Calculate yearly NBA homecourt advantage

nba_hca.R

library(sportsTools)
library(ggplot2)
library(grid)

# All years since NBA-ABA Merger
years <- 1977:2016

# Create empty data frame to hold results from each year
df <- data.frame(matrix(nrow = length(years), ncol = 4))
colnames(df) <- c('year', 'mean', 'sd', 'n')
df$year <- years

# Loop through years and save HCA mean, HCA SD, and number of reg season games
for (i in 1:nrow(df)) {
  sched <- GetSchedule(df[i, 'year'])
  sched <- sched[!is.na(sched$home.points), ]
  
  df[i, 'mean'] <- mean(sched$home.margin)
  df[i, 'sd'] <- sd(sched$home.margin)
  df[i, 'n'] <- length(sched$home.margin)
  
  cat(i, '\n')
}

# Include significant years and label them
sig.years <- c(1988, 1997, 2003, 2013, 2016)
sig.means <- round(sapply(sig.years, function(x) df[which(df$year == x), 'mean']), 2)

# Plot HCA over time without CI for Instagram
ggplot(df, aes(year)) +
  geom_line(aes(y = mean), color = 'darkblue', size = 0.7) +
  annotate("text", label = sig.means[1], x = sig.years[1], y = sig.means[1] + .2, size = 6) + 
  annotate("text", label = sig.means[2], x = sig.years[2], y = sig.means[2] - .2, size = 6) + 
  annotate("text", label = sig.means[3], x = sig.years[3], y = sig.means[3] + .2, size = 6) + 
  annotate("text", label = sig.means[4], x = sig.years[4], y = sig.means[4] + .2, size = 6) + 
  annotate("text", label = sig.means[5], x = sig.years[5], y = sig.means[5] - .2, size = 6) + 
  scale_x_continuous(expand = c(0, 0), breaks = seq(1976, 2016, 4), limits = c(1977, 2016.7)) +
  scale_y_continuous(expand = c(0, 0), breaks = seq(0, 7, 1), limits = c(0, 6.25)) +
  labs(title = "NBA Regular Season: Average Margin of Victory for Home Team") + 
  theme(axis.title = element_blank(),
        axis.line = element_line(size = 0.7, color = 'grey70'),
        axis.ticks = element_blank(),
        panel.border = element_blank(), panel.background = element_blank(),
        panel.grid.minor = element_blank(),
        panel.grid.major = element_line(size = 0.5, color = 'grey90'),
        plot.title = element_text(size = rel(2), face = 'bold', vjust = 1),
        axis.text = element_text(colour = 'black', size = rel(2)),
        plot.margin = unit(c(1, 1, 1, 1), 'line'))
ggsave(filename = './nba_hca.png', width = 14, height = 10, units = 'in')

# Calculate 95% confidence interval using T-Distribution
df$se <- df$sd / sqrt(df$n)
df$t.star <- qt(.975, df = df$n - 1)
df$margin <- df$t.star * df$se
df$lower <- df$mean - df$margin
df$upper <- df$mean + df$margin

# Plot HCA over time with CI for website
ggplot(df, aes(year)) +
  geom_line(aes(y = mean, color = 'Average Margin of Victory for Home Team')) +
  geom_line(aes(y = lower, color = '95% Confidence Interval for Homecourt Advantage'), linetype = 'dashed') +
  geom_line(aes(y = upper, color = '95% Confidence Interval for Homecourt Advantage'), linetype = 'dashed') +
  scale_color_manual(values = c('grey40', 'darkblue')) +
  scale_x_continuous(expand = c(0, 0), breaks = seq(1976, 2016, 4), limits = c(1977, 2016.7)) +
  scale_y_continuous(expand = c(0, 0), breaks = seq(-1, 7, 1), limits = c(-1, 7.25)) +
  labs(title = "NBA Regular Season Homecourt Advantage") + 
  theme(axis.title = element_blank(),
        axis.line = element_line(size = 0.7, color = 'grey70'),
        panel.border = element_blank(), panel.background = element_blank(),
        panel.grid.minor = element_blank(),
        panel.grid.major = element_line(size = 0.5, color = 'grey90'),
        legend.position = c(0.35, 0.16), legend.key = element_blank(),
        legend.title = element_blank(), legend.background = element_blank(),
        legend.text = element_text(colour = 'black', size = rel(1.3)),
        legend.key.size = unit(1.5, 'line'),
        plot.title = element_text(size = rel(2), face = 'bold', vjust = 1),
        axis.text = element_text(colour = 'black', size = rel(1.3)))
ggsave(filename = './nba_hca_ci.png', width = 14, height = 6, units = 'in')