 of her team's receipts, that would yield a score of .27.

The more I thought about it, however, I felt greater weight should be given to punishing serve-receipt errors. To do so, I simply converted the first part of the formula to receipt-success-rate squared. If a player makes no reception errors, her success rate of 1.000 remains the same when squared. A player with a low error rate (e.g., success rate of .90) is punished mildly, as the success-rate squared becomes .81. A serve-receiver who is mediocre at best (e.g., success rate of .60) suffers more heavily under squaring, as her value plummets to .36. Thus, my current working formula is as follows (^2 stands for squared):

(non-error receipts over total receipt opportunities)^2   X   (proportion of team's serve receipts player took)

Although detailed box scores do not seem to be widely available, Hagglund's school, USC, nearly always provides them. I examined all of SC's Pac-12 conference matches this season, except for the Trojans' loss to Arizona last weekend (for which I couldn't find a detailed box score). Hagglund very rarely botches a serve-receipt, doing so only twice in the seven matches (and 70 reception opportunities) summarized in the following table.

Natalie Hagglund
Serve Receipt
% Success 
% of Team's
Receipts (R)
UCLA11-0-111.000.136 (11/81).136
.909.220 (11/50).182
.136 (8/59)
.237 (23/97)
Oregon St.
.086 (3/35)
.113 (7/62)
Arizona St.
.149 (7/47)
Average of values in final column = .143

However, Hagglund doesn't receive a lot of serves, either, usually between .10-.15 of opponents' launches. Her greatest serve-receipt activity occurred in a five-game win over Cal, in which Hagglund flawlessly passed along 23 of the Golden Bears' 97 non-error serves (.237). Even though the squaring of serve-receipt success rates helps Hagglund (because her success rate is usually 1.000, which is undiminished by squaring), her final scores are not that high, due to her low share of Trojan serve-receipts. (SC outside hitters Sara Shaw and Samantha Bricio regularly field more opposing serves than does Hagglund.) Averaging Hagglund's final scores for the seven matches, she ends up with a value of .143, which is much lower than those for other players we'll review.

Next up is Kristen Hahn. Because of the dearth of this year's Big 12 conference matches for which I could find detailed box scores, I dipped back into the 2012 NCAA tournament (to ensure high quality of opposition). Hahn appears more error-prone than Hagglund, compiling six blown receipts in the five matches examined. However, Hahn is much more active on serve-receipt than Hagglund, a trend that has intensified with Cyclone outside hitter (and frequent serve-receiver) Rachel Hockaday finishing her college career last season.

Kristen Hahn (Iowa State)
Serve Receipt
% Success
% of Team's
Receipts (R)
IPFW*13-3-16.812.222 (16/72).146
N Carolina*27-0-271.000.284 (27/95).284
Stanford*26-2-28.929.394 (28/71).340
.364 (16/44)
Kansas47-1-48.979.490 (48/98).470
Average of values in final column = .321
*2012 NCAA Tournament

Hahn's most recent performance, in Iowa State's five-game win Wednesday night at Kansas, was quite remarkable. She received nearly half of the Jayhawks' serves (48 of 98) and made only one error in the 48 attempted receipts. Hahn compiled a .470 score in the Kansas match and averaged a .321 for the five matches of hers that were studied.

Hahn recently told the Iowa State Daily that, "I think serve receive is a very mental game. I think just making sure that mentally, I’m ready to go... I like to observe their servers during warm-ups and know who the starting servers are and what their go-to zone is and what they practice."

Our third contender is Michigan State's Kori Moster, who Spartan coach Cathy George discussed on the October 21 edition of the Internet radio volleyball show, The Net Live.

Kori Moster (Michigan State)
Serve Receipt
% Success
% of Team's
Receipts (R)
Ohio St.
 .484 (30/62)
.273 (12/44)
.308 (33/107)
.176 (15/85)
Average of values in final column = .288

Again focusing on this year's conference matches, I found only four of Moster's Big 10 matches that had detailed box scores. In three of these, Moster received serve impeccably. However, Moster made five serve-reception errors October 17 vs. Minnesota, an unusually high number for a potentially elite libero. In that match, the Gophers' Daly Santana served seven aces, at least some of which were at Moster's expense. For the quartet of matches studied for Moster, she averaged a score of .288, slightly below Hahn's.

One other libero I wanted to mention is Rachel Brummitt of Texas Tech (where I'm on the faculty). During 2012, Brummitt twice won Big 12 Defensive Player of the Week (to Hahn's eight times). I was only able to find a detailed box score for one of the Red Raiders' matches this season, against Kansas. In that match, Brummitt successfully received all 11 serves she faced, but these 11 made up a small proportion of non-error Jayhawk serves (11/73 = .151).

If anyone has additional liberos to suggest, please let us know in the Comments section to this blog. Remember, the current entry was only on serve-receipt. Liberos' digging statistics will be examined in a future posting.


*As I later learned from VolleyTalk, the FIVB's serve-receipt statistics for international play include a count of how many receipts were "excellent."

Thursday, September 19, 2013

Torrid Toreros Take Season by Storm

The break-out team of the women's college season thus far is the University of San Diego, which has risen to No. 2 in the national rankings. The Toreros, featured in this article from ESPN-W's Mechelle Voepel, have played a very tough non-conference schedule and lost only once -- in five games to defending NCAA champion Texas. Last Friday's win over USC really put the Toreros on the national map. West Coast Conference play begins for USD tomorrow night, as the Toreros play at BYU in a match-up of last year's WCC co-champions (the match will be shown on BYU TV, which some cable/satellite systems carry).

Longtime readers of this blog know that I consider hitting percentage the key volleyball statistic. That was the first thing I looked at for USD and indeed the Toreros have shined on this metric. I have plotted the match-by-match hitting percentages for five USD players who get the bulk of the team's spike attempts and for the team as a whole, in the following graph (you can click on the graph to enlarge it).

So far, the Toreros have been riding the arms of senior middle-hitters Chloe Ferrari and Katie Hoekman. Ferrari, who missed the last month of the 2012 season due to ACL injury, has consistently hit at or around the .400 level so far this season, except for the UC Santa Barbara match, in which she hit a perfect 1.000 (6 kills in 6 attempts). Hoekman had a big match in the opener against Texas-El Paso, hitting .708, with 18 kills and only 1 error in 24 attempts. She has hit nearly as high as Ferrari in most of USD's matches. Alaysia Brown, listed as both a middle and outside hitter, is a barometer of how the team as a whole is doing offensively, in the sense that her match-by-match hitting percentages (depicted in the bright blue small dashed line) tracks closely with the team's overall hitting percentage (the heavy bright blue line).

It hasn't only been offense that's taken the Toreros to where they are. They've shut down prominent opposing hitters such as Texas's Haley Eckerman (-.095 hitting percentage), Iowa State's Victoria Hurtt (.071), and USC's Samantha Bricio (.068).

Against 'SC, digging played a big role in the Toreros' four-game victory. USD retrieved 56% of the Trojans' non-error spike attempts (81 digs of 145 'SC hits that were either kills or kept in play). The Trojans' rate of digging USD's non-error spike attempts was only 37%. The Toreros' big digging advantage offset a large USC edge in total team blocks (15-4).

Monday, September 9, 2013

Nike Big Four Tournament in Austin

Four of the nation's top collegiate women's volleyball teams gathered in Austin, Texas this past weekend and it was the host University of Texas Longhorns compiling the best record of the teams, 2-0. All of the match-ups were prearranged, rather than a format of semi-finals and finals being used. UT's wins were both close: 25-27, 25-17, 13-25, 25-21, 15-10 over Penn State, and 29-27, 18-25, 25-16, 27-25 over Stanford. Box scores of the four matches are available at the following links: Texas-Penn State, Texas-Stanford, Florida-Stanford, and Florida-Penn State.

My initial interest was in looking at which players hit at a consistently high level over their teams' two matches. I created the following chart (on which you can click to enlarge), focusing on players who took 20 or more hitting attempts in a match. Highlighted in blue are players who hit (roughly) .300 or better in both of their matches.

Four middle-blockers hit well in both of their matches: Penn State's Katie Slay, Florida's Chloe Mann, and Stanford's Inky Ajanaku and Carly Wopat. Cardinal outside-hitter Brittany Howard recorded attack percentages of .290 and .300 against Florida and Texas, respectively.

Oddly, the team that compiled the best record, Texas, had no players who hit for a high percentage in both matches, and the team with the worst record, Stanford, had three such players. I therefore decided to probe Stanford's matches a little more closely, as shown in the next chart.

Stanford was outplayed by Florida across the statistical indicators examined, although not by a lot. The Gators' winning score of 28-26, 25-17, 18-25, 25-22 thus seems consistent with how the teams played. The Texas-Stanford match was a different story. Stanford statistically outperformed UT in three categories -- hitting, blocking, and digging -- yet still lost in four. The "Cardinal sin" occurred in the serving game. Stanford botched 16 serves against Texas, while scoring only three aces. The Longhorns, in contrast, had much more balanced numbers of aces (9) and errors (11).

Texas also benefited from errant opposition serving vs. Penn State, as the Nittany Lions amassed 22 service errors (with six aces). It may be a case where the Longhorns' reputation precedes them; out of respect or fear of the UT offensive attack, opposing teams may feel it necessary to serve extremely aggressively.

Saturday, August 31, 2013

Hawai'i Tops Texas on 2013 Opening Night

Defending NCAA champion and preseason No. 1 Texas had neither an answer for host Hawai'i's offensive prowess nor much of an attack itself (except for Game 2), as the Rainbow Wahine prevailed in four games, 25-19, 19-25, 27-25, 25-16 (box score).

Youth was served for Hawai'i, as frosh OH Nikki Taylor (10-1-18) and soph MB Jade Vorster (7-0-14) each hit .500 for the match, committing only one hitting error between them. Junior middle Kalei Adolpho added a .421 night for the 'Bows (11-3-19). The following graph shows each team's hitting percentage per game, with the winning team in each game listed at the bottom. You can click on the graphic to enlarge it.

Frosh OH Pilar Victoria (10-3-23, .304) was a rare bright spot for the Longhorns.

In Friday night's other marquee match (in my view), USC defeated host Purdue in four.

Friday, August 30, 2013

Opening of 2013 Women's College Season

The 2013 women's college season gets underway today. ESPN "W" (for women's sports) writer Mechelle Voepel has a pair of articles (here and here) previewing the season, so I'll just add some brief remarks at this point. I've put together the following chart, showing the preseason polls for three major outlets: the American Volleyball Coaches Association, ESPN W, and Volleyball Magazine (you can click on each column heading to be taken to the full reports on the preseason polls). The polls are identical in their top three. Further, the same 10 schools appear in all three polls' Top 10.

Penn St.
Penn St.Penn St.

Defending NCAA champion Texas, which returns front-court stars Haley Eckerman (outside), Bailey Webster (outside), and Khat Bell (middle), is the consensus preseason No. 1. The Longhorns' main loss from last year is versatile Sha'Dare McNeal, who hit .364 and recorded 2.41 digs per game as a senior last year, both second-highest on the team (on hitting percentage, I'm excluding a player with only 25 attempts all season). In addition, McNeal's 0.85 blocks per game were third on the squad. Texas opens up the season tonight at Hawai'i, in a match that starts after midnight for those of you in the Eastern and Central time zones.

Another interesting opening-night match features USC at Purdue. The Trojans return nearly all of their top players from a year ago, led by OH Samantha Bricio. The same can't be said for the Boilermakers, with Ariel Turner having moved on. Turner alone took 37% of Purdue's hitting attempts  a season ago (1622/4347); further, no other Boilermaker was within 1,000 spike attempts of Turner.

Friday, August 23, 2013

40 Years of Following Volleyball

The summer of 2013 marks my 40th year of following volleyball. The following photo montage shows me in Israel in 1973, as I traveled with my father, a USA team organizer, to the Maccabiah Games, an international competition for Jewish athletes. Among the players pictured in action is Ed Becker, a  former UCLA star who was once mentioned in Sports Illustrated (in the middle, going from left to right, of the five USA players shown).

Further information on my volleyball background is available here. With next week's start to the NCAA women's season, I look forward to my 41st year of following the sport and my seventh year of analyzing and writing for this blog.

Thursday, June 6, 2013

Guest Contributor Adam White Suggests New Statistic: "Gift" Points to the Opponent

Adam White is a graduate student at Bowling Green State University. Here is his idea:

I call the statistic “gifts.” Like any worthy measure, gifts are simple to understand and to calculate, yet aspire to novel explanatory force.

Gifts are the difference between a team’s total points scored and the points they scored on kills. This reflects the number of non-kill or “unearned” points handed to them by the other team.

For example, say [Texas Tech] beats BGSU 25-21, but the Falcons outhit the Red Raiders 15-14. This entails that BGSU gave TTU 11 gifts (25-14), while TTU gave BGSU only 6 gifts (21-15).

The point is that, in this case at least, the gifts explain the TTU victory better than do the kills. (This is an exception, but not a rarity.)

“Gifts per set” is probably the best way to track and analyze the measure. Teams would have both a “gifts received” and a “gifts given” figure.

Arguably "gifts given" measures the consistency and effectiveness of a team’s all-around ball handling. A team with a high "gift received" average is likely capable of longer rallies.

These assumptions have been supported by my modest research, as the BGSU varsity team has better gift statistics than does the group of very mediocre sandlot players I measured.

Sunday, May 5, 2013

UC Irvine Block Party Sweeps BYU in Men's Final

The University of California Irvine repeated as men's NCAA volleyball champions last night, sweeping Brigham Young University in three tightly contested games, 25-23, 25-22, 26-24, in Los Angeles.

UCI was even more dominant on paper, as the Anteaters outblocked and outhit the Cougars by wide margins. At the net, UCI prevailed 17-6 in total team blocks (block assists divided by two, plus solo blocks). Whereas the Anteaters' Colin Mehring was the nation's top blocker on a contending team this season by my calculation, it was his teammate, junior middle-blocker Scott Kevorken, who stepped up in the championship match with one solo block and 11 assists.

Irvine's blocking dominance last night was very different from the teams' relatively even blocking in their two regular-season head-to-head matches, both of which were won by BYU. (Box scores for the three matches are available by clicking on the relevant link: first regular-season match, second regular-season match, NCAA title match). I have graphed BYU and UCI's total team blocks in the teams' three head-to-head matches:

UCI's blocking prowess was the story of the match, all the way down to the final points. The Anteaters trailed 24-21 in Game 3, meaning the Cougars had three set points. However, Irvine ran off five straight points to win 26-24, the last three all coming on blocks!

Irvine outhit BYU for the match, .383-.274. In Game 1, UCI recorded 17 kills with only 4 errors in 29 attempts, for a .448 hitting percentage. In Game 2, the Anteaters reduced their hitting errors even further, again scoring 17 kills, but committing only 2 miscues (on 35 attempts) for a .429 percentage. Kevorken and Mehring each had identical hitting lines for UCI: 7 kills, 1 error, 10 attempts, .600 percentage. Outside hitter Connor Hughes (11-3-22, .364) and opposite (right-side) hitter Zack La Cavera (11-1-21, .476) also contributed offensively for the Anteaters.

BYU OH Taylor Sander, Mountain Pacific Sports Federation Player of the Year and first-team All-America, hit .308 on the night (20-8-39). However, six of his eight hitting errors resulted from being blocked. The Cougars' highly touted frosh opposite Ben Patch had a rougher night (7-5-22, .091), with four of his five hitting errors due to being blocked.

Wednesday, May 1, 2013

NCAA Men's Final Four Begins Thursday

The Net Set offers some fairly statistically oriented previews of the Loyola (Chicago)-UC Irvine and Penn State-BYU semifinal matches of this year's NCAA men's Final Four.

Sunday, April 14, 2013

My Ballot for the 2013 Off the Block Men's Awards (Blocking and Serving)

Once again this year, Vinnie Lopes, operator of the "Off the Block" men's college-volleyball blog, invited me to be a judge for his site's Blocker of the Year award, as well as another award, new this year, for best server.

As seen in this previous posting, I created an elaborate system for determining my vote for 2011 Blocker of the Year. The NCAA compiles the statistic of blocks per set (or game), but I had a few quibbles with that. First, players can commit blocking errors (e.g., touching the net) in addition to successful blocks, so I thought errors should be subtracted from successes (akin to how hitting percentage subtracts hitting errors from kills). Also, a set is not a very fine-grained unit, as a game that goes 25-23, for example, provides greater opportunity to amass blocks (and other statistics) than one that goes 25-13. Therefore, I use total points in a given match as the denominator for my statistic. This year, to save some time, I revised my procedures in two ways relative to 2011:
  • Like before, this year I only considered players for Blocker of the Year who were on a top-5 nationally ranked team. However, whereas before I narrowed my candidates to players in the top 50 in the NCAA's blocks-per-set statistic, this year I only looked at players in the top 25 of this category (as of when I conducted my analysis).
  • Whereas before I examined box scores for all conference matches played by a candidate, this year I only looked at matches played against the top-5 hitting-percentage teams in the candidate's conference.
The most recent national rankings I could locate listed BYU, UC Irvine, UCLA, Long Beach State, and Pepperdine, in that order, as the top 5 teams. All are from the Mountain Pacific Sports Federation (MPSF). NCAA individual leaders in blocks per set were obtained here. Top blockers (as of last week) who are on a top-5 team include:

2. Taylor Gregory (Long Beach St.) 1.43 blocks/set
6. Colin Mehring (UC Irvine) 1.26
10. Parker Kalmbach (Pepperdine) 1.22
11. Russell Lavaja (BYU) 1.21
23. Nikola Antonijevic (Pepperdine) 1.05

For each of these players, I examined his blocking performances in each match against opponents who ranked in the top 5 of MPSF team hitting-percentage. As shown here, these teams were (as of April 7, 2013): UCLA (.336), BYU (.329), Long Beach St. (.305), UCI (.303), and Pepperdine (.297).

Just to provide a concrete example, I would look at each match Taylor Gregory (Long Beach St.) played this season (excluding preseason tournaments) against UCLA, BYU, UCI, and Pepperdine. For each match, I would take his total number of blocks (solo plus assists), subtract blocking errors, and divide the difference by total points played in the match. I also did the analogous thing for UCI's Mehring (against UCLA, BYU, Long Beach St., and Pepperdine); Pepperdine's Kalmbach and Antonijevic (against UCLA, BYU, Long Beach St., and UCI); and BYU's Lavaja (against UCLA, Long Beach St., UCI, and Pepperdine).

The following graph (on which you may click to enlarge) shows the results. Above each player's name, vertically, are his blocking performances against relevant opponents. These numbers tend to be low, given that a player's blocks minus blocking errors typically will yield a number from 0 to 5, which then gets divided by the total number of points played in a match (often over 100 and sometimes even exceeding 200 for a tight five-gamer!).

Even though Gregory had the best blocks/set among my five finalists, he did not fare that well under my elaborate blocking formula, recording values consistently between .000 and .024 against the best-hitting MPSF teams. The highest individual-match value was .064 for Mehring in UCI's second match against BYU. As seen in the box score from that match, Mehring had an unusually high 14 blocks (all assists) with no errors. There were 219 points in the match (a five-game come-from-behind win by the Cougars), yielding 14/219 = .064.

Even if one considers that match an outlier for Mehring, his median value (half of his matches above it and half below it; shown as a little red bar) exceeds those of his four competitors. Thus, in a close finish, my votes for Blocker of the Year are:

1st Place: Mehring
2nd Place: Lavaja
3rd Place: Antonijevic

Turning our attention to serving, my starting point was the NCAA's aces per set statistic. As with blocking, using set as the denominator is not a sensitive measure of match length. That's the least of the problem with aces per set, in my view. The bigger problem is the statistic does not take into account service errors. A highly aggressive server (probably using a jump-serve) may get a lot of aces, but he or she will probably also amass a lot of service errors. Service errors cost a team an immediate point, so they are no trivial matter.

Indeed, if we look at the top 7 players nationally in aces per set (as of April 7), we see that nearly all of them have far more service errors than aces! (Service-error numbers can be obtained by going to the athletics website of a given player's school.)

NCAA Stats Total Aces Aces/Set Total Errors Aces/Errors
Smalzer (LUC) 61 .67 77 .79
Ferrer (Coker) 45 .59 82 .55
Christenson (USC) 40 (41*) .47 44 .93
Quiroga (UCLA) 47 (48*) .47 82 .59
Dache (Mt. Olive) 38 (39*) .47 63 .62
Herceg (Ball St.) 35 .47 53 .66
Cabral (Cal Baptist) 53 .46 118 .45
*Number from player's school website (this is used in computing the player's aces-to-errors ratio). Minor discrepancies regarding aces per set sometimes occurred between NCAA and school-website statistics, possibly because different cut-off dates were used in documents I consulted. Also, on a few school pages, column headings were incorrect (e.g., serving statistics listed under setting statistics).

Because USC setter Micah Christenson is the only player above who has close to a 1-to-1 ratio of service aces to errors, I give him my first-place vote for Server of the Year. I vote Joseph Smalzer of Loyola University Chicago 2nd, and Greg Herceg of Ball St. 3rd.

More refined analyses of serve proficiency are, of course, possible. One could look at a team's typical rate of scoring points on its own serve. Let's say a team scores on 40% of its serves (i.e., its opponents side-out 60% of the time). In other words, each time a team serves, on average, it scores .40 point. An ace, as an automatic point, increases the serving team's "yield" per serve by .60 (1 - its usual .40), whereas a service error (worth 0 points) decreases it by .40 (0 - .40). Such analyses were beyond my available time, however.

Sunday, April 7, 2013

New Article on Attack Speed by BYU's Fellingham and Colleagues

The latest issue of the Journal of Quantitative Analysis in Sports includes an article entitled "Importance of attack speed in volleyball" by Gilbert Fellingham, Lee Hinkle, and Iain Hunter of Brigham Young University.

The abstract (brief summary) of the article can be found here. According to this summary, the authors used high-speed photography to measure the time a set was in the air (to .01 of a second) in a number of men's collegiate matches. Set speed was then examined for correlation with kill probability. Quoting from the abstract:

...sets that traveled a further distance had significant increases in the probability of success with a faster set. No trends were seen with sets that were delivered to hitters that were closer to the setter.

A video on Dr. Fellingham's work with BYU's volleyball programs, this time with the women's team, is available here.

Wednesday, January 16, 2013

UCLA Men's String of Five-Game Matches

During his weekly appearance on Internet-radio's "The Net Live," UCLA athletics administrator and volleyball analyst Mike Sondheimer pointed out an unusual stretch of matches played by the Bruin men. As shown on the team's schedule/results log, the Bruins played seven straight five-game matches from January 4-12, winning five of them and losing two.

Last night (after Sondheimer's appearance on the Monday show), UCLA lost 3-1 to Long Beach State, ending the streak.

Mike challenged all listeners and volleyball observers to come up with any other streak of seven (or presumably more) straight five-game matches played by a team at any level of volleyball. If you know of any, please add it in the Comments section, preferably with a link to corroborating documentation.

Broadcasts of "The Net Live" are archived, for later listening. See the Volleyball Magazine link in the right-hand column to access these archives.

Hawai'i Sweeps Long Beach State to Claim Second Straight NCAA Men's Championship

Hawai'i swept Long Beach State last night in Los Angeles to win its second straight NCAA men's championship. Scores were 25-22, 25-...