leloss.R

packages <- c('rvest','dplyr','pipeR', 'knitr')
lapply(packages, library, character.only = T)

css_page <- "#pgl_basic"

getYear <- function(year='2017') {
  url <- "https://www.basketball-reference.com/players/j/jamesle01/gamelog/%s"
  url <- sprintf(url, as.character(year))
  url %>>%
      read_html %>>%
      html_nodes(css_page) %>>%
      html_table(header = F) %>>%
      data.frame() %>>%
      tbl_df() -> lbj
  ## cleanup
  lbj %>>%
      filter(X1 == 'Rk') %>% unique -> names
  lbj %>% 
      filter(X1 != 'Rk') -> lbj

  colnames(lbj) <- names
  lbj
}

lbj_li <- lapply(2004:2017, getYear)
lbj_df <- do.call('rbind', lbj_li)

colnames(lbj_df)[6] <- 'Away'
colnames(lbj_df)[8] <- 'Outcome'

## Pull out point differential
lbj_df$differential <- as.numeric(paste0(gsub("[^-+]", "\\1", 
      as.matrix(lbj_df[,'Outcome'])), gsub("\\D", "\\1", as.matrix(lbj_df[,'Outcome']))))

## convert dates
lbj_df %>% mutate(Date=as.Date(Date)) -> lbj_df

second <- which(c(0, diff(lbj_df$Date)) == 1)
first <- second-1

firstLoses <- grepl('L', as.matrix(lbj_df[first,'Outcome']))
biglosses <- lbj_df[first,'differential'] <= -20

## Record of the second night of a back to back, after losing by at least 20 on the first night
lbj_df[second[biglosses],1:8]