Hitting Efficiency is a Kill Rate (Kills per Attempt – KR) minus an Error Rate (Errors per Attempt – ER). Take NYU, for example- Their season long Kill Rate = 42.5% and season long Error Rate = 17.1%. This makes NYU’s H% = 25.4% for the year (42.5%-17.1%). Check out where their logo is placed below – Directly on the cell where its KR & ER join together on the table. Under their logo you may see a 0.60 (or maybe not?). Certainly not their H% = 25.4%, but it is the expected live-ball win rate the moment they swung, i.e. 60%. The first 42.5% in the moment they earned a kill, and the additional 17.5% ultimately winning that same contested point after not erroring, either. (Markov Chain Estimation). You can see the farther a logo goes to the upper left, producing higher Kill Rates & lower Error Rates, the better efficiency becomes. There is no team’s logo in the top left cell, but if one was this efficient, they would kill 52% and only error 13.4% making their H% an astronomical .386, with a point win expectancy of 68%! Nobody like that this year, given who they played against!
Did you catch that last condition, “…, given who they played against?” NYU played the #1 ranked SOS of the year. (See the button indicating as much next to their logo?) Maybe now making sense why their attacking stats are skewed lower & to the right relative to others thought not to be as skilled at terminating. I can’t be certain for how far to the upper left their #1 button should move NYU in my mind, but my guess is at least to where you see Cal-Lu. However, Cal-Lu played the #3 ranked SOS, so they probably warrant a shift to almost where Steven’s presently sits. Not for long though, because Stevens played the #6 ranked SOS, so they might need to shift to the upper left themselves, maybe just a diagonal or two. While your at it, scan your eyes to notice Southern Virginia’s #29 button to know that means a shift reverting back to the lower right, at least as much as Stevens moved to the upper left. Get the idea? The magic number is #20. If a button has a larger number than 20, the team needs to shift to the lower right a distance related to how much higher than 20 it is. If it is less than 20, then the team’s logo shifts to the upper left a distance related to how much below 20 it is. Just like NYU, Cal -Lu, and Stevens did. Wentworth is pretty much like Baby Bear in Goldilocks lore would explain it, “Just right!” Because their button is #19.
Getting back to NYU, who is playing Baruch later today. That #86 button on Baruch likely takes them to the lower right off the chart, no less than a few diagonals. Taken together with NYU’s shift of about 3 diagonals to the upper left, this mindfully puts NYU and Baruch’s attacking performance all year into perspective. You see, the NYU defense, or at least what it has become since these two teams met early in the season, shouldn’t allow Baruch their season long attacking rates, while Baruch likely will allow NYU theirs plus more. I hope to see a match like they had earlier this season, but with something like that happening once every 100 years, me thinks I might be asking for too much.
I gave some serious thought to simply moving the teams above to their rightful “relative place” on the attacking grid, but decided to respect the stats as they are offered at each team’s website. I figured the SOS button would allow the graphic to work, and this way, not alter actual H% data from what it is reported to be.
Everything I see above seems about right knowing how the T100 is forecasting the games today, except for Juniata vs. Springfield. Juniata has a higher T100 rating and is the only higher seed (#9) in round one to be favored by T100, projected to defeat #8 Springfield. The live ball attacking stats suggest both teams shift a little to the upper left making Springfield the more functional offense in live ball play. Seems a little contradictory to the projection, but live ball play is only about 75% of the game’s score, even less for Springfield because of the nature for serving points that never get the chance to be live. Potential to be a really good thing, but also carrying with it a high risk, too. This brings us to the next graphic seen below.
When it comes to Aces per set and service errors per set, I have less of an interest to leave them be, so I standardized all of these teams actual data to reflect the expectation for each playing a top 10 opponent. Now, it is apples to apples, requiring no need to have a button or to mindfully shift any team’s logo based on their SOS. This adjustment dampens the ace per set metric most, because aces are far more dependent on the opponent you are serving than errors could ever be. Think about it this way. An ace requires a serve to get by a worthy opponent with a receive to match. An error means the serving team just missed, but “just missed” is not totally independent of whose on the other side of the net, but a lot less so than aces, however.
Take a look at Springfield below to continue the vein of thought from above. They, together with Wentworth & Southern Virginia are clearly the higher aggressors all year in their serving games. This is why they statistically drive about 1 more ace per set than the rest on average, and simultaneously give up about 1 additional error per set, too. That is not a push, though. It is true the net total of terminal points will be roughly 1-1 from these facts, but the aggression which facilitates a “one for one” in a terminal game, guarantees more live ball wins than otherwise would have been produced. (Aggression promotes out of system passes more often, etc.) In the case of Wentworth and Springfield, I would guess their losses this year were related to the volatility high aggression serving sometimes brings. In the case of Southern Virginia, I suspect the volatility in combination with a 29th strongest SOS simply resulted in less exposure to that type of risk, though I would like to look in on their Rutgers match to see if it was the culprit for giving up 2 of its 3 sets all year in that one match. It seems to me the live ball games for SVU & Wentworth show comfortable margins for their first match today, so serving as aggressive as the graphic below suggest they often do would make very little tactical sense, believing chances to lose today’s match is heightened by that aggression. Though there is something to be said about preparing for a future contest by having a consistent approach with a similar mindset. Maybe it’s worth an additional 2-3% to lose today in order to improve the chances to win by 5-6% tomorrow? I would like to note that Baruch and St. Joes were the only other two serving teams with more than 2 aces per set in their raw numbers, but given their respective SOS, you can see the adjustment brought them back to the pack below.
I think I know why the T100 model is forecasting Juniata rather than Springfield, now. It thinks Springfield wins live ball, but Juniata’s receive is good enough to counter. If Springfield’s aggression puts them at higher risk in the terminal game, then that risk, in theory, can be enhanced: Pass better and error less on the service line. (Get more aces, too – but that is less within their locus of control than the other two, I think.) If I am Juniata, I remember the live ball game was a 76-76 push at St. John Fisher, just like I’d want with Springfield tonight. Then, I’d reflect on having loss the terminal game 28-23 to lose that match 3-1, while considering how best to improve terminal outcomes in the game against Springfield. That Fisher loss for most of the year seemed like it could have the leverage to keep Juniata out of this weekend’s festivities, but a similar one today with Springfield takes them out for good. Ironic that Fisher loss could teach them something about how to defeat Springfield and arrive in a regional final against Stevens tomorrow. Should it happen, it would offer solace to a group of Juniata guys 10 years their Senior who took it on the chin against Springfield and its same coach in 2014 to lose the National Championship match that year. It was the last time they have played each other 8 days shy of a full decade ago.
The gold line gravitates from an Error to Ace ratio from 2.4 to 2.8. As aggressive serving behavior escalates, it has enough other benefits to live with a little higher ratio. This is the reason the green region does not have a gradient as steep as equivalent E/A ratios. As aggressive as the 3 teams on the left are, so might you say about the Hiram Terriers. The difference however is that they weren’t playing a top 10 SOS, and had they been, likely would have made adjustments to that aggression because it looks like it would not have worked as well. Stevens seems to serve in a way to generate its fair share of aces, but rarely at a high price for its serving errors. It makes the game more dependent on their live ball play, thereby risking less. Had that happened against Wentworth this year, we would be talking about two undefeated teams in the tournament. And with knowing how functional Stevens live ball game really is, can anybody blame them for what seems to be a tactic to play to their strength? Certainly not I.
I realized last evening that when I imported T100 ratings into the National Tournament bracket I posted earlier, I did not adjust them to reflect the regional home matches for the favorites, and the potential Final 4 home matches for Loras should they arrive there. The propagation of compounding a home gym factor by .5 points per set advantage for Loras took it from a 10% chance to win the championship up to 18%.* The other regional favorites’ championship win probability changed very little because their gains in winning their regional essentially leaked when potentially having to play a home team in subsequent rounds. Two years ago Carthage won their second consecutive championship at home against a favored and #1 seeded Springfield, so it isn’t unprecedented. The latest link to the 2024 updated bracket reflecting home court is updated and can still be found here.
One of the metrics used all year was the CZR, Central Z-Score Rating – The median of the Z-Scores for a team from the 4 major computer models available online. It has been a worthy “Wisdom of the Expert Crowd” model to accurately assess both strength of teams, and by extension, a composite strength of the schedule they played. In order to see where the buttons came from on the first graphic and to understand where the composite model has these teams ranked before the tournament starts in about 8 hours or so, I offer you that information below. Beneath it is the matrix of round by round win probabilities in a more streamlined format, identical to those found on the bracket linked above.
If you have any interest in seeing how SOS produced by these computer models compare to the NCAA SOS metric related to Opponent Wins & Opponents of Opponent Wins, like I have done extensively for a couple years now, check out the NCAA D3 Volleyball page with its 4 tabs under the “Rankings” header. You can click on each region to see those, just like I would have if they hadn’t posted 2023 data rather than this year’s. Same thing happened a year ago, too. “C’mon Man!” I was thinking I need to get a hold of some guy named Frank to let him know. Until I realized the staticpdfrank name of the file isn’t the embedded name of the guy who does the posting. LOL! (Static – PDF – Rank)
Enjoy the competition today. May the better teams earn their opportunities for tomorrow’s Regional Finals!