Possession analysis: Time vs Passes

Manipulating the Data

Taking the event level data I loaded earlier, I filter it for pass events only and then count the number of passes per team. The final chart will be split into the first and second halves, so I also split the dataset into two parts.

# Pull the pass events from the base dataset
pass_poss_tib <- match_1_clean %>%
  filter(type.name == "Pass") %>%
  select(ElapsedTime, team.name, type.name, pass.outcome.name, period) %>%
  group_by(team.name) %>%
  mutate(pass_count = row_number(team.name)) %>%
  select(-pass.outcome.name) %>%
  ungroup()

# Set dataset for first half
pass_poss_tib_1 <- pass_poss_tib %>%
  filter(period == 1)

# Extract the adjustment amount to reset the start time
# for the second half to 2,700 seconds (45min)
adj_amt_2 <- pass_poss_tib %>%
  filter(period == 2) %>%
  filter(team.name == "Tottenham Hotspur") %>%
  filter(pass_count == 291) %>%
  pull(ElapsedTime) - (45 * 60)

# Set data for second half
pass_poss_tib_2 <- pass_poss_tib %>%
  filter(period == 2) %>%
  mutate(ElapsedTime = (ElapsedTime - adj_amt_2))

Creating the chart

The chart consists of two ggplot2 charts displayed together using the gridExtra package. Firstly, I create the two ggplot2 charts, one for each half. I reformat the x-axis, so it displays in minutes and I scale the y-axis, so that it is scaled for both halves and is only displayed in the first chart, which will be on the left of the final plot. I only keep the legend in the second plot.

# Create the left hand side of the chart
# Plotting the count of passes per team in the first half
pass_poss_plot_1 <- pass_poss_tib_1 %>%
  ggplot(aes(x = ElapsedTime)) +
  geom_line(aes(y = pass_count, colour = team.name), size = 1) +
  scale_color_manual(values = setNames(
    c(home_team_colour, away_team_colour),
    c(home_team, away_team)
  )) +
  scale_x_time(
    breaks = c(0, (15 * 60), (30 * 60), (45 * 60)),
    labels = c(0, (15), (30), (45)),
    limits = c(0, max(pass_poss_tib_1$ElapsedTime))
  ) +
  scale_y_continuous(expand = c(0, 0), limits = c(0, 575)) +
  theme(
    legend.position = "none",
    axis.ticks.y = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.major.x = element_line(color = gridline_colour),
    panel.grid.major.y = element_line(color = gridline_colour),
    panel.grid.minor = element_blank(),
    panel.background = element_blank()
  ) +
  labs(x = "First Half", y = "Pass Count")

# Create the right hand side of the chart
# Plotting the count of passes per team in the second half
# This part of the plot also includes the legend
pass_poss_plot_2 <- pass_poss_tib_2 %>%
  ggplot(aes(x = ElapsedTime)) +
  geom_line(aes(y = pass_count, colour = team.name), size = 1) +
  scale_color_manual(values = setNames(
    c(home_team_colour, away_team_colour),
    c(home_team, away_team)
  )) +
  scale_x_time(
    breaks = c((45 * 60), (60 * 60), (75 * 60), (90 * 60)),
    labels = c((45), (60), (75), (90)),
    limits = c(
      min(pass_poss_tib_2$ElapsedTime),
      max(pass_poss_tib_2$ElapsedTime)
    )
  ) +
  scale_y_continuous(expand = c(0, 0), limits = c(0, 575)) +
  theme(
    legend.key = element_rect(fill = "white"), # background of legend icon
    axis.title.y = element_blank(),
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.major.x = element_line(color = gridline_colour),
    panel.grid.major.y = element_line(color = gridline_colour),
    panel.grid.minor = element_blank(),
    panel.background = element_blank()
  ) +
  labs(x = "Second Half", colour = "Team")

I now have two charts that need to be combined into one final chart:

• pass_poss_plot_1 (first half chart)
• pass_poss_plot_2 (second half chart)

# Using the gridExtra package arrange the two plots into one chart
grid.arrange(pass_poss_plot_1, pass_poss_plot_2,
  ncol = 2, widths = c(3.8, 5.2), # c(3.5,3,1),
  top = textGrob("Number of passes per team", gp = gpar(fontsize = 14))
)

Manipulating the Data

In the second part of this post I create the datasets containing the cumulative amount of time across all the events for each team. Again, taking the event level data, I calculate the start and end time of each possession and I delete all the events for each possession except the first and last events. I also delete possessions that are zero seconds long. I then calculate the cumulative time in possession for each team. As the final chart will again be split into two halves, I do the same with the data.

# Calculate the cumulative time in possession for each team
# A given possession can be made up of many events
# I am only interested in the time of the first and last events of each possession
poss_tib <- match_1_clean %>%
  select(
    index, timestamp, minute, second, period, possession, possession_team.id,
    possession_team.name, duration, type.name, play_pattern.name,
    ElapsedTime, StartOfPossession, TimeInPoss, TimeToPossEnd
  ) %>%
  mutate(EventFlag = case_when(
    TimeInPoss == 0 ~ "StartTime",
    TimeToPossEnd == 0 ~ "EndTime",
    TRUE ~ "Event"
  )) %>% # distinguish between first, last
  # and middle events for a given possession
  select(
    possession, possession_team.name, EventFlag,
    ElapsedTime, TimeInPoss, TimeToPossEnd, period
  ) %>%
  filter(!EventFlag == "Event") %>% # remove the non first or last events
  mutate(TimeDiff = (TimeInPoss - TimeToPossEnd)) %>%
  filter(!TimeDiff == 0) %>% # remove the possessions that are zero seconds long
  group_by(possession, possession_team.name, EventFlag, ElapsedTime, period) %>%
  count() %>% # this puts mutiple entries with the same ElapsedTime on the sane row
  ungroup() %>%
  select(!n) %>% # remove the count
  arrange(ElapsedTime) %>% # make sure it is ordered correctly
  pivot_wider(
    names_from = EventFlag,
    values_from = ElapsedTime
  ) %>% # StartTime and EndTime as columns
  filter(!is.na(StartTime)) %>% # remove the NA; first row of 2nd half
  mutate(PossTime = EndTime - StartTime) %>%
  group_by((possession_team.name)) %>%
  mutate(CumTime = cumsum(PossTime)) %>%
  ungroup() %>%
  select(StartTime, CumTime, possession_team.name, period)

# Extract the adjustment amount to reset the start time
# for the second half to 2,700 seconds (45min)
adj_amt <- poss_tib %>%
  filter(period == 2) %>%
  filter(CumTime == 1002.514) %>% # first row of the 2nd half
  pull(StartTime) - (45 * 60)

# Split the data out for the first half into a mini dataset
time_poss_tib_1 <- poss_tib %>%
  filter(period == 1) # keep the first half only

# Split the data out for the second half into a mini dataset
# Adjust the start time of the seconf half back to 45mins (2,700 seconds)
time_poss_tib_2 <- poss_tib %>%
  filter(period == 2) %>% # keep the second half only
  mutate(StartTime = (StartTime - adj_amt)) # %>% # adjust the StartTime
# so it starts at 2,700 seconds

Creating the chart

To create the time based possession chart, I follow a very similar process to the one I used for the pass possession chart above. Additionally, I format the y-axis to be displayed in minutes rather than seconds.

# Create the left hand side of the chart
# Plotting the amount of time in possession per team in the first half
time_poss_plot_1 <- time_poss_tib_1 %>%
  ggplot(aes(x = StartTime)) +
  geom_line(aes(y = CumTime, colour = possession_team.name), size = 1) +
  scale_color_manual(values = setNames(
    c(home_team_colour, away_team_colour),
    c(home_team, away_team)
  )) +
  scale_x_time(
    breaks = c(0, (15 * 60), (30 * 60), (45 * 60)),
    labels = c(0, (15), (30), (45)),
    limits = c(
      min(time_poss_tib_1$StartTime),
      max(time_poss_tib_1$StartTime)
    )
  ) +
  scale_y_time(
    breaks = c(
      0, (5 * 60), (10 * 60), (15 * 60), (20 * 60),
      (25 * 60), (30 * 60), (35 * 60), (40 * 60)
    ),
    labels = c(0, (5), (10), (15), (20), (25), (30), (35), (40)),
    limits = c(0, 2200)
  ) +
  theme(
    legend.position = "none",
    axis.ticks.y = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.major.x = element_line(color = gridline_colour),
    panel.grid.major.y = element_line(color = gridline_colour),
    panel.grid.minor = element_blank(),
    panel.background = element_blank()
  ) +
  labs(x = "First Half", y = "Cumulative Time in Possession")

# Create the right hand side of the chart
# Plotting the amount of time in possession per team in the second half
# This part of the plot also includes the legend
time_poss_plot_2 <- time_poss_tib_2 %>%
  ggplot(aes(x = StartTime)) +
  geom_line(aes(y = CumTime, colour = possession_team.name), size = 1) +
  scale_color_manual(values = setNames(
    c(home_team_colour, away_team_colour),
    c(home_team, away_team)
  )) +
  scale_x_time(
    breaks = c((45 * 60), (60 * 60), (75 * 60), (90 * 60)),
    labels = c((45), (60), (75), (90)),
    limits = c(
      min(time_poss_tib_2$StartTime),
      max(time_poss_tib_2$StartTime)
    )
  ) +
  scale_y_time(
    breaks = c(
      0, (5 * 60), (10 * 60), (15 * 60), (20 * 60),
      (25 * 60), (30 * 60), (35 * 60), (40 * 60)
    ),
    labels = c(0, (5), (10), (15), (20), (25), (30), (35), (40)),
    limits = c(0, 2200)
  ) +
  theme(
    legend.key = element_rect(fill = "white"),
    axis.title.y = element_blank(),
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.major.x = element_line(color = gridline_colour),
    panel.grid.major.y = element_line(color = gridline_colour),
    panel.grid.minor = element_blank(),
    panel.background = element_blank()
  ) +
  labs(x = "Second Half", colour = "Team")

I now have two charts that need to be combined into one final chart:

• time_poss_plot_1 (first half chart)
• time_poss_plot_2 (second half chart)

# Using the gridExtra package arrange all the plots into one plot
grid.arrange(time_poss_plot_1, time_poss_plot_2,
  ncol = 2, widths = c(3.8, 5.2), # c(3.5,3,1),
  top = textGrob("Time in possession", gp = gpar(fontsize = 14))
)

As expected both charts follow a similar pattern. For possession measured by number of passes, Spurs finished with 564 passes (63%) and Liverpool had a total of 326 passes (37%). Whereas, for cumulative time in possession of the ball the final breakdown is Spurs with 36m 05s in possession of the ball (68%) and Liverpool with 17m 22s (32%).