NBA Wingspan & Performance Solutions

In professional basketball, physical traits can have a major impact on how a player performs. One key trait that often draws attention from scouts and analysts is wingspan, the distance from fingertip to fingertip with arms fully extended.

This worksheet explores the question: How does wingspan, especially wingspan relative to height, relate to player performance in the NBA?

You’ll analyze data combining NBA player profiles (including height and wingspan) with in-game statistics from the 2024–25 season. A key variable is wingspan_advantage: the difference between a player’s wingspan and height, often viewed as a potential edge in defense and rebounding, but a potential disadvantage in shooting.

Throughout this worksheet, you’ll clean and combine messy datasets, visualize relationships, and explore meaningful patterns, just like a data analyst working for an NBA front office. Your goal is to investigate whether a player’s wingspan advantage is linked to any performance metrics, and to think critically about how physical traits might (or might not) translate to on-court impact.

0. Load the Following Packages

library(tidyverse)
library(rvest)
library(janitor)

1. Load in CSV File

Load in the nba_wingspan_2025.csv from the data folder.

wingspan <- read_csv("nba_wingspan_2025.csv")

2. Clean the Dataset

Clean the dataset using the following packages from tidyverse: dplyr, tidyr, readr and stringr.

Once you are finished cleaning this dataset, these should be the following variables.

Variable	Type	Description
name	chr	Full name of the NBA player
team	chr	Three-letter abbreviation of the team the player is on
position	chr	Player’s primary on-court position in abbreviated form
height_inches	num	Player’s height in inches
wingspan_inches	num	Player’s wingspan in inches
wingspan_advantage	num	Difference between wingspan and height in inches (wingspan - height)

1. Convert the height variable (which is currently a character string like 6’4”) into a new numeric variable called height_inches that represents each player’s height in total inches. Hint using helper variables (such as separate variables for feet and inches) can make this process easier.

Click for solution

wingspan <-
  wingspan %>%
  separate_wider_delim(
    cols = height,
    names = c("feet", "inches"),
    delim = "'"
  ) %>%
  mutate(
    feet = parse_number(feet),
    inches = parse_number(inches),
    height_inches = feet * 12 + inches
  ) %>%
  select(-feet, -inches)

2. Do the same thing as above, but with the wingspan variable. Create a new numeric variable called wingspan_inches that represents each player’s wingspan in total inches.

Click for solution

wingspan <-
  wingspan %>%
  separate_wider_delim(
    cols = wingspan,
    names = c("feet", "inches"),
    delim = "'"
  ) %>%
  mutate(
    feet = parse_number(feet),
    inches = parse_number(inches),
    wingspan_inches = feet * 12 + inches
  ) %>%
  select(-feet, -inches) %>%
  relocate(wingspan_advantage, .after = wingspan_inches)

3. Extract the player’s position (e.g. “SG” or “C”) from the name variable and store it in a new variable called position, so that the position is no longer part of the name column.

Click for solution

wingspan <-
  wingspan %>%
  separate_wider_delim(
    cols = name,
    names = c("name", "position"),
    delim = " | "
  )

4. Remove players who are not currently on a team. In the name column, these players do not have a three-letter team abbreviation at the end of their name (e.g. “LeBron JamesLAL” vs. “LeBron James”). Then, split the name and team abbreviation into two separate variables: name and team. Hint: Use stringr functions to complete this step.

Click for solution

wingspan <-
  wingspan %>%
  filter(
    str_detect(name, pattern = "[A-Z]{3}$")
  ) %>%
  mutate(
    team = str_extract(name, pattern = "[A-Z]{3}$"),
    name = str_remove(name, pattern = "[A-Z]{3}$")
  ) %>%
  relocate(team, .after = name)

5. At this point, there may still be errors in the dataset, such as invalid or incorrectly extracted team abbreviations. To catch these, use the reference dataset of valid NBA team abbreviations: nba_team_abbreviations.csv.

teams <- read_csv("nba_team_abbreviations.csv")

• First use anti_join() to identify any rows in the wingspan dataset with invalid team names.

Click for solution

anti_join(wingspan, teams, by = "team")

• Then remove those rows from the dataset. Although this could be done with filters, instead practice using semi_join().

Click for solution

wingspan <- semi_join(wingspan, teams, by = "team")

3. Scrape Data From Basketball Reference

• Go to this link, Basketball-Reference: Per 100 Possesions and scrape the Per 100 Possessions data table.

Why Per 100 Possessions: Per 100 possessions stats are often preferred over per game stats in basketball because some teams have more possessions simply due to pace. These stats adjusts for that by standardizing performance across the same number of plays, making it easier to compare players fairly and evaluate efficiency and impact.

From the janitor library, use the help section to get more details on the functions.

Function	Description
clean_names()	Standardizes column names to be lowercase, snake_case, and R-friendly.
row_to_names()	Promotes a row (usually the first) to become the column names of the table.

1. Scrape the data from the Per 100 Possessions table.

Click for solution

page <- 
  "https://www.basketball-reference.com/leagues/NBA_2025_per_poss.html" %>%
    read_html() %>%
    html_nodes("table") %>%
    html_table()

per_100_poss <- page[[1]]

2. To tidy the dataset by completing the following steps:

First, standardize all variable names using the clean_names() function.

Second, remove duplicate player entries by keeping only the row that represents a player’s full season total. (For example, Luka Dončić was traded mid-season, so he might appear three times: once for his stats with DAL, once with LAL, and once for his total 2024–25 season stats.) Hint: Use group_by() and ungroup() to help identify duplicates. Also, games played will be the highest for rows that represent a full season total.

Optional: Use select() to keep only the variables you’re interested in. Also use rename() and relocate() to improve clarity in the dataset.

Click for solution

per_100_poss <- 
  per_100_poss %>%
  clean_names() %>%
  group_by(player) %>%
  slice_max(order_by = g, n = 1, with_ties = FALSE) %>%
  ungroup() %>%
  # Optional: use select, rename and relocate to improve clarity
  select(2, 6, 8, 18, 21, 22, 23, 24, 25, 26, 27, 30, 31, 32) %>%
  relocate(e_fg_percent, .after = d_rtg) %>%
  relocate(ft_percent, .after = e_fg_percent) %>%
  relocate(pts, .after = mp)

3. This table also contained a row for the League Average. Remove it from the dataset. (Tip: The only variables with non-missing entries for the League Average are e_fg_percent and ft_percent. This might help you more easily find it in the dataset.)

Click for solution

per_100_poss <-
  per_100_poss %>%
  drop_na(g)

4. Do the same steps as in parts a - c, but this time use the Shooting data table. Here is the link: Basketball-Reference: Shooting

Note: These tables contain two rows of column headers. To fix this issue, use the row_to_names() function. As a result some columns will have the same names, the clean_names() function fixes this issue.

Click for solution

url2 <- "https://www.basketball-reference.com/leagues/NBA_2025_shooting.html"
objs2 <-
  url2 %>%
  read_html() %>%
  html_nodes("table") %>%
  html_table()

shooting <- objs2[[1]]

shooting <-
  shooting %>%
  row_to_names(1) %>%
  clean_names() %>%
  group_by(player) %>%
  slice_max(order_by = g, n = 1, with_ties = FALSE) %>%
  ungroup() %>%
  # remove leagyue average
  filter(player != "League Average") %>%
   # Optional: use select, rename and relocate to improve clarity
  select(2, 9, 10, 11, 16, 17, 22, 25) %>%
  rename(
    avg_dist_of_fg = dist,
    `2pt_rate` = x2p,
    `3pt_rate` = x3p,
    `2pt_percent` = x2p_2,
    `3pt_percent` = x3p_2,
    `dunk_rate` = percent_fga
  ) %>%
  # needed to do this because of the use of row_to_names()
  mutate(
    across(2:8, parse_number)
  )

4. Combine the two cleaned datasets you scraped from Basketball Reference into a single dataset.

bball_ref <-
  per_100_poss %>%
  left_join(shooting, by = "player")

4. Combine Datasets

Combine the cleaned wingspan dataset from part 2, with the combined Basketball Reference dataset you created in part 3.e.

In the nba_wingspan_2025.csv file, sevearl names were misspelled compared to their spellings on Basketball Reference. Keep this in mind as you work through the questions.

1. Explain why anti_join would allow us to identify the players with misspelled names.

Click for solution

Because, when joined by the player’s name, it will identify rows that do not have a match.

2. Use anti_join() to identify the players in the wingspan dataset without a match in the merged data from 3.e.

Click for solution

wingspan %>%
  anti_join(bball_ref, by = c("name" = "player"))

3. Manually explore the player names in the merged data from 3.e (e.g., using the View function available with the R Studio IDE) to determine which players were misspelled and which players with wingspans are not in the performance statistics dataset. Summarize your findings here.

Click for solution

You should notice that Alperen Şengün and Jimmy Butler are the only two that should have a match but didn’t.

4. After comparing the differences, use mutate() to correct the two misspelled names in the wingspan dataset. Then, recombine the datasets and this time there should be no mismatches.

Click for solution

combined <- 
  wingspan %>%
  mutate(
    name = if_else(name == "Alperen Sengun", "Alperen Şengün", name),
    name = if_else(name == "Jimmy Butler III", "Jimmy Butler", name)
  ) %>%
  left_join(bball_ref, by = c("name" = "player"))

5. Explore the Combined Dataset

Use the newly created dataset to investigate potential relationships between a player’s physical traits and their performance on the court.

1. Create a histogram of the wingspan_advantage variable to see how common different levels of advantage (or disadvantage) are across players in the NBA. Provide a brief summary of the distribution.

Click for solution

combined %>%
  ggplot(., aes(x = wingspan_advantage)) +
  geom_histogram(binwidth = 1, color = "black")

Nearly all players have a larger wingspan than height (i.e., a positive advantage). The distribution is approximately normally distributed with a center around 4 inches. Most advantages are in the 1 - 8 inch range.

2. Create a scatterplot using wingspan_advantage as the explanatory variable and blk as the response variable. What kind of relationship, if any, do you observe? Are there any outliers?

Click for solution

combined %>%
  ggplot(., aes(x = wingspan_advantage, y = blk)) +
  geom_point() +
  geom_smooth(method = "loess", se = TRUE)

Not surprisingly, as the advantage increases there is a “Blocks per 100 possessions” statistic. While a steady/consistent trend, there is still plenty of variability associated with it though.

3. Create a scatterplot using wingspan_inches as the explanatory variable and 3pt_rate as the response variable. Include a regression line, a title, and labels for the x and y axes. Then separate the plot by position.What patterns or relationships stand out within or across positions? (Hint: Recall that 3pt_rate measures the proportion of a player’s shots that are 3 point attempt.)

Click for solution

combined %>%
  ggplot(aes(x = wingspan_advantage, y = `3pt_rate`)) +
  geom_point() +
  geom_smooth(method = "loess", se = TRUE) +
  facet_wrap(~position) +
  labs(
    title = "Relationship Between Wingspan and\n3PT Rate by Position",
    x = "Wingspan Advantage (inches)",
    y = "3PT Shooting Rate"
  )

Besides Point Guards (PG), there tends to be a slight negative relationship between the two variables. This is most pronounced in the Shooting Guard (SG) and Small Forward (SF) positions - both of which tend to be the higher scoring positions “outside shooters” on a team. An overall conclusion would be that 3pt shooter effectiveness decreases as the wingspan advantage increases. (The additional readings section of the module page provides some references that have some similar findings.) It is unclear why point guards would be immune to this, although they do tend to be the shortest players on a team - which might indicate that the longer arms may compensate for their shorter bodies, allowing them to more effectively shoot 3pointers.

4. Optional: Investigate any other combination of physical traits and on-court performance metric you find interesting. Create a visualization to explore the relationship and try to explain any patterns or outliers you observe.

Click for solution

Answers will vary