Tuesday, October 28, 2014

First Contact

I've entered another dimension.  The dimension of coaching volleyball.  Well, let's clarify this:  Technically, what I am coaching really isn't VOLLEYball.  Given that the team is made up of 7 and 8 year old girls, many of which have never played on a sports team before, there really isn't much volleying going on.  The ball rarely clears the net on a serve and comes back to the team that initiated first contact.  The concept of rally is completely non-existent and "volleys" are rare and far between.  I've noticed there is plenty of squealing, though.

Virtually every practice is comprised of derivatives of two basic drills:  one focusing on serving and one focused on passing.  There is no "setting" and "attacking" in seven year old volleyball.  But, there are kneepads.  Boy, do they love to slide around on them when they should be listening to "Coach Greg".

Despite this young offshoot of the game we all love, I have been struck with some similarities to the college game.  Never were the basics of "first contact" more on display than last week when SFA hosted Northwestern State.  The match - almost entirely - could be described as a battle between serve and serve receive.

Two years ago, I was interviewing Paige Holland for this blog.  I had a legal pad with a page full of setting related questions.  At one point, I asked Holland what things a beginning player should work on.  Since she was a setter, I initially thought her answer might in some way have to do with setting, footwork used by setters, how to hold one's hands while trying to set, etc.  Of course, about a millisecond after I had asked the question, I realized that any reasonable answer wasn't going to have anything to do with setting in the sense that we see Paige Holland or any other college setter perform.

Paige succinctly said that she would have the girls work on serve and receive, since they are the two basic tenants of the game.  Two years later, she has, of course, been proven very right by my brief experiences.  The answer made sense to me then.  It makes more sense to me now, but it is truly profound that volleyball can at many times have basically NOTHING to do with volleying.

The point of first contact for each team is so very important.  How many times have you heard coaches refer to a "service run" as a momentum changer?  How many times have you seen timeouts taken after a couple of shanked returns?  Why is there a position called "defensive specialist" and not "offensive specialist"?  Why are certain players "hidden" in the corner in various receive patterns?  Why did Paige work to perfect the short serve and why has Jill Ivy continued to serve from the neighboring county?

Because serve and receive are the most fundamental skills in the game.

I basically can't write a post without stating that I love digs.  I love, love , love - in rally, back row defense.  But, you can't even have a dig until the ball returns to the serving team.  The concept of the "dig" requires the beginning of a rally.  An ace serve, a serve into the net or out of bounds, a shanked receive, a miscommunication on the back row - all of these preclude there being even one dig on the play.  Yes, even I have to admit (and it's really an easy call when you think about it), serving and receiving serve are generally more important than digging up attacks.

Because of my career profession, I have a penchant for numbers.  I love sports statistics and at times am able to make convincing arguments using them and at other times probably over state their importance.  Honestly, volleyball is sorely lacking in truly meaningful statistics.  There are statistics that could be created that would be more representative of talent, but they'd require review of film to accurately record.  Either that or many people on the sideline tracking very specific information.  Now, it's really hard to record the volleyball statistics that we do have at our disposal due to the pace of the game.  Any time someone new works at a volleyball media table, one of the first things they remark about is the "stat calling" that takes place.  Often times it takes three people to record the statistics that you see in a typical volleyball box score:  an "inputter" , a "caller" and a "writer". The input person basically doesn't see the match.  He/she is typing at a furious pace based at what they hear from the caller.  The caller just barks out codes - for two hours straight.

"Serve Home 11, receive 5, attack 9, dig 12, attack 2, over, dig 11, attack 15, kill, assist 1"

This would be what a caller would say on a very short rally.  Now imagine the ball going over the net more than three or four times in a rally.  It gets INSANE.  That's why you often have a "writer".  The writers job?  Simply to write down - in shorthand - everything the caller says.  Why?  Because over the course of two hours there probably are going to be close to a thousand calls.  At least several hundred.  Do you think that can be done without occasionally mistaking a "3" for an "8" on a jersey?  Or occasionally not saying "over" when the ball is blocked or batted back on to the attacking side without a dig?  Corrections in the flow of play have to be made.

Two years ago, we played at Louisiana-Monroe and the "caller" and the "inputter" were both rookies.  Plus, they didn't use a "writer".  It was horrible.  Absolutely horrible.  At the end of three sets, we had scored more than 70 points and our setter had 5 assists in the box score.  This is almost physically impossible to actually happen.  That is, unless it is 7 and 8 year old league!  The entire box score was redone later that night, in part by using Katzy Randall's stats that she had recorded on the sideline just for coaching purposes.  It was an utter disaster and without Katzy's help, the statistics would have had to been completely redone by video.

So, it is hard to imagine that volleyball will ever adopt many more "official statistics".  What the game needs is a stat that accurately measures serve and receive effectiveness.  Aces are a nice stat, but they are too course.  As mentioned in previous posts, many teams use a passing "point system" whereby points are given to passers based on how accurately their passes go to target.  But, despite these numbers being important they miss one huge component:  the correlation between the quality of the serve and the quality of the pass.  It simply is not true that the accuracy of a pass can be claimed to be independent of the quality of the serve.

Simply put, if I am serving, an opposing team will score better on their passing score than if Jill Ivy or Paige Holland is serving.  An extreme example, but still one that makes the point.  Some players simply don't have the serving skill that others do.  So, if you are tracking passing scores on the sideline, your recorded numbers are artificially inflated when you play poor serving teams.  Likewise, they are systematically depressed when you play a tough serving team.  So, your recorded ability to receive is correlated to the quality of serves you face.  No volleyball metric I know factors these two things in TOGETHER.  They are always separated.  That's at least marginally deceiving.  In some cases, it may render what you are recording close to useless.

On a similar note, take a look at this:  Last Tuesday, SFA won the first set against Northwestern State despite being outhit .200 to .194.  Now, of course, that difference is negligible.   There is no meaningful difference in those attack percentages across dozens of games, much less a single set.  It's just that one starts with a "2" and the other starts with a "1", so it has a different feel.  In the second set against the Demons, SFA won the set despite hitting only .041.  You hardly ever win sets in which you hit that low.  Northwestern State hit only .065 in that second set.  If all you saw was the box score, you might conclude that both teams were having awful attacking nights.  Actually, that wasn't really the case.

Additionally, if you look only at the numbers you'll find that SFA had six aces compared to just one for Northwestern State.  This isn't overly impressive in and of itself.  Six aces is not a HUGE total for three sets.  SFA averages about four aces per three sets, so six isn't ridiculously large compared to average.

Now, if you were to go back and watch the tape of the match, you'd be inclined to believe that Demon OH Caiti O'Connell had a nice attacking match.  However, the stats say she had 9 kills with 7 errors on 32 swings for paltry .062 attack percentage.   Still, I claim O'Connell kept the Demons in the first two sets with her attacking.

Why?  What gives here with all these seemingly poor numbers?  They don't come close to telling the real story.  And there is a simple explanation why.  All of these numbers depend upon first contact for each side being clean in order to have a high level of relevance.  O'Connell's nine kills were virtually all skillful attacks on out-of-system balls.  Many of the 16 attacks that didn't result in a kill or error were artful plays just to keep the ball in alive.  Those nine kills were hardly EVER in system.  Most of them were off junk sets by Jaeger or Johnson - balls just flipped to O'Connell because she was the only place the ball could go.  Yet, O'Connell scored on some of them and kept others in a rally.

The missing link here?  Attack percentage is correlated to receive quality.  Yet, we have no true measure of this correlation.  Northwestern State REALLY struggled to pass against us last week.  Their poor first contact led (in part) to poor attacking numbers. 

Why?  How much of it is due to Bailey Martin having an off match?  How much of it is due to SFA serving the Demons tough?  See, I think both of those things are true.  I think SFA served REALLY well, but yet six aces don't tell that story completely.  I also think that Bailey Martin really struggled.  Bad.  Her four reception errors tell part of that story, but they don't come close to explaining all the times that the Demon setters were sent sprinting all over the court for second touch.

Now, hey, this is not to dog on Martin.  Martin actually has played well in place of Keelie Arneson.  Plus, the other back row serve receive players for NWLA weren't exactly blameless either.  The Demons just did not pass well and if you go by the recorded stats alone you really wouldn't think it was the biggest key to the match - yet, it was.

One number we can see is .103.  Northwestern State hit .103 for the entire match.  This number is almost entirely due to each teams' first contact.  That number is FAR better explained by SFA's serving and NWLA's receipt of serve as opposed to the swings of the Demon hitters.  So, here what I am saying.  In THIS case, the poor "attack" percentage of .103 had little to do with "attacking".  It had far more to do with serving and passing.

The point being made here is one that we all need to keep in mind when looking at just box scores, or things like GameTracker.  The point is especially relevant for statheads like me.  This is certainly one of those "talk in the mirror" type posts.

We love numbers in sports.  Some of us adore them more than others.  It's often said that the numbers don't lie - and to a point, that is true.  However, in volleyball so many of our main numbers are correlated or associated with variables that we can "see" when we watch matches, but that we don't have a statistic for.  This is particularly true when it comes to serve and receive.

A better measure of serve quality than aces would be:  "What fraction of serves lead to the other team becoming out-of-system over and above what would be expected by an average reception team?"  This is almost impossible to measure.

A better measure of reception would be "What fraction of balls on serves of average difficulty are passed to target?"  Then, ask the question again replacing the phrase "average difficulty" with "low/high degree of difficulty".  This might could be measured, but we'd have to try and define what average/low/high degree of difficulty serves would be.  This is difficult and surely subjective.

Instead, we relegate ourselves to the yes/no granularity of "ace or no ace".  Likewise, we subjectively rate serve receives as things like "1's" or "2's".  Finally, serve receive is almost always rated by the team doing the receiving and not someone acting in a neutral capacity.  So, I claim passing scores rated by the bench are subject to some degree of human error and bias.  This doesn't render these numbers useless.  It just means they are a function of the actual person doing the recording. 

On the other hand:  A kill is a kill.  An ace is an ace.  An assist is an assist.  A dig is a dig.  There is little to no subjectivity in their definitions.  So, we record these things.

Volleyball needs both statistics and the "eye test".  That said, to measure true talent, volleyball needs to consider adopting progressive performance measures that utilize the concept of correlation in some succinct way. 

I'm not holding my breath on the last of these suggestions become a reality soon.  It just isn't functional.  The game is too fast to ask for recording much more than what we already record.  So, in the meantime, we should try and use numbers in the proper context.  We should try and realize they are informative, but not without multiple causes for their creation.  We should always look to not only describe WHAT happened, but HOW and WHY it happened.  For those harder questions, box scores like those created in the game between Northwestern State and SFA last Tuesday should be relegated to lesser importance.

Northwestern State lost to SFA in large part because they passed poorly.  The box score doesn't provide much more to that main story line.

"Numbers Never Lie" isn't the same thing as saying they tell the whole truth.  For last week's match against the Demons, they most certainly did not.