Rally Length: Let’s Talk Turkey
Having a firm grasp of what’s happening on the court in terms of rally length is one of the building blocks of success in tennis. Players should be aware of whether or not they are winning or losing the shorter or the longer points. And, coaches need to be making their students aware of how their most effective style of play translates into rally length.
Craig O’Shannessy at Brain Game Tennis has done outstanding work in bringing rally length to the fore in tennis analytics. His key contribution has been to point out the higher success rates for the top players in the 0-4 rally length category. The 0-4 category is followed by 5-8 shots and 9+ shots. Often, O’Shannessy points out, the top players are either even or actually trail their opponents in winning share of the 9+ rallies.
However, it is not often appreciated that these categories are somewhat arbitrary and may sometimes not lead to the most striking conclusions from the raw data. For example, there is no a priori reason why the categories could not be 0-4, 5-10, and 11+. In what follows, I am going to break down the rally length into 4 categories, not 3.
The four categories are 0-1, 2-4, 5-8, and 9+. There is an a priori reason for breaking down the analysis in this way. The first category now isolates the serve and return of serve, arguably the two most important shots in tennis, into their own category, while also allowing for a more detailed examination of what happened within the 2-4 shot category. The 0-1 category would now include double faults, aces, service winners, and unforced errors on the return of serve. Using data from the Dominic Theim/Daniil Medvedev 2020 US Open semi-final, the results of changing the categories takes a fairly standard result and turns it into a REVELATION.
Here are the standard results:
|
Total: |
% of win |
|||||||
|
Thiem |
0-4 |
5- thru 8 |
9+ |
Total |
0-4 |
5- thru 8 |
9+ |
|
|
84 |
22 |
22 |
128 |
65.63% |
17.19% |
17.19% |
100.00% |
|
|
61 |
24 |
27 |
112 |
|||||
|
% of win |
||||||||
|
Med |
0-4 |
5- thru 8 |
9+ |
Total |
0-4 |
5- thru 8 |
9+ |
|
|
61 |
24 |
27 |
112 |
54.46% |
21.43% |
24.11% |
100.00% |
|
|
84 |
22 |
22 |
128 |
|||||
|
Thiem |
0-4 |
5- thru 8 |
9+ |
|||||
|
35.00% |
9.17% |
9.17% |
53.33% |
|||||
|
25.42% |
10.00% |
11.25% |
46.67% |
|||||
|
Within cat: |
||||||||
|
57.93% |
47.83% |
44.90% |
||||||
|
42.07% |
52.17% |
55.10% |
The table informs us of the following:
Overall, Theim won 128 points to Medvedev’s 112. He was dominant in the 0-4 category as O’Shannessy’s work would predict by 84 points to 61. Theim won 58% of the 0-4 points and was just below 50% in the other two categories. These results suggest looking within the 0-4 category to see what Theim did better.
Here are the REVELATION results:
|
Total: |
|||||
|
Thiem |
0-1 |
2 thru 4 |
5- thru 8 |
9+ |
|
|
35 |
49 |
22 |
22 |
128 |
|
|
36 |
25 |
24 |
27 |
112 |
|
|
Med |
0-1 |
2 thru 4 |
5- thru 8 |
9+ |
|
|
36 |
25 |
24 |
27 |
112 |
|
|
35 |
49 |
22 |
22 |
128 |
|
|
Thiem |
0-1 |
2 thru 4 |
5- thru 8 |
9+ |
|
|
14.58% |
20.42% |
9.17% |
9.17% |
53.33% |
|
|
15.00% |
10.42% |
10.00% |
11.25% |
46.67% |
|
|
Within cat: |
|||||
|
49.30% |
66.22% |
47.83% |
44.90% |
||
|
50.70% |
33.78% |
52.17% |
55.10% |
Obviously, the total number of points won/lost remains the same, but with the new categories, we get to see two eye-opening results: 1) Within the serve/return of serve category the players were essentially equal (35 vs. 36); 2) therefore, ALL of the 0-4 edge appeared in the 2-4 category (49 to 25). Theim had NEARLY DOUBLE THE NUMBER OF POINTS WON IN THIS CATEGORY! This information is far more valuable to both the player and the coach when looking to make improvements. Taking a fresh point of view makes for a new take on rally length. Let’s hope that more coaches will pursue it.
