Investigating the Impact of Tournament Length and Surface Type on Win Percentages in Professional Tennis

Visualizations
Distribution
Interpretation
Describing distribution shapes and interpreting visualizations or Men’s Professional Tennis data.
Authors
Affiliation

Eric Seltzer

St. Lawrence University

Ivan Ramler

St. Lawrence University

Published

June 24, 2024

Module

Please note that these material have not yet completed the required pedagogical and industry peer-reviews to become a published module on the SCORE Network. However, instructors are still welcome to use these materials if they are so inclined.

Introduction

In professional tennis, there are four tiers of events, Grand Slams, Masters 1000, ATP 500 and ATP 250. Players receive the most points in their Association of Tennis Professionals ATP rankings for winning a Grand Slam, and the least from winning an ATP 250 tournament. There are four Grand Slams throughout the year, the Australian Open (Hard surface), Roland Garros (Clay surface), Wimbledon (Grass surface), and the U.S. Open (Hard surface). This leads to more competitive players wanting to play in the Grand Slams. In the ATP, Grand Slam tournaments are played in a best of five format, and non-Grand Slam tournaments in a best of three.

In Tennis, there are three different types of surfaces that are played on. The options are Grass, Hard, and Clay. The surfaces are important to keep track of as the speed of tennis changes, e.g., clay generally slows the ball down whereas grass speeds it up. Individual players perform better on certain surfaces and may favor playing on their preferred surface.

This dataset contains information for each player on each surface. To be included in the data set, players must have played a minimum of 10 matches overall or 5 matches on a particular surface. This data was filtered so only players who have recorded data on all three surfaces are present. This data is available in the file atp_2023.csv located in Data section below.

By the end of this activity, you will be able to:

  1. Understand the shapes of distribution as well as terms used to describe distribution.

  2. Be able to discuss open ended questions regarding why distributions might differ

  3. Compare differences between two populations through visualizations.

Technology requirement: The activity handout will require nothing more than the handout and a writing utensil to complete.

  1. Knowledge of distribution types and terms that come along with it.

  2. Knowledge of interpreting visualizations and what they represent.

Data

There is one data set for this exercise. The data is made up of 5194 rows and 20 columns and is provided by Jeff Sackman on GitHub. Each row represents a match played between two players.

Download non tidy data: atp_2023

Tennis

Variable Descriptions
Variable Description
tourney_name The name of the tournament
GS Whether this tournament was a grand slam or not
surface The type of surface that was played on for this match
result Whether this player was a winner or not
player The name of the player
rank The ATP ranking of the player
hand The handedness of the player
height The height of the player (cm)
country The country the player is representing
age The age of the player
minutes How long the match was in minutes
numAces The number of aces
numDF The number of double faults
servePoints The number of serve points
first_serve_won The number of serve points won
second_serve_won The number of second serve points won
serve_games_won The number of serve games won
first_serve_in The number of first serves in
break_points_saved The number of break points saved
break_points_faced The number of break points faced

The two summarized datasets used in the handout can be downloaded here.

Grand Slam

Download Grand Slam vs Non-Grand Slam data: grand_slam_summarized.csv

Variable Descriptions
Variable Description
player The name of the player
best_of Non-Grand Slam (best of 3) or Grand Slam (best of 5)
winPct win percentage (recorded as a proportion)
matches number of matches played (used to calculate winPct)

Surface Type

Download Surface Type Win Percentage data: court_surface_summarized.csv

Variable Descriptions
Variable Description
player The name of the player
surface The type of surface that was played on for this match
winPct win percentage (recorded as a proportion)
matches number of matches played (used to calculate winPct)

Data Source

Jeff Sackman

Materials

Class Handout

Class Handout - Solutions

By the end of the activity students should be able to see how longer format tournaments increase the variability in win percentages - implying that the best-of-five used in Men’s Grand Slams favors the better player.

Further, they can see how looking at just the top players (or, more specifically, just the players that actively play in many matches) can give interesting displays of distributions for win percentages. (By excluding the lesser talented professionals that have low win rates, but aren’t included in the given data.)