20250222 NHL goals saved above expected.R

# -- Load libraries
library(tidyverse)
library(glue)
library(ggpath)


# -- Read data
df <- read_csv("Downloads/20250222_shots_2024.csv") %>% 
  janitor::clean_names()



df %>% tail()


# # -- Look at goalies goals allowed over expected
# df %>%
#   filter(
#     # shot_was_on_goal == 1,
#     event %in% c("SHOT", "GOAL"),
#     shot_on_empty_net == 0
#   ) %>%
#   group_by(goalie_name_for_shot) %>%
#   summarise(goals_allowed = sum(goal),
#             xg = sum(x_goal),
#             diff = xg - goals_allowed) %>%
#   arrange(diff)



prepped_for_viz <- 
  df %>% 
  filter(
    # shot_was_on_goal == 1, 
    # event %in% c("SHOT", "GOAL"),
    shot_on_empty_net == 0
  ) %>% 
  transmute(
    id,
    team = if_else(is_home_team == 1, away_team_code, home_team_code), 
    goalie_name_for_shot, 
    xg = x_goal,
    goal
  ) %>% 
  group_by(team, goalie_name_for_shot) %>% 
  mutate(
    goalie_shot_id = row_number(), 
    cum_xg = cumsum(xg), 
    cum_ga = cumsum(goal), 
    goal_xg_diff = cum_ga - cum_xg
  )


# -- Pull in NHL logos
# Courtesy of Ivo Villanueva CSV
# Need to manually adjust some team abbrevation in order to join tables
data_git_nhl <- read.csv("https://raw.githubusercontent.com/IvoVillanueva/NHL/main/dataNHL.csv")
logos_df <- 
  data_git_nhl %>% 
  transmute(
    team = str_to_upper(url_abr), 
    logo = espn_logo, 
    primary
  ) %>% 
  distinct() %>% 
  add_row(team = "SEA") %>% 
  add_row(team = "UTA") %>% 
  mutate(
    team = 
      case_when(
        team == "TB" ~ "TBL", 
        team == "VGS" ~ "VGK", 
        team == "LA" ~ "LAK", 
        team == "NJ" ~ "NJD", 
        team == "SJ" ~ "SJS", 
        team == "SEA" ~ "SEA", 
        team == "UTA" ~ "UTA", 
        TRUE ~ team
      )
  ) %>% 
  mutate(logo = case_when(
    team == "VGK" ~ "https://a.espncdn.com/i/teamlogos/nhl/500/vgs.png", 
    team == "SEA" ~ "https://a.espncdn.com/i/teamlogos/nhl/500/sea.png", 
    team == "UTA" ~ "https://a.espncdn.com/i/teamlogos/nhl/500/utah.png", 
    TRUE ~ logo), 
    primary = case_when(
      team == "SEA" ~ "#001628", 
      team == "UTA" ~ "#71AFE5", 
      TRUE ~ primary
    )
    )



prepped_for_viz %>% 
  left_join(
    logos_df, by = "team"
  ) %>% 
  mutate(x = if_else(goalie_shot_id == 100, 100, NA), 
         y = if_else(goalie_shot_id == 100, 25, NA)) %>% 
  mutate(logo = if_else(!is.na(x), logo, NA)) %>% 
  # filter(any(goalie_shot_id > 750)) %>% 
  ggplot(
    aes(
      x = goalie_shot_id, 
      goal_xg_diff, 
      color = primary)
    ) + 
  geom_from_path(
     # data = logos_df %>% filter(team != "ARI"), 
             aes(x = x, y = y, path = logo), 
             width = 0.15, inherit.aes = FALSE) + 
  geom_line( 
      aes(group = goalie_name_for_shot), 
      size = .4
  ) + 
  geom_hline(yintercept = 0) + 
  scale_x_continuous(minor_breaks = NULL,
                     labels = scales::comma_format(),
                     # labels = c(0, "", 500, "", "1,000", "")
                     ) +
  scale_y_continuous(
    minor_breaks = NULL, 
    limits = c(-35, 35), 
    breaks = seq(-30, 30, by = 10), expand = c(0,0)) + 
  scale_color_identity() + 
  facet_wrap(
    ~team, 
    ncol = 4
    ) + 
  theme_minimal() + 
  theme(
    plot.title = element_text(hjust = 0, size = 16, face = "bold"), 
    plot.subtitle = element_text(hjust = 0, size = 12), 
    axis.title = element_text(size = 8), 
    axis.text = element_text(size = 6), 
    panel.grid.major = element_line(linewidth = .4), 
    strip.background = element_blank(), 
    strip.text = element_blank(), 
    plot.margin = margin(12, 12, 12, 12)
  ) + 
  labs(
    title = "NHL 2025: Goals Prevented by Goaltender", 
    subtitle = "Cumulative goals prevented from all shots faced in NHL | Each line is a different goaltender", 
    y = "", 
    x = "Cumulative Shot Faced", 
    caption = "Created by @OTStats\nData as of 2/24/2025\nData Source: @MoneyPuck.com"
  )
  
ggsave(
  filename = "20250225 NHL goaltender goals prevented.png", 
  height = 12, width = 8, dpi = "print", bg = "white"
)


install.packages("gghighlight")
library(gghighlight)
prepped_for_viz %>% 
  left_join(
    logos_df, by = "team"
  ) %>% 
  ggplot(
    aes(
      x = goalie_shot_id, 
      y = goal_xg_diff, 
      group = str_c(goalie_name_for_shot, "-", team)
    )
  ) + 
  geom_line(aes(color = primary)) + 
  gghighlight(any(goal_xg_diff < -22) || any(goal_xg_diff > 10), use_group_by = T) + 
  scale_color_identity()



#   event %in% c("SHOT", "GOAL")
# 
# goalieNameForShot
# team
# teamcode # team that shot
# shotWasOnGoal
# shotOnEmptyNet