construct measures of competitiveness of european leagues using match level data from https://github.com/jalapic/engsoccerdata

eur_leagues_competitiveness.R

# %% ####################################################
rm(list = ls())
library(LalRUtils)
LalRUtils::libreq(ggplot2, data.table, fixest, magrittr,
                  patchwork, IRdisplay, did, panelView, plotly)
theme_set(lal_plot_theme())
options(repr.plot.width=12, repr.plot.height=9)
options(ggplot2.discrete.fill = RColorBrewer::brewer.pal(7, "Set2"))
options(ggplot2.discrete.colour = RColorBrewer::brewer.pal(7, "Set2"))
options(ggplot2.continuous.fill = "viridis"); options(ggplot2.continuous.colour = "viridis")
set.seed(42)
chr = function(...) as.character(...) %>% display_html()
# %% ####################################################
data = "../data/engsoccerdata-master/data"
list.files(data) %>% print
# %%
bel = import(file.path(data, "belgium.rda"))  %>% setDT
eng = import(file.path(data, "england.rda"))  %>% setDT
fra = import(file.path(data, "france.rda"))  %>% setDT
ger = import(file.path(data, "germany.rda"))  %>% setDT
hol = import(file.path(data, "holland.rda"))  %>% setDT
ita = import(file.path(data, "italy.rda"))  %>% setDT
esp = import(file.path(data, "spain.rda"))  %>% setDT
por = import(file.path(data, "portugal.rda"))  %>% setDT

# %%
long_matches = function(df){
  df[, id := .I] # create match id
  df[, RESULT := fcase(hgoal > vgoal, "HW", hgoal < vgoal, "VW", default = "draw")]
  ids = setdiff(colnames(df), c("home", "visitor"))
  matches_long = melt(df, id.vars = ids, measure = c("home", "visitor"),
    variable.name = 'type', value.name = 'team')
  matches_long[,
      goals := ifelse(type == "home", hgoal, vgoal)][,
      points := fcase(
        type == "home"    & RESULT == "HW",  3,
        type == "visitor" & RESULT == "VW",  3,
        RESULT == "draw", 1,
        default = 0
      )]
  setorder(matches_long, Season, id)
  return(matches_long)
}

competition_index = function(df){
  df[, season_total := sum(points), by = .(Season, team)]
  team_tallies = df[, .(total_points = season_total[1]), by = .(Season, team), ]
  team_tallies[, rk := floor(frank(-total_points)), by = Season]
  setorder(team_tallies, Season, rk)
  team_tallies[, margin := total_points - lead(total_points), by = Season]
  team_tallies[,
    tot_points_all := sum(total_points), by = Season][,
    share_points   := (total_points / tot_points_all)][,
    hhi            := sum((share_points * 100)^2), by = Season][,
    enop           := 1/sum(share_points^2), by = Season]
  summ_seasons = team_tallies[, .SD[1], Season, .SDcols = c("margin", "hhi", "enop")]
  return(summ_seasons)
}

# %%
eng_comp = eng[division == 1]  |> long_matches() |> competition_index()
fra_comp = fra[division == 1]  |> long_matches() |> competition_index()
ger_comp = ger[division == 1]  |> long_matches() |> competition_index()
hol_comp = hol[tier == 1]  |> long_matches() |> competition_index()
ita_comp = ita[tier == 1]  |> long_matches() |> competition_index()
esp_comp = esp[tier == 1 & round == "league"]  |> long_matches() |> competition_index()
por_comp = por[division == "P1"]  |> long_matches() |> competition_index()
# %%
all_leagues = rbindlist(mget(ls(pattern = "*_comp")), idcol = TRUE)
all_leagues[, country := fcase(
  .id == "bel_comp", "bel",
  .id == "eng_comp", "eng",
  .id == "fra_comp", "fra",
  .id == "gre_comp", "gre",
  .id == "ger_comp", "ger",
  .id == "hol_comp", "hol",
  .id == "ita_comp", "ita",
  .id == "esp_comp", "esp",
  .id == "por_comp", "por",
  .id == "tur_comp", "tur", default = ""
)][, .id := NULL]

# %% figures
f1 = all_leagues[Season>=1950]  |>
  ggplot(aes(Season, margin, group = country, colour = country)) +
  geom_point() + geom_smooth(se = FALSE) +
  scale_color_brewer(palette = "Set3") +
  lal_plot_theme_d() + ggtitle("Title Winning Margin")
f2 = all_leagues[Season>=1950]  |>
  ggplot(aes(Season, hhi, group = country, colour = country)) +
  geom_point() + geom_smooth(se = FALSE) +
  scale_color_brewer(palette = "Set3") +
  lal_plot_theme_d() + ggtitle("HHI of Points Tallies")

# %%

f = (f1 | f2)
ggsave("../figs/competitiveness.png", f, width = 15, height = 10)
# %%