Has average rally length really decreased?

[anjoooo]

I've been watching a series of matches from the Gilbert Cup in 1996. I assume that this is before the speed glue ban as well as when they were still playing with the 38mm celluloid balls (forgive me, I wasn't playing table tennis back then and am not sure of when these things came into effect).

Here's one such match between Kim Taek Soo and Saive: https://youtu.be/pzmVXT0jYBQ

Very enjoyable matches to watch as the camera angle really shows how much the ball used to curve in the air and how fast the 38mm ball was.

One thing I noticed is that the length of the rallies seemed to be quite long. Perhaps I am overestimating the length of the rallies but there are lots of nice loop to loop rallies.

I've heard people say that the average length of rallies has decreased since they got rid of speed glue and increased the size of the ball. However, when I watch matches nowadays it seems to me that rallies are very short when compared to the matches I mentioned from the 90s above. One theory that I have is that the larger ball (and perhaps no speed glue) has indeed decreased the speed of shots and so players are able to get in closer to the table as they have a bit more time to react. This could have the effect of actually shortening rallies as players are now so close to the table that they are either playing more outright winners or making more errors.

Additionally, from what I understand, the old 38mm ball was easier to loop and create spin with and so my theory is that perhaps players were able to stay in rallies longer as they could use spin to help keep the ball on the table?

Another possibility is that players are now using the banana flick to end rallies outright from the serve.

Thoughts? Is this just observation bias on my part?

 

[Loopadoop]

My guesses

Pro tour - yes

Amateur club level - no, serve return is a lot easier now.

 

[Hysteresis]

I believe statistically, rallies are actually slightly longer now.

While returns have gotten alot more aggressive, you have to keep in mind that people used to outright fail to return the serve ALOT.

There is an actual academic paper by Zoran Djokic if you google it, addressing serve/return in TT, where he says serves are some 21% of all strokes in the game.

Given how few, statistically, of the strokes in the game are in long rallies (longer than 4/5 strokes total), the shift of rallies from failing to return serve, to outright winning on receive is actually a net increase in rally length (2 strokes instead of 1 on a significant number of points).

 

[Zaccai]

Interesting. I have only played with old recreational celluloid balls as a novelty, as a newer player to the game I am unfamiliar with boosting and celluloid balls, and the 38mm ball.

 

[Lula]

Should be longer. Bigger ball and not hide the ball was changed because of it.

But at the same time. With less spin in the plastic ball they can attack the serve much more aggressive and Maybe because of this win the point earlier?

 

[yoass]

That footage shows exceptional play.

Way back when 38mm and speedglue, it was rather rare to see anything after the 3rd ball. Many outright serve points, the overwhelming rest 3rd ball kills.

Our memory is treacherous and these relatively quick and unspectacular exchanges are eclipsed by the more riveting, anomalous ones.

This is a bit hyperbolic snd simplistic. The speedglue ban, larger ball, and then the plastic one took out a few of the game’s teeth. Serves are less dangerous, so the 2nd ball has become more forceful. The 3rd ball is less devastating.

I am no stranger to nostalgia and miss some of that, but even so I rather like the extra measure of tenacity, smarts, and focus needed to finish rallies now.

 

[zeio]

We had an extremely fun series of tournaments in Los Angeles in the late '90s.
The 1996 Gilbert Cup was played at the Beverly Hilton, and the participants were J.O. Waldner, Zoran Primorac, Andrzej Grubba, Kim Taek Soo, Jean Michel Saive, Philip Saive, Erik Lindh, Chen Xinhua, Patrick Chila and some of the top players in the US at the time: Cheng Yinghua, Khoa Nguyen and Jim Butler. The prize money generated an entertaining mix of competition and exhibition.

To give you an idea of the pinnacle of the 38mm era:

 

[yogi_bear]

Opposite. Bigger ball is slower and has lesser spin equals to longer and more returns.

 

[kenshiro]

in teory it should be slower and longer plays but i see table tennis is totaly diferent todays, its faster and the ralliess in general a bit shorter, moore close to the table and players are using way moore the backhand fast play. after watching some good old late 90s and beggining 2000s games i tink also penholders players had a super forehand comparing to the new genaration of penhoders. this ball is good for short pips, maybe will apear moore players like mattias falck and zan jian

 

[anjoooo]

I can't believe it. I completely mistyped what I really meant to say. I meant to say "Has average rally length really increased?"

Sorry for the confusion! My post probably makes more sense now.

I keep hearing that rally length has increased because of the larger slower ball and no speed glue. But I feel like it's actually shorter after watching the videos in my post above. Also nowadays people are so aggressive with the banana flick making rallies even shorter.

 

[Hysteresis]

I can't believe it. I completely mistyped what I really meant to say. I meant to say "Has average rally length really increased?"

Sorry for the confusion! My post probably makes more sense now.

I keep hearing that rally length has increased because of the larger slower ball and no speed glue. But I feel like it's actually shorter after watching the videos in my post above. Also nowadays people are so aggressive with the banana flick making rallies even shorter.

I think it's just a disconnect of expectation to reality.

The average length of the rally has increased, but not by alot, and in not a way that creates a lot more actually long rallies.

My original post still stands, the major factor in the increase in length of rallies, is that there are far fewer outright errors on receive because serves spin less and bounce higher (also because of rules of no hiding the ball during serve). This does mean receives are now more aggressive, so doesn't really create more long rallies. However, because serve and 3rd ball attack were already such an oppressive part of the game's meta anyway, the shift of advantage away from the server actually creates a longer average rally just from the return hitting the table more.

I don't think incremental changes can ever create a meta with "more longer rallies". The meta has immediately adjusted to the weaker serves with a stronger, more aggressive return of serve because it is not the goal of elite professional players to keep the ball in play and have a good long hit, it's their goal to find the best way to win the point, and that means seizing the first opportunity to end the point in their favour.

But statistically, the ease of long rallies is almost irrelevent. We have on average more strokes because we have moved to a "very short rally" with more strokes than the "very short rally" that we used to have.

 

[UpSideDownCarl]

To give you an idea of the pinnacle of the 38mm era:

Zeio, this shows the issue really well.

 

[usualsuspect]

I believe statistically, rallies are actually slightly longer now.

While returns have gotten alot more aggressive, you have to keep in mind that people used to outright fail to return the serve ALOT.

There is an actual academic paper by Zoran Djokic if you google it, addressing serve/return in TT, where he says serves are some 21% of all strokes in the game.

Given how few, statistically, of the strokes in the game are in long rallies (longer than 4/5 strokes total), the shift of rallies from failing to return serve, to outright winning on receive is actually a net increase in rally length (2 strokes instead of 1 on a significant number of points).

This is a pretty sound argument. The percentage of successful service returns has increased significantly since switching to the 40+ balls. This change alone will noticeably increase the average number of hits played per point.

 

[Lula]

Do you guys think todays game if of benefit for the asians that train very much from the beginning? My thought is that with the old ball, and the first ball there was much more spin so you could be less of a machine? And maybe have "feeling" for the ball and spin that is not as easily to just train in? the same with the hidden serves, need somewhat of a talent to become really good at it.

I mean that nowdays it is more pang pang and less finesse i feel. Feel like this should benefit the ones that just train like robots or machines. But of course you need talent and feed for the ball nowdays too, so maybe i am just thinking and speculating completely wrong haha

 

[NextLevel]

Do you guys think todays game if of benefit for the asians that train very much from the beginning? My thought is that with the old ball, and the first ball there was much more spin so you could be less of a machine? And maybe have "feeling" for the ball and spin that is not as easily to just train in? the same with the hidden serves, need somewhat of a talent to become really good at it.

I mean that nowdays it is more pang pang and less finesse i feel. Feel like this should benefit the ones that just train like robots or machines. But of course you need talent and feed for the ball nowdays too, so maybe i am just thinking and speculating completely wrong haha

I think if more physically oriented played the sport, the ge.would have become like it is today, no matter what ball was used. Wang Liqin also got to World #1 during the 39mm ball era, it was team politics not the ball that may have slowed his rise to dominance.

 

[igorponger]

[size=+1] Rallies did get shortened indeed. Since the crazy speedy rubbers causing much more lapses by the players.[/size]

 

[zeio]

Now for some numbers. Data collected from QF, SF and F for Men's World Cup 2014-2018.

2014-2016: DHS 40+(Cellulose acetate)
2017-2018: DHS D40+(ABS)

Definitions of relevant statistical concepts:
(1) Rally: The period during which the ball is in play. A rally starts from a serve and ends when a point is scored; (2) Number of strokes per rally: the total number of strokes hit by either player in one rally; (3) Nth scoring/losing stroke: points scored/lost on a player's nth stroke. Among them, the 0th scoring stroke means the opponent's service error, and there is no 0th losing stroke.

https://i.imgur.com/nRhFOof.jpg
Year/Matches/Games/Rallies(Points)/Total strokes/Average rallies per game/Average strokes per rally
Table 1: Number of strokes for 2014-2018 Men's World Cup

https://i.imgur.com/ZlfBVRa.jpg
Figure 1: Frequency distribution of the number of strokes per rally

https://i.imgur.com/9LZrqUj.jpg
Year/Max strokes/Times/Max strokes that happened more than 7 times/Number of strokes that happened the most(1st/2nd/3rd/4th/5th/%)
Table 2: Characteristics of the strokes

https://i.imgur.com/OAcKtPI.jpg
X-axis: strokes per rally
Y-axis: average frequency per game
Figure 2: Statistical chart for the average frequency per game for 2-6 strokes per rally

https://i.imgur.com/lAD4Wdg.jpg
Blue: Winning players
Brown: Losing players
X-axis: Scoring stroke
Y-axis: Frequency
Figure 3: Frequency distribution of the scoring stroke for the winning and losing players

https://i.imgur.com/HGTjOHP.jpg
Blue: Winning players
Brown: Losing players
Black: Overall
Figure 4: Statistical chart for the scoring percentage per stroke for the winning and losing players

https://i.imgur.com/RmUPJ59.jpg
Table 3: Pearson's correlation for points won and lost per stroke
Note: ** indicates significant correlation at .01 level (both sides); * indicates significant correlation at 0.05 level (both sides)

https://i.imgur.com/iztiuyQ.jpg
Red: Points won
Black: Points lost
Figure 5: Logical diagram of the correlation for points won and lost per stroke

 

[SFF_lib]

Now for some numbers. Data collected from QF, SF and F for Men's World Cup 2014-2018.

2014-2016: DHS 40+(Cellulose acetate)
2017-2018: DHS D40+(ABS)

Definitions of relevant statistical concepts:
(1) Rally: The period during which the ball is in play. A rally starts from a serve and ends when a point is scored; (2) Number of strokes per rally: the total number of strokes hit by either player in one rally; (3) Nth scoring/losing stroke: points scored/lost on a player's nth stroke. Among them, the 0th scoring stroke means the opponent's service error, and there is no 0th losing stroke.

Year/Matches/Games/Rallies(Points)/Total strokes/Average rallies per game/Average strokes per rally
Table 1: Number of strokes for 2014-2018 Men's World Cup

Figure 1: Frequency distribution of the number of strokes per rally

Year/Max strokes/Times/Max strokes that happened more than 7 times/Number of strokes that happened the most(1st/2nd/3rd/4th/5th/%)
Table 2: Characteristics of the strokes

X-axis: strokes per rally
Y-axis: average frequency per game
Figure 2: Statistical chart for the average frequency per game for 2-6 strokes per rally

Blue: Winning players
Brown: Losing players
X-axis: Scoring stroke
Y-axis: Frequency
Figure 3: Frequency distribution of the scoring stroke for the winning and losing players

Blue: Winning players
Brown: Losing players
Black: Overall
Figure 4: Statistical chart for the scoring percentage per stroke for the winning and losing players

Table 3: Pearson's correlation for points won and lost per stroke
Note: ** indicates significant correlation at .01 level (both sides); * indicates significant correlation at 0.05 level (both sides)

Red: Points won
Black: Points lost
Figure 5: Logical diagram of the correlation for points won and lost per stroke

Awesome data. Is there any similar analysis done in the 40mm era, speed glue era and 38mm era?

Sent from my PAR-LX9 using Tapatalk

 

[zeio]

The author noted that:
1. Aggressive receive has displaced serve-and-attack as the core of table tennis tactics.
2. Points lost when receiving are higher now because of that.
3. Points lost when returning long serves were less than 6% during 2015-2017, but 9% in 2018.
4. For 2017-2018, average strokes per rally have slightly decreased for 6 strokes and below, but have slightly increased for 7 strokes and beyond.
5. 3rd ball kill still plays a pivotal role despite all that.
6. 5th and 6th strokes are the transition point b/w initiation of attack and sustained rally. They're also the critical point b/w offense and defense, after which the winning player has a higher scoring percentage on 7th and 9th stroke.

 

[zeio]

Awesome data. Is there any similar analysis done in the 40mm era, speed glue era and 38mm era?

Sent from my PAR-LX9 using Tapatalk

Here's one from Japan for London 2012 for both men's and women's events.

https://jcoachings.jp/jcoachings2012/wp-content/uploads/2016/03/a3458153ee64b0e377db620e80d394a3.pdf

Characteristics of world-class rally in table tennis:
Focusing on the number of shots played per rally

Kazuto Yoshida[sup]1)[/sup]，Koshi Yamada[sup]2)[/sup]，Sho Tamaki[sup]3)[/sup]，Hisashi Naito[sup]4)[/sup] and Masaru Kaga[sup]5)[/sup]

Abstract

A notational analysis of 12 , 428 rallies in 149 singles matches at the London Olympic Games was conducted to clarify the characteristics of world-class rallies in table tennis. The number of shots played per rally, which was the sum of a correct service and correct returns, was measured. And the winning ratios of server and receiver were determined by the number of shots played per rally (Odd numbers indicate the server won the point, even numbers indicate the receiver won the point). The playing styles of the players were classified into all-round types and defensive types (chopper type), and the types of matches were classified into 3 by the combination of playing styles as: all-rounder vs. all-rounder (AA type), all-rounder vs. defensive (AD type) and defensive vs. defensive (DD type). The numbers of shots and the two ratios, and relations to both gender and type of match were considered. The DD type was excluded from this analysis because there were too few measured data. As a result, the mode of the number of shots for men and women was 3 each. The maximums for men and women were 38 and 60 , respectively. The minimum for men and women were 0 each. The mean number of shots for women was significantly larger than that for men for AD type (p< 0 . 01 ), and that for AD type was significantly larger than that for AA type for both men (p< 0 . 01 ) and women (p< 0 . 001 ). The mean winning ratios for servers for both men and women for AA type and AD type were all at the same level at about 55 ％, which were significantly higher than those for receivers of about 45 ％ (p< 0 . 01 ). Additionally, the evaluation criteria for the mean number of shots in a match, and those for the winning ratios of a server and a receiver were proposed. It was suggested that these results could be valuable for table tennis coaching.

Here's one from Germany. Abstract only.

http://www.ittf.com/wp-content/uploads/2016/10/Book_of_Abstracts_Suzhou_2015.pdf

Michael Fuchs and Martin Lames
Technische Universität München, Germany

RALLY LENGTH IN TOP LEVEL TABLE TENNIS

Abstract
Problem: In table tennis, the position of a stroke in the rally stands for tactical meanings and options e.g. the second stroke (return) has to deal with quality of service and to give a response ideally in creating an advantage. Therefore, rally length and success/failure rates of different strokes are of particular interest in studying the structure of table tennis (Zhang, Liu, Hu & Liu, 2013).
Method: The sample consisted of 81 women’s and 141 men’s single matches (only matches between offensive players). The matches were chosen from final stages of world table tennis top events. Data were collected by recently developed match analysis software for real-time match analysis in table tennis. Impact of sex, rankings and service on the rally length was statistically analysed.
Results: Whereas there were no differences in rally length in matches between different performance levels of male athletes (F=0.906, p=0.479, eta2=0.032) there were significant differences in female athletes (F=3.811, p=0.004, eta2=0.203). We also found a significantly higher rally length in female compared to male table tennis. This difference is mostly due to differences between the sexes in matches of top class players, e.g. top class vs. top class: mean male=4.73, mean female=5.79, t=4.112, p=0.002. In matches between lower ranked athletes there were no significant differences. Conclusion: Rally length is a first and crude variable to characterize table tennis matches. We found several characteristic relationships to start from in further studies.

Here's one from China that compared the matches and training between Jan, 2013 and May, 2016. Abstract only.

Ruizhi Liu[sup]1[/sup]
[sup]1[/sup]China Table Tennis College, Shanghai University of Sport, China
Correspondence: [email protected]

THE COMPARATIVE RESEARCH ON THE FEATURES OF MATCH AND TRAINING FOR MAIN
PLAYERS OF CHINA TABLE TENNIS FEMALE TEAM — BASED ON CELLULOID AND PLASTIC
BALLS

Abstract
The ITTF passed the resolution that plastic balls would be used at all world title and ITTF sanctioned events from 1st July 2014 before Paris world championships, 2013. From subjective and objective reflection of Chinese national team players and coaches, this small adjustment has made significant changes on technique and tactic.
The paper analysed the statistics of matches and training for five main players of China table tennis female team (Li Xiaoxia, Ding Ning, Liu Shiwen, Zhu Yuling, Chen Meng), which playing with celluloid and plastic balls in RIO Olympic period from January 2013 to May 2016. 132 matches were randomly chosen from world important tournaments while 917 training data were from closed trainings preparation for world major events. All the opponents were limited in world top 35 based on the world ranking in June 2016. All the training partners were Chinese table tennis female players.
Through using five phases’ evaluation method and analysis of variance, this research made 3 kinds of comparative analysis: Matches analysis between celluloid and plastic balls, training analysis between celluloid and plastic balls, performance differences analysis between matches and training with celluloid and plastic balls.
Conclusions as follows: (1) The phases of attack after serve, attack after receive still were the wining factors of China table tennis female team, which always kept high scoring rate and usage in tournaments; (2) The rally length of each point with plastic balls became shorter than celluloid, and appeared a tendency towards the first forth strokes from long rally. (3) The performance diversity between matches and training with plastic balls was smaller than celluloid, players performed more stable.

Studies for the transition from 38mm to 40mm are hard to come by. Here's one from Japan comparing the All Japan Championships 1993 and 2000.

https://www.ittfeducation.com/effect-40mm-diameter-ball-table-tennis-rallies-elite-players/

The effect of the 40mm diameter ball on
table tennis rallies by elite players

Toshiko TAKEUCHII, Yoshio KOBAYASHII, Shuichi HIRUTA
and Nobuo YUZA
Chukyo University, Japan; Nagoya University, Japan and Chukyo
Women 's University, Japan

Abstract This study dealt with the influence of the ball's mass and size
on the number of hits per point (rally hit counts) in table tennis
matches. The difference was studied between the conventional 38mm
diameter ball that is 2.5 grams in weight, and the currently adopted
40mm diameter ball that is 2.7 grams in weight. Rally hit counts for
the 38mm ball were conducted in the All Japan Championships in
1993 and those for the 40mm ball in the same tournament in 2000. The
rally hit counts for the 38mm ball were taken from the quarterfinals
and the semifinals in both men's and women's singles. The rally hit
counts for the 40mm ball were taken from the quarterfinals and the
semifinals in both men's and women's singles and from the
quarterfinals and the semifinals in both junior men's and junior
women's singles. The defensive players' matches have been excluded.
The mean of rally hits per point for 38mm and 40mm balls
respectively was 3.1 vs. 4.1 in men, and 3.8 vs. 4.6 in women. The
same counts were 3.0 vs. 3.3 in junior men and 3.6 vs. 4.6 in junior
women. The combined means for men and junior men, and women and
junior women were 3.1 vs. 3.8 and 3.7 vs. 4.6 for 38mm and 40mm
balls, respectively. The frequency distribution for the rally hit counts in
all groups for both the 38mm and 40mm balls were concentrated
toward the low end of the scale. In conclusion, the rallies in elite
players during table tennis matches are lengthened with a 40mm ball,
which may result from slower flight and spin characteristics of the
larger ball.