There are at least two ways to measure the length of rallies in a volleyball match. The first is obviously by time. This is obviously problematic. How do you define the start and finish of the rally? Obviously the whistle. But then you include the several seconds before the serve, including the seconds the toss is in the air, and the several seconds that the ball is in the air after it bounces off a person. That doesn’t seem like a true reflection of what is in a rally.
Another way is to count the number of net crosses in a rally. That seems like it would also give a measure of how much volleyball there is. A ball going over the net as a free ball is not volleyball in any interesting sense, and shouldn’t be included. An attack that is covered and re-attacked is extra volleyball and should count more than a ball that simply crosses the net. So while we use the term ‘net crosses’, we are really counting the number of attacks per rally.
Now that definitions are out of the way, how many rallies are there in an international volleyball match? We (me and the gentlemen at Science Untangled) chose four full data sets of matches: 2004 Men’s Olympic Tournament, 2021 Men’s and Women’s Volleyball Nations Leagure and the 2021 (I refuse to say 2020) Men’s Olympic tournament. Here are the results.
|TOURNAMENT||AVE NET CROSSES|
|2004 Olympics M (38 matches)||2.08|
|2021 Olympics M (38 matches)||2.00|
|2021 VNL M (128 matches)||1.97|
|2021 VNL W (128 matches)||2.43|
From first glance we see some things we expect to see. International women’s volleyball has longer rallies than men’s volleyball, to the degree that on average every two rallies there is an extra net cross. From watching matches at the 2004 and 2021 Olympics (here, here and here) we know that the game has evolved but the length of rallies is more or less unchanged. So if the game has changed but the rally length is the same / similar something must be going on.
(Aces + other non attacks)
(Serve + 1st attack)
(Serve, 1st attack + min 1 transition attack)
|2004 Olympics M||12.7||6.97||56.3||24.0|
|2021 Olympics M||18.4||4.84||52.0||24.8|
|2021 VNL M||18.9||6.05||50.7||24.4|
|2021 VNL W||10.4||6.48||47.3||35.8|
Considering just the two men’s Olympic tournaments, if you guessed that the difference was serving, you would have been correct. We would say that serving is much stronger now in men’s volleyball than in 2004 and that is manifested itself in more errors. But strangely not more aces. However, the number of rallies that ended after the first attack is greater in 2004 confirming that serving was in fact weaker (i.e. better reception = better first ball sideout). The number of longer rallies is very similar, but can we infer something about block and defence from these numbers? Why, yes we can.
Removing aces and service errors from our totals we can calculate the percentage of ‘extended’ rallies from the number of opportunities. In 2004, 29.9% of sideout attacks led to extra net crosses. In 2021, the number was 32.3%. Which suggests that block and defence has improved. For reference, the number for women’s VNL is 43.1%.
In conclusion, the average length of rallies in men’s volleyball is fairly stagnant despite the evolution of the game. Serving has improved, but receivers (and tactics, perhaps) allow fewer aces. Serving does have a positive affect on rally length by reducing the quality of reception (shown by the decrease in 2 net cross rallies), as does improved block and defence (shown by more multiple net crosses).
In extra conclusion, we can extrapolate what might happen if rule changes encouraged less pressure on the serve. In fact we don’t have to extrapolate. The numbers are there in 2004. Less service errors means more 2 net cross rallies and less 3 net cross rallies, and no overall change. The more things change the more they stay the same.
One extra, extra conclusion. The improvement in block and defence might be because of the power in serving. In which case in our thought experiment reducing service errors would end up shortening rallies.