kc_victory.R

library(rvest)
library(tidyverse)
library(ggrepel)
library(glue)
library(lubridate)
library(gt)
library(ggforce)

url <- "https://www.pro-football-reference.com/teams/comebacks.htm"

list_of_teams <- c("atl", "buf", "car", "chi", "cin", "cle", "clt", "crd", "dal", "den", "det",
                   "gnb", "htx", "jax", "kan", "mia", "min", "nor", "nwe", "nyg", "nyj",
                   "oti", "phi", "pit", "rai", "ram", "rav", "sdg", 'sea', "sfo", "tam", "was")


scrape_comeback <- function(team){
  
  message(paste("Scraping", team))
  
  Sys.sleep(5)
  
  url <- glue::glue("https://www.pro-football-reference.com/teams/{team}/comebacks.htm")
  
  url %>% 
    read_html() %>% 
    html_table() %>% 
    .[[1]] %>% 
    janitor::clean_names() %>% 
    as_tibble()
}

chiefs_score <- tibble(
  tm = "Kansas City Chiefs",
  x = "",
  opp = "Houston Texans",
  date = "Jan 12, 2020",
  pf = 51,
  pa = 31,
  deficit = 24,
  margin = 20
)

all_comebacks <- list_of_teams %>% 
  map_dfr(scrape_comeback)




plot_df <- all_comebacks %>% 
  mutate_at(vars(pf, pa, deficit), as.numeric) %>% 
  rowwise() %>% 
  mutate(margin = pf - pa) %>% 
  bind_rows(chiefs_score) %>% 
  mutate(swing = margin + deficit) %>% 
  mutate(date = str_remove(date, "\\*"),
         date_iso = readr::parse_date(date, format = "%b %d, %Y"),
         year = lubridate::year(date_iso)) %>% 
  arrange(desc(swing)) %>% 
  filter(year >= 1968)

margin_plot <- plot_df %>%
  ggplot(aes(x = deficit, y= margin)) +
  annotate('segment',
           xend = 24, x = 24, y = 0, yend = 20, color = "red", size = 3, alpha = 0.2) +
  geom_point(alpha =  0.5, aes(color = if_else(deficit >= 24 & margin >= 20, "red", "black")),
             size = 3) +
  # geom_text(x = 24, y = 20, label = "(24,20)",vjust = -1) +
  geom_mark_circle(data = filter(plot_df, year == 2020, deficit == 24), 
                   aes(label = paste(tm, 2020, sep = " - ")), 
                   color = "red", label.colour = "red", con.colour = "red",
                   label.fill = NA) +
  geom_hline(yintercept = 0) +
  scale_color_identity() +
  theme_minimal() + 
  labs(caption = "\nPlot: @thomas_mock | Data: ProFootballReference",
       x = "\nLargest Deficit",
       y = "Margin of Victory\n",
       title = "KC's comeback from 24 points down was historic",
       subtitle = "Teams with deficit of 24 points or more & margin of victory of 20 or more in red") +
  theme(axis.title = element_text(face = "bold"),
        panel.grid.minor = element_blank(),
        plot.title = element_text(size = 20, face = "bold"),
        plot.subtitle = element_text(size = 16),
        axis.text = element_text(size = 16),
        axis.title.x = element_text(size = 20),
        axis.title.y = element_text(size = 20)) +
  scale_x_continuous(breaks = seq(5, 30, 5)) +
  scale_y_continuous(breaks = seq(0, 30, 5))

margin_plot

ggsave("margin_plot.png", margin_plot, height = 10, width = 10, units = "in", dpi = "retina")

plot_df

gt_table <- plot_df %>% 
  filter(deficit >= 24) %>% 
  select(tm, opp, date_iso, pf:swing) %>% 
  gt::gt() %>% 
  cols_merge(vars(pf), 
             vars(pa),
             pattern = "{1}-{2}") %>% 
  cols_label(tm = "Team", opp = "Opp", date_iso = "Date", 
             margin = "Margin", swing = "Swing", pf = "Score", deficit = "Deficit") %>% 
  data_color(
    columns = vars(margin),
    colors = scales::col_numeric(
      palette = c("#F8F8F8","#30a2da"),
      domain = NULL
    )
  ) %>% 
  data_color(
    columns = vars(swing),
    colors = scales::col_numeric(
      palette = c("#F8F8F8","#30a2da"),
      domain = NULL
    )
  ) %>% 
  tab_style(
    style = cell_text(
      weight = "bold"
    ),
    locations = cells_data(
      rows = tm == "Kansas City Chiefs"
    )
  ) %>% 
  tab_header(title = "Kansas City's comeback was historic",
  subtitle = "The largest swing and margin of victory ever for a deficit of 24 points or more") %>% 
  tab_source_note("Table: @thomas_mock | Data: Pro Football Reference") %>% 
  tab_footnote(footnote = "Margin = margin of victory",
               locations = cells_column_labels(
                 columns = vars(margin)
               )
  ) %>% 
  tab_footnote(footnote = "Deficit = largest deficit at any point in the game",
               locations = cells_column_labels(
                 columns = vars(deficit)
               )
  ) %>% 
  tab_footnote(footnote = "Swing = margin of victory + deficit",
               locations = cells_column_labels(
                 columns = vars(swing)
                 )
               )

gt_table

gtsave(gt_table, "kc_table.png")

kc_bar <- plot_df %>% 
  filter(deficit >= 24) %>% 
  mutate(tm_year = paste(tm, as.character(year), sep = "-"),
    tm_year = fct_reorder(tm_year, swing)) %>% 
  select(tm_year, pf:swing) %>% 
  ggplot(aes(x = fct_reorder(tm_year, swing), y = swing)) +
  geom_col(aes(fill = if_else(tm_year == "Kansas City Chiefs-2020", "red", "grey"))) +
  geom_text(aes(label = margin), hjust = 1.5, color = "white", fontface = "bold", size = 6) +
  geom_hline(yintercept = 0, size = 1, color = "black") +
  scale_fill_identity() +
  coord_flip() +
  theme_minimal() +
  theme(panel.grid.minor = element_blank(),
        panel.grid.major.y = element_blank(),
        axis.text = element_text(face = "bold", size = 14, color = "black"),
        axis.title = element_text(size = 16),
        plot.title = element_text(face = "bold", size = 20)) +
  labs(x = "",
       y = "\nPoint Swing",
       title = "Kansas City's comeback from 24 points down was historic",
       subtitle = "Point Swing = Deficit + Margin of Victory\nInset number = Margin of Victory",
       caption = "\nPlot: @thomas_mock | Data: Pro Football Reference")

kc_bar
ggsave("kc_bar.png", kc_bar, height = 12, width = 14, units = "in", dpi = "retina")