20220929 MLS 2022 xG evolution.R

# -- Created by Owen Thompson (@OTStats)
# 2022-09-29

# -- Load libraries
library(tidyverse)
library(ggtext)
library(ggforce)
library(lubridate)
library(worldfootballR)
library(glue)
library(ggimage)
library(showtext)

font_add_google("Lato")
showtext_auto()

# -- Import Data
spi_raw <- read_csv("https://projects.fivethirtyeight.com/soccer-api/club/spi_matches.csv")

# Prep data
matches <- spi_raw %>% 
  transmute(date, 
            league, 
            league_id, 
            team = team1, 
            opponent = team2, 
            teamGoal = score1, 
            oppGoal = score2, 
            result = case_when(score1 > score2  ~ "W", 
                               score1 < score2  ~ "L", 
                               score1 == score2 ~ "D"), 
            ha = "Home", 
            team_spi = spi1, 
            opponent_spi = spi2, 
            team_xg = xg1, 
            opponent_xg = xg2) %>% 
  bind_rows(
    spi_raw %>% 
      transmute(date, 
                league, 
                league_id, 
                team = team2, 
                opponent = team1, 
                teamGoal = score2, 
                oppGoal = score1, 
                result = case_when(score1 < score2  ~ "W", 
                                   score1 > score2  ~ "L", 
                                   score1 == score2 ~ "D"), 
                ha = "Away", 
                team_spi = spi2, 
                opponent_spi = spi1, 
                team_xg = xg2, 
                opponent_xg = xg1)) %>% 
  mutate(game_goal_diff = teamGoal - oppGoal) %>% 
  mutate(result_points = case_when(result == "W" ~ 3, 
                                   result == "D" ~ 1, 
                                   TRUE ~ 0))



doParallel::registerDoParallel()
dates <- seq(ymd("20220226"), ymd("20220310"), by = "day")
results <- fotmob_get_matches_by_date(date = dates)
filtered_results <- results %>%
  dplyr::select(primary_id, ccode, league_name = name, matches) %>%
  dplyr::filter(league_name == "Major League Soccer", ccode == "USA")

# one way of getting data out of the results
unnested_results <- filtered_results %>% 
  tidyr::unnest_longer(col = matches) %>% 
  unnest(everything()) %>% 
  unnest(cols = c("home", "away"), names_sep = "_") %>% 
  unnest(cols = status) %>% 
  janitor::clean_names()


match_ids <- unnested_results %>% 
  distinct() %>% 
  pull(id)


details_df <- fotmob_get_match_details(match_ids)

shots <- details_df %>% 
  unnest(shots) %>% 
  mutate(team = if_else(team_id == home_team_id, home_team, away_team))

mls_team_color_df <- shots %>% 
  distinct(team, team_color, team_id)



mls_df <- matches %>% 
  filter(league_id == 1951) %>% 
  mutate(season = 
           case_when(between(date, ymd("20220226"), ymd("20220913")) ~ 2022, 
                     # between(date, ymd("20210416"), ymd("20211107")) ~ 2021, 
                     # between(date, ymd("20200229"), ymd("20200308")) ~ 2020, 
                     # between(date, ymd("20200812"), ymd("20201108")) ~ 2020, 
                     # between(date, ymd("20190302"), ymd("20191008")) ~ 2019, 
                     TRUE ~ 0)) %>% 
  filter(season == 2022) %>% 
  arrange(team, date) %>% 
  with_groups(team, mutate, matchday = row_number()) %>% 
  mutate(half = if_else(matchday <= 17, "first", "second")) %>% 
  group_by(team, half) %>% 
  summarize(xGF = mean(team_xg, na.rm = TRUE), 
            xGC = mean(opponent_xg, na.rm = TRUE), 
            .groups = "drop") %>% 
  # pivot_wider(id_cols = team, names_from = season, values_from = c(xGF, xGC)) %>% 
  mutate(team = case_when(
    team == "Los Angeles Galaxy" ~ "LA Galaxy", 
    team == "Atlanta United FC" ~ "Atlanta United", 
    team == "Chicago Fire" ~ "Chicago Fire FC", 
    team == "Minnesota United FC" ~ "Minnesota United", 
    team == "Houston Dynamo" ~ "Houston Dynamo FC", 
    team == "Montreal Impact" ~ "CF Montreal", 
    team == "Orlando City SC" ~ "Orlando City", 
    TRUE ~ team
  )) %>% 
  inner_join(mls_team_color_df, by = "team") %>% 
  mutate(logo = if_else(half == "second", glue("https://images.fotmob.com/image_resources/logo/teamlogo/{team_id}.png"), 
                        NULL))

pacman::p_load(ggpath)
# library(gghighlight)
# p <- 
mls_df %>% 
  ggplot() + 
  geom_link2(aes(x = xGF,
                y = xGC,
                # alpha = stat(index),
                size = stat(index),
                color = team_color,
                group = team), n = 100, lineend = "round") +
  scale_size(range = c(.01, 1.8)) + 
  guides(size = "none", alpha = "none") + 
  scale_color_identity() + 
  # geom_image(aes(x = xGF, y = xGC, image = logo), size = .05) + 
  geom_from_path(aes(x = xGF, y = xGC, path = logo), width = .045, alpha = 0.9) + 
  coord_fixed() + 
  # scale_y_continuous(breaks = seq(.75, 2, by = 0.25)) +
  scale_x_continuous(breaks = seq(.75, 2.25, by = 0.25)) +
  scale_y_reverse(breaks = seq(.75, 2, by = 0.25)) +
  labs(title = "Team xG evolution during the 2022 season", 
       subtitle = "Start of tail: First 17 matches of 2022; End of tail: Second half of 2022 matches to date\nCreated by: Owen Thompson | @OTStats", 
       x = "Average xG Created", 
       y = "Average xG Conceded", 
       caption = glue("Data as of {format(Sys.Date()-1)}\nSource: 538")) + 
  theme_minimal() + 
  theme(text = element_text(family = "Lato"), 
        plot.title.position = "plot", 
        plot.title = element_text(size = 20), 
        plot.background = element_rect(fill = "#f2f4f5", color = "#f2f4f5")) + 
  
  # --- Upper Right ---
  annotate(geom = "text", 
           x = 2, y = .98, 
           label = "There's Good...", family = "Lato", 
           hjust = 1, vjust = 0.3, alpha = 0.5, size = 3) + 
  # --- Upper Right... and there's Philly ---
  annotate(geom = "text", 
           x = 2.2, y = 1.025, 
           label = "...and then there's Philly", family = "Lato", 
           hjust = 0.5, vjust = 0.3, alpha = 0.5, size = 3) + 
  # --- Bad, bad, bad ---
  annotate(geom = "text", 
           x = 1.1, y = 2, 
           label = "All around bad", family = "Lato", 
           # hjust = 0.5, vjust = 0, 
           alpha = 0.5, size = 3) + 
  # --- Fun to watch ---
  annotate(geom = "text", 
           x = 1.85, y = 1.9, 
           label = "Fun to watch", family = "Lato", 
           # hjust = 0.5, vjust = 0, 
           alpha = 0.5, size = 3) + 
  # --- Boring ---
  annotate(geom = "text", 
           x = 1.1, y = 1.05, 
           label = "Defensively sound, can't create", family = "Lato", 
           # hjust = 0.5, vjust = 0, 
           alpha = 0.5, size = 3)
  

source("Developer/OT/gameplots/R/add_logo.R")

ggsave(filename = "20220929 MLS 2022 season xG evolution.png", 
       plot = last_plot(), 
       width = 9, height = 8, dpi = "retina")


# -- Add Leauge logo to plot
plot_with_logo <- add_logo(plot_path = "20220929 MLS 2022 season xG evolution.png",
                           logo_path = glue("https://images.fotmob.com/image_resources/logo/leaguelogo/130.png"),
                           logo_position = "top right",
                           logo_scale = 12)


magick::image_write(plot_with_logo, "20220929 MLS 2022 season xG evolution.png")