Capstone - Complete Analytics System

# Calculate basic defensive statistics
library(StatsBombR)
library(dplyr)

# Load data
comps <- FreeCompetitions() %>%
  filter(competition_id == 43, season_id == 106)
matches <- FreeMatches(comps)
events <- free_allevents(MatchesDF = matches)

# Defensive actions
defensive_stats <- events %>%
  filter(type.name %in% c("Tackle", "Interception", "Clearance",
                          "Block", "Duel")) %>%
  group_by(player.name, team.name) %>%
  summarise(
    matches = n_distinct(match_id),

    # Tackles
    tackles = sum(type.name == "Tackle"),
    tackles_won = sum(type.name == "Tackle" &
                      duel.outcome.name %in% c("Won", "Success")),

    # Interceptions
    interceptions = sum(type.name == "Interception"),

    # Clearances
    clearances = sum(type.name == "Clearance"),

    # Blocks
    blocks = sum(type.name == "Block"),
    shot_blocks = sum(type.name == "Block" &
                      block.offensive == FALSE),

    # Duels
    duels = sum(type.name == "Duel"),
    duels_won = sum(type.name == "Duel" &
                    duel.outcome.name %in% c("Won", "Success")),

    .groups = "drop"
  ) %>%
  mutate(
    # Rates
    tackle_success = round(tackles_won / tackles * 100, 1),
    duel_success = round(duels_won / duels * 100, 1),

    # Per 90
    tackles_per_90 = round(tackles / matches, 2),
    interceptions_per_90 = round(interceptions / matches, 2),
    clearances_per_90 = round(clearances / matches, 2),

    # Ball-winning actions
    ball_winning = tackles_won + interceptions,
    ball_winning_per_90 = round(ball_winning / matches, 2)
  )

print("Defensive Action Leaders:")
print(defensive_stats %>%
        filter(matches >= 3) %>%
        arrange(desc(ball_winning_per_90)) %>%
        select(player.name, team.name, matches, tackles_won,
               interceptions, ball_winning_per_90, tackle_success) %>%
        head(15))

# Possession-adjusted defensive metrics
# Calculate opponent possession to adjust

# First, get team possession by match
match_possession <- events %>%
  filter(type.name == "Pass") %>%
  group_by(match_id, team.name) %>%
  summarise(passes = n(), .groups = "drop") %>%
  group_by(match_id) %>%
  mutate(
    total_passes = sum(passes),
    possession_pct = passes / total_passes * 100,
    opp_possession_pct = 100 - possession_pct
  )

# Join to player defensive stats
player_match_defense <- events %>%
  filter(type.name %in% c("Tackle", "Interception")) %>%
  group_by(player.name, team.name, match_id) %>%
  summarise(
    defensive_actions = n(),
    .groups = "drop"
  ) %>%
  left_join(match_possession, by = c("match_id", "team.name"))

# Possession-adjusted rate
# More opponent possession = more opportunities to defend
poss_adjusted <- player_match_defense %>%
  group_by(player.name, team.name) %>%
  summarise(
    matches = n(),
    total_def_actions = sum(defensive_actions),
    avg_opp_possession = mean(opp_possession_pct),

    # Raw rate
    def_actions_per_match = total_def_actions / matches,

    # Adjusted: normalize to 50% opponent possession baseline
    adj_def_actions = total_def_actions / (avg_opp_possession / 50),
    adj_def_per_match = adj_def_actions / matches,

    .groups = "drop"
  ) %>%
  filter(matches >= 3) %>%
  mutate(
    adjustment_factor = round(50 / avg_opp_possession, 2)
  )

print("Possession-Adjusted Defensive Actions:")
print(poss_adjusted %>%
        arrange(desc(adj_def_per_match)) %>%
        select(player.name, matches, def_actions_per_match,
               avg_opp_possession, adjustment_factor, adj_def_per_match) %>%
        head(15))

# Calculate PPDA (Passes Per Defensive Action)
# Lower PPDA = more intense pressing

calculate_ppda <- function(events_df, team_name, match_id) {
  match_events <- events_df %>%
    filter(match_id == !!match_id)

  # Opponent passes in their defensive 60%
  opp_passes <- match_events %>%
    filter(team.name != team_name,
           type.name == "Pass",
           location.x <= 72)  # Opponent defensive 60%

  # Team defensive actions in opponent defensive 60%
  team_def_actions <- match_events %>%
    filter(team.name == team_name,
           type.name %in% c("Tackle", "Interception", "Foul Committed"),
           location.x >= 48)  # Same zone, flipped

  if (nrow(team_def_actions) == 0) return(NA)

  ppda <- nrow(opp_passes) / nrow(team_def_actions)
  return(ppda)
}

# Calculate PPDA for all teams across tournament
team_ppda <- events %>%
  select(match_id, team.name) %>%
  distinct() %>%
  group_by(team.name) %>%
  summarise(
    matches = n_distinct(match_id)
  )

# Simplified PPDA calculation
team_ppda_stats <- events %>%
  group_by(team.name) %>%
  summarise(
    matches = n_distinct(match_id),

    # Approximate PPDA
    opp_passes_faced = sum(type.name == "Pass" &
                           !is.na(team.name)),  # Would need opponent filter
    high_press_actions = sum(type.name %in% c("Tackle", "Interception",
                                               "Pressure") &
                             location.x >= 60),  # High press zone

    .groups = "drop"
  )

print("Team Pressing Intensity:")
print(team_ppda_stats %>%
        mutate(press_actions_per_match = round(high_press_actions / matches, 1)) %>%
        arrange(desc(press_actions_per_match)))

# Calculate xT (Expected Threat) prevented by defensive actions
# xT assigns values to pitch zones based on goal probability

# Simple xT grid (12x8 zones)
# Values represent probability of scoring from that zone
xt_grid <- matrix(c(
  0.01, 0.01, 0.01, 0.02, 0.02, 0.02, 0.02, 0.01,  # Row 1 (own goal)
  0.01, 0.01, 0.02, 0.02, 0.03, 0.03, 0.02, 0.01,
  0.01, 0.02, 0.03, 0.04, 0.05, 0.04, 0.03, 0.02,
  0.02, 0.03, 0.04, 0.06, 0.08, 0.06, 0.04, 0.03,
  0.03, 0.04, 0.06, 0.10, 0.15, 0.10, 0.06, 0.04,
  0.04, 0.06, 0.10, 0.20, 0.30, 0.20, 0.10, 0.06,
  0.08, 0.15, 0.25, 0.35, 0.40, 0.35, 0.25, 0.15,  # Row 7 (opp box)
  0.15, 0.30, 0.40, 0.50, 0.50, 0.50, 0.40, 0.30   # Row 8 (6-yard box)
), nrow = 8, byrow = TRUE)

# Function to get xT value from coordinates
get_xt <- function(x, y) {
  # Convert StatsBomb coords (120x80) to grid (8x12)
  grid_x <- ceiling(x / 10)
  grid_y <- ceiling(y / 10)

  grid_x <- max(1, min(12, grid_x))
  grid_y <- max(1, min(8, grid_y))

  return(xt_grid[grid_y, grid_x])
}

# Calculate xT prevented by defensive actions
def_xt <- events %>%
  filter(type.name %in% c("Tackle", "Interception", "Block"),
         duel.outcome.name %in% c("Won", "Success") |
         type.name %in% c("Interception", "Block")) %>%
  mutate(
    xt_at_action = mapply(get_xt, location.x, location.y),
    # Opponent was progressing, so xT prevented is current position value
    xt_prevented = xt_at_action
  )

# Player xT prevented
xt_prevented_by_player <- def_xt %>%
  group_by(player.name, team.name) %>%
  summarise(
    matches = n_distinct(match_id),
    defensive_actions = n(),
    total_xt_prevented = sum(xt_prevented),
    xt_prevented_per_90 = round(total_xt_prevented / matches, 3),
    .groups = "drop"
  ) %>%
  filter(matches >= 3) %>%
  arrange(desc(xt_prevented_per_90))

print("xT Prevented Leaders:")
print(head(xt_prevented_by_player, 15))

# Track defensive failures
defensive_failures <- events %>%
  filter(
    # Dribbled past (from dribble events where defender lost)
    (type.name == "Dribbled Past") |
    # Fouls leading to dangerous free kicks
    (type.name == "Foul Committed" & location.x >= 80) |
    # Errors
    (type.name == "Error")
  ) %>%
  group_by(player.name, team.name) %>%
  summarise(
    matches = n_distinct(match_id),
    dribbled_past = sum(type.name == "Dribbled Past"),
    fouls_danger_zone = sum(type.name == "Foul Committed"),
    errors = sum(type.name == "Error"),
    .groups = "drop"
  ) %>%
  mutate(
    failures_per_90 = round((dribbled_past + errors) / matches, 2)
  )

# Combine with positive actions for net contribution
net_defensive <- defensive_stats %>%
  select(player.name, team.name, matches, ball_winning_per_90) %>%
  left_join(
    defensive_failures %>%
      select(player.name, failures_per_90),
    by = "player.name"
  ) %>%
  mutate(
    failures_per_90 = ifelse(is.na(failures_per_90), 0, failures_per_90),
    net_defensive = ball_winning_per_90 - failures_per_90
  ) %>%
  filter(matches >= 3) %>%
  arrange(desc(net_defensive))

print("Net Defensive Contribution (Wins - Failures):")
print(head(net_defensive, 15))

# Aerial duel analysis
aerial_stats <- events %>%
  filter(type.name == "Aerial Lost" | type.name == "Clearance" |
         (type.name == "Duel" & duel.type.name == "Aerial Lost")) %>%
  # This is simplified - actual aerial tracking needs both won/lost
  group_by(player.name, team.name) %>%
  summarise(
    matches = n_distinct(match_id),
    aerials = n(),
    .groups = "drop"
  ) %>%
  mutate(
    aerials_per_90 = round(aerials / matches, 2)
  ) %>%
  filter(matches >= 3)

# For proper aerial win rate, need both aerial won and lost events
print("Aerial Activity Leaders:")
print(head(aerial_stats %>% arrange(desc(aerials_per_90)), 10))