In Kazink36/cfb4th: Functions to Calculate Optimal Fourth Down Decisions for NCAA Football

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
options(dplyr.summarise.inform = FALSE)
options(warn = -1) # silences a warning caused by a dplyr bug in the gt package

First load the packages and some data. load_4th_pbp() loads cfbfastR data and computes 4th down probabilities (depending on your computer, this may take up to a minute or two per season).

library(cfb4th)
library(dplyr)
library(knitr)
library(tibble)
library(gt)

Easy mode: using cfbfastR data

Here's what the data obtained using load_4th_pbp() looks like:

suppressWarnings(
  data <- cfb4th::load_4th_pbp(2020) %>% 
    dplyr::filter(.data$week %in% 1:14)
)
data %>%
  dplyr::filter(!is.na(go_boost)) %>%
  utils::head(10) %>%
  dplyr::select(
    pos_team, distance, yards_to_goal, go_boost, first_down_prob, 
    wp_fail, wp_succeed, go_wp, fg_make_prob, miss_fg_wp, make_fg_wp, 
    fg_wp, punt_wp
  ) %>%
  knitr::kable(digits = 2)

Or we can add some filters to look up a certain game:

data %>%
  dplyr::filter(week == 12, pos_team == "Utah", down == 4) %>%
  dplyr::select(
    pos_team, distance, yards_to_goal, go_boost, first_down_prob, 
    wp_fail, wp_succeed, go_wp, fg_make_prob, miss_fg_wp, make_fg_wp, 
    fg_wp, punt_wp
  ) %>%
  knitr::kable(digits = 2)

Calculations from user input

The below shows the bare minimum amount of information that has to be fed to cfb4th in order to compute 4th down decision recommendations. The main function on user-input data is add_4th_probs().

Teams are included to help the model easily track the simulations.

one_play <-
  tibble::tibble(
      # Game Info
      home = "Utah",
      away = "BYU",
      pos_team = "Utah",
      def_pos_team = "BYU",
      spread = -7,
      over_under = 55,

      # Situation Info
      half = 2,
      period = 3, # Quarter
      TimeSecsRem = 900, # Half Seconds Remaining
      adj_TimeSecsRem = 900, # Game Seconds Remaining
      down = 4,
      distance = 4,
      yards_to_goal = 40,
      pos_score_diff_start = 7,

      pos_team_receives_2H_kickoff = 1,
      pos_team_timeouts_rem_before = 3,
      def_pos_team_timeouts_rem_before = 3

    )
one_play %>%
  cfb4th::add_4th_probs() %>%
  dplyr::select(
    pos_team, distance, yards_to_goal, go_boost, first_down_prob, 
    wp_fail, wp_succeed, go_wp, fg_make_prob, miss_fg_wp, make_fg_wp, 
    fg_wp, punt_wp
  ) %>%
  knitr::kable(digits = 2)

Make a summary table

Let's put the play above into a table using the provided function make_table_data(), which makes it easier to interpret the recommendations for a play. This function only works with one play at a time since it makes a table using the results from the play.

one_play %>%
  cfb4th::add_4th_probs() %>%
  cfb4th::make_table_data() %>%
  knitr::kable(digits = 1)

Looking at the table, the offense would be expected to have 86.5% win probability if they had gone for it and 85% if they punted.

Getting 4th down plays from a live game

cfbfastR isn't available for live games and typing all the plays in by hand is annoying. So how does the 4th down bot work? With thanks to the ESPN API, which can be accessed using get_4th_plays().

game <- cfbfastR::cfbd_game_info(year = 2019, team = "Utah", week = 4)
plays <- cfb4th::get_4th_plays(game) %>% 
  tail(1)
plays %>% 
  dplyr::select("desc", "TimeSecsRem")
plays %>% 
  cfb4th::add_4th_probs() %>%
  cfb4th::make_table_data() %>%
  knitr::kable(digits = 1)

Kazink36/cfb4th documentation built on Jan. 25, 2025, 12:19 a.m.