Tag Archives: Volleyball

Are We Doing It Wrong?

go

There was a story a week or so ago about a computer program beating the world’s best Go players. Apparently Go is an ancient Chinese game that has more or less an infinite number of possible moves and is therefore considered to be the ultimate test of artificial intelligence (AI)*.  I know nothing about either Go or AI but apparently this is a big deal.  The original article is hidden behind a pay wall, but I was able to pull out a couple of quotes that sparked a spot of thinking.

“The (computer program) made moves that seemed foolish but inevitably led to victory over the world’s best players.”

This quote seems to suggest that the computer understood the game and played it in a completely different way to humans have been playing it.  On that theme the current world champion was quoted as saying,

“After humanity spent thousands of years improving tactics, computers tell us humans are completely wrong.  I would go as far as to say that not a single human has touched the edge of the truth of Go.”

As I am avowed questioner of conventional wisdom these thoughts really piqued my interest and obviously I thought about applying them to volleyball.  Like everything, there is a set of parameters about the game that are accepted as conventional wisdom.  For example, according to the rules a team is allowed only three contacts.  The conventional wisdom is that using all three contacts is the most effective way of playing.  But is it?  As I have written about earlier, Frenchman Earvin N’Gapeth has become famous for, among other things, not always using three contacts. Watching him live I was struck by how obvious those plays actually are. Once you accept that it is possible, his actions are the easiest and best solutions.  I would say that nearly everything we do In practice, is in some way based on conventional wisdom.  For some coaches more than others, but there is a lot of it there.

The computer who won in Go won by playing in a different way than people who were locked into a way of thinking going back thousands of years.  What would happen if that computer decided to try to play volleyball?  Would it use three contacts every time?  I think, deep down,  we already know the answer is no.  Would spikers jump off two feet?  Would there be such a thing as the underarm pass?  Would we train in the same way?  And if the answer to any of those questions is no, what would the alternative be?  How would the computer solve the problem of the game?

I don’t think any of us has touched the edge of the truth of volleyball.

The Coaching Is Not In The Interventions

There is a common quote applying to music that I first heard in a Phil Jackson book but have heard in varying forms many times since,

“Music is the space between the notes.”

The quote has been attributed among others to Claude Debussy, and it always makes me think about things like the interactions and relationships in the playing of whatever game is being talked about at the time.  A few days ago I read something that made me think of this idea directly in relation to coaching.

Most people think of coaching as being what the coach does during the game, the timeouts, the substitutions or if we want to go into real ‘depth’, the starting rotation.  Some smarter people understand that what happens in practice is equally important, the drills done, the feedback given, the time taken, the conduct of practice.

The moment I had was when it occurred to me that all of those things are interventions.  The notes, if you will.  But just as music is not in the notes, the coaching is not in the interventions.  The coaching is in the timing of the interventions.  It is choosing the moment when the feedback will have the greatest impact.  It is not giving any verbal feedback at all but allowing the player or team to learn the lesson by themselves.  It is allowing the errors that lead to learning.  It is not jumping up and down on the sideline berating players or the referee but trusting the team to carry out the vision of the game you have taught them in practice.

In short, the coaching is not in the interventions. The coaching is in the space between the interventions.


Read about the great new Vyacheslav Platonov coaching book here.

Cover v2

Zagrywać czy nie zagrywać? – Oto jest pytanie

Artykuł przetłumaczony na język polski przez Zuzannę Dulnik.  Originalne tutaj.

Na początku chciałbym szczerze przeprosić Williama Szekspira. To było nie na miejscu. Przejdźmy dalej.

Każdy wie, że kiedy wygrywasz losowanie na początku meczu siatkówki, powinieneś wybrać tak, aby zacząć od przyjęcia. To jest tak samo jasne jak moja aluzja do Szekspira była idiotyczna. Dla tych, którzy nie znają powodów tego wyboru, aby wygrać seta zagrywająca drużyna musi uzyskać przewagę o jeden raz więcej niż drużyna przyjmująca. Biorąc pod uwagę to jak trudno jest uzyskać kolejny punkt przewagi, przyjmująca drużyna może czerpać z tego korzyści w każdym secie. Istnieje jednak sugestia, że skoro serwująca drużyna może faktycznie uzyskać tę dodatkową przewagę, to opisany efekt będzie zanegowany.

W przerwie między sezonami wziąłem udział w projekcie badawczym z Benem Raymondem (tzn. Ja mu dałem pewne dane a on je zbadał), który zasugerował (lekko konstrowersyjnie), że przerwy taktyczne nie były aż tak efektywne aż mogło nam się wydawać (wszystkie linki znajdują się tutaj). Kiedy skończyliśmy to badanie, zaczęliśmy patrzeć na inne aspekty gry. Pytanie “wybrać zagrywkę czy przyjęcie na początku gry” wydało nam się dobrym do rozpatrzenia. Jeśli chcesz przejrzeć wszystkie dane, kliknij w ten link. Jeśli chcesz podsumowanie, czytaj dalej.

W pierwszym przypadku wzięliśmy procentową średnią przyjęcia w Pluslidze i przeprowadziliśmy symulację komputerową 10 tysięcy setów. Symulacja pokazała, że drużyna, która zaczynała set od przyjęcia wygrała ich 4.4% więcej niż drużyna, która rozpoczynała od zagrywki – 52.2% v 47.8%. Wyraźnie widać, że rozpoczynanie meczu od przyjęcia powinno wyjść na korzyść pomiędzy drużynami o podobnym poziomie. Podobne wyniki, w różnym stopniu, zostały uzyskanie przy przyjęciu powyżej 50%. Ten rezultat wydaje się sugerować, że pomiędzy drużynami o zbliżonym poziomie, wartość posiadania dodatkowej szansy na uzyskanie przewagi niekoniecznie wyrównuje tę niekorzyść wynikającą z rozpoczęcia seta od serwisu.

Wiedząc, czego się spodziewać, spojrzeliśmy później na faktyczne wydarzenia z ligi polskiej i włoskiej. Te wyniki były interesujące. Przyjmująca drużyna wygrała:

Plus Liga – 50.1% setów

Włoska SuperLega – 46.8%

To znaczy, we włoskiej lidze przyjmowanie na początku seta wydaje się być niekorzystne. Dlaczego te wyniki tak się różnią? Wpadliśmy na dwie możliwości. Po pierwsze, wielkość próbki aktualnych danych to ok 500 setów, o wiele mniej niż 10 tysięcy z symulacji komputerowej. Po drugie w rzeczywistości drużyny nie mają równego procenta przyjęcia, jedna drużyna jest lepsza niż druga.

Spróbowaliśmy zagłębić się trochę dalej, by zobaczyć czy możemy znaleźć inne przyczyny. Na przykład czy różnicę robił fakt, że różnica punktowa w secie była mała (trzy i mniej punktów). Jak sie okazało tak i nie. W Polsce, drużyna przyjmująca wygrywała 54.1% setów (jak się spodziewaliśmy), ale we Włoszech tylko 43% setów (odwrotnie do spodziewanych wyników). Włoska liga jest najwyraźniej bardzo dziwna.

Kiedy poszliśmy jeszcze głębiej, odkryliśmy, że we włoskiej lidze pomiędzy drużynami o podobnym poziomie (drużyny z miejsc 1-4 i 5-8 grające między sobą), przewaga drużyny przyjmującej przy wygrywaniu setów była znacząca (56.9% oraz 54.3%). Jednak ten związek nie wystąpił w Pluslidze. Jest to bardzo dezorientujące i zaskakujące biorąc pod uwagę wyniki symulacji komputerowej.

Co możemy powiedzieć na koniec? Przyjmująca drużyna powinna mieć wyraźną przewagę z biegiem czasu, jeśli nie zawsze w okresie krótkoterminowym. Możemy się spodziewać, że przewaga będzie jeszcze większa w wyrównanych setach i pomiędzy wyrównanymi drużynami. Życie nie zawsze toczy się tak jak byśmy się tego spodziewali.

Kilka innych wskaźników, które wyłapaliśmy z aktualnych danych, a które mogą wydać się interesujące.

  • Drużyna, która zdobyła pierwszy punkt miała około 57% szans na wygranie seta
  • Jeśli drużyna rozpoczynała seta od zagrywki i wygrała pierwszy punkt, ich szansa na wygranie seta wynosiła ponad 60% (Plus Liga 60%, Superlega 66%)
  • Drużyna, która jako pierwsza uzyskała ósmy punkt, miała 70% szans na wygranie seta. Ten procent wzrastał do 83%, jeśli przewaga przy ósmym punkcie wynosiła trzy lub więcej.
  • Drużyna, która jako pierwsza uzyskała 16-ty punkt miała 83% szans na wygranie seta. Ta wartość wzrastała do 92%, jeśli przewaga wynosiła trzy lub więcej punktów.
  • Jesli sety były wyrównane (trzy lub mniej punktów), wszystkie powyżej wymienione wskaźniki miały mniejszą rację bytu.

To Serve Or Not To Serve? – That Is The Question

Firstly, my sincerest apologies to William Shakespeare. That was uncalled for. Let’s move on.

Everyone knows that when you win the toss in volleyball you should choose to receive. That is as clear as my Shakespeare allusion was idiotic. For those who don’t know the reasoning behind it, the serving team must win one break point more than the receiving team in order to win a set. Given that it is very difficult to win a break point the receiving team has an advantage in any / every given set. There is, however, some suggestion that as the serving team has an extra opportunity to win that break point, the effect is negated.

During the off season, I was involved in a research project with Ben Raymond (i.e. I gave him some raw data and he researched it) that suggested (somewhat controversially) that timeouts were not as effective as we might think (all the links are here). When we were finished with that we started looking at some other things. The ‘serve or receive first’ question seemed like a pretty good question to investigate. If you want to go through all the data, click on this link. If you just want the executive summary, read on.

In the first instance, we took the average sideout percentage  for the Polish Plus Liga and ran a computer simulation of 10,000 sets. The simulation showed that the team receiving first won 4.4% more sets than the team serving first, 52.2% v 47.8%. Clearly it should be an advantage to receive first between closely matched teams.  The same holds, to varying degrees for any sideout percentage above 50%. This result seems to suggest that the between closely matched teams the value of having an extra chance for a break point does not (completely) even out the disadvantage

Knowing what we expected to happen, we then looked at actual events from the Polish and Italian leagues. Those results were interesting. The team receiving first won:

Polish Plus Liga – 50.1%

Italian Superlega – 46.8%

That is, in the Italian League it seems to actually be a disadvantage to receive first.  Why are the results different.  We came up with two possibilities.  Firstly, the sample size in each league was about 500 sets, a lot less than the 10,000 in the simulation. Secondly, in real life the teams do not have equal sideout percentages, one team is better than the other.

We tried to dig down a little bit deeper to see if we could find other factors. For example, did it make a difference if the set was close (three points or less).  Yes and no.  In Poland, the team receiving first won 54.1% of the time (as expected). But in Italy, they won only 43% of the time (the opposite of expected). That Italian league is apparently very strange.

When we dug down even deeper, we found that in the Italian League among similarly matched teams (teams 1-4 and 5-8 playing against each other) in close sets there was a strong advantage in receiving first (56.9%, 54.3%).  But this relationship didn’t hold in the Polish league. It is all very confusing, and surprising in the context of our simulation results.

So what can we say in the end? The receiving team should have a clear advantage over time, if not always in the short term. We can expect that the advantage is even greater in close sets and between evenly matched teams. Life doesn’t always happen as we expect.

A couple of other indicators we picked out of the actual data that could be interesting.

  • The team winning the first point had a roughly 57% chance of winning the set
  • If the team serving first won the first point, their chance of winning the set was over 60%. (Plus Liga 60%, Superlega 66%)
  • The team reaching 8 points first, had a 70% chance of winning the set. This increased to 85% if the margin was 3 or more points.
  • The team reaching 16 points first, had an 83% chance of winning the set. This increased to 92% if the margin was 3 or more points.
  • The team reaching 20 points first, with a margin of 3 or more points, had a 95% chance of winning the set.
  • If the sets were close (three points or less), all of the above indicators were less likely.

The legendary Platonov, now on iTunes.

https://itun.es/au/SNTxV.l

Blocking And Serving Effectiveness – The Answer?

Thanks for all comments and suggestions.  They were all thoughtful and helpful.

I did a poor job of explaining the situation so a few people misinterpreted the information and thought I was referring to sideout percentage.  The problem is that I never directly refer to serve quality, I only ever relate a serve to reception quality.  I understand that is a complicated way of thinking about it.

To rephrase my point, after a positive (+) serve, the team in question was much more likely to win a point if the serve was a jump serve than if it was a float serve.  For the record these are the actual figures.

opp rec pts total  ratio jump ratio float ratio
R- 187 396 0.4722 98 197 0.4975 88 197 0.4467

The ‘winning’ suggestion was the observation that the float servers are most often middle blockers who then must defend.  That is, the libero is not on court.  Breaking down the above situation by whether or not the middle blocker is defending, we get the following figures for the float serve.

Middle in defence 48 119 0.4034
Libero in defence 40 78 0.5128

The sample isn’t really big, but it seems to show that there is a fairly large ‘libero effect’, at least with this team.

Oddly, for the ‘2’ reception, the ‘libero effect’ is much smaller and doesn’t explain the differences so well, but for the moment I am satisfied.

Blocking And Serving Effectiveness

I have an observation that does not make sense to me, for which I can think of no explanation.  Perhaps you can help.

I have found over the years that reception from jump serves and reception from jump float serves is not the same.  I don’t mean that the quality of reception is different, I mean that given a particular quality of reception, the likelihood of winning a point is different.  In some cases vastly different.

For example, studying a team recently, I discovered that if the jump servers can force a negative reception, (i.e. only one possible attacker, basically a high ball attack) that team wins the point roughly 50% of the time.  However, if the jump float servers can the same negative reception, the team wins the point only 45% of the time.  The difference is even more stark with good reception without first tempo (i.e. both the outside can still attack a fast ball, a classic ‘2’ pass).  In that case the jump servers won 43% of the time, but the float servers only 26%.

In each case the definitions are the same, the scoutman/recorder is the same, the team is the same, the figures are based on a whole season’s worth of data.  The best I can come up with is that there is a difference somehow in blocking the two situations, but that is the best I can come up with.

To quote a popular figure from my childhood, “Why is it so?”

 

 

Shooting Blind – A Life Without Feedback

This week my club president invited the team and staff for a casual get together / get to know you / team building activity at a local gun club.  If you think about it, it is a logical place to hold a team get together.  I mean what brings a group of men more enjoyment than shooting stuff?  Oh, you can think of a few things, eh? Well, anyway that is where we went.  After struggling for a few minutes with the personal morality of shooting a gun at all (particularly as I don’t want my son to have even a toy gun), I decided to join in.  It was an interesting experience.

The first problem I had was that I wasn’t wearing my glasses.  This wasn’t an issue about seeing the target, or not, but an issue of not being able to see what I hit.  I took careful aim at the target, carefully squeezed the trigger and off in the distance there was a cloud of dust.  I had no clue whether I had hit anything in between those events.  I was shooting blind.  I realised that without being able to see the target I had no feedback on what I was doing.  Between series I was able to see the target and eventually piece together some information.

20160913_161314

In the picture on the right you can see the 8 and 6 below the bullseye were in my 3rd series.  The 10, 9 and 8 were in my 4th series.  With feedback, I could quickly improve.

Oddly, considering how many millions of times I have seen it, I am much better able to ‘see’ where a ball lands after having spent a season working with the video challenge system in the Polish League.  For the first time in my over 30 year involvement with volleyball I have had actual feedback on where a ball has landed.  It turns out that is important too.  Who would have thought.

The lesson is, as always, there is no learning without feedback.