Compute CFB Game Results in Season Simulations

The default compute_results function used by cfb_simulations(). It is a faithful adaptation of nflseedR's nflseedR_compute_results() (a variant of 538's ELO model, originally coded by Lee Sharpe and rewritten by Sebastian Carl): a simple dynamic ELO carried through the teams table, updated each week from (real or simulated) results.

Usage

cfbseedR_compute_results(teams, games, week_num, ...)

Arguments

teams

A data frame of teams by simulation with at least the columns sim and team. May carry an elo column between weeks.

games

A data frame of games with at least sim, game_type, week, home_team, away_team, result, and optionally neutral (0/1; a home-field ELO bonus of 20 applies when not neutral).

week_num

The (numeric) week to simulate. Only games with week == week_num & is.na(result) receive a simulated result.

...

Optionally pass elo, a named vector of initial ELO ratings (names = team names). If absent, teams start at random ratings drawn from rnorm(1500, 150).

Value

A list with two elements, as required by the compute_results contract (see simulations_verify_fct()):

Element	Type	Description
teams	data.frame	The input teams table with the elo column added/updated from this week's results.
games	data.frame	The input games table with result filled for this week's previously missing results (home margin, integer).

Details

Mechanics (constants identical to nflseedR): ELO difference is home - away plus 20 for true home games, multiplied by 1.2 in the postseason (game_type != "REG"). Win probability is 1 / (10^(-elo_diff / 400) + 1), the margin estimate is elo_diff / 25, and simulated margins are drawn from rnorm(mean = estimate, sd = 13) and rounded away from zero. Ties are possible in the regular season only; a simulated postseason tie is re-decided by the win probability with a 3-point margin. nflseedR's rest-day adjustment is dropped (no rest data in the CFB schema).

See also

cfb_simulations(), simulations_verify_fct()

Examples

games <- read.csv(system.file("extdata", "toy_games.csv", package = "cfbseedR"))
teams <- read.csv(system.file("extdata", "toy_teams.csv", package = "cfbseedR"))
games$result[games$week >= 3] <- NA
teams$sim <- 1
games$sim <- 1
out <- cfbseedR_compute_results(teams, games, week_num = 3)
out$games[out$games$week == 3, ]
#>    sim week game_type home_team away_team result neutral
#> 9    1    3       REG        A1        A4     24       0
#> 10   1    3       REG        A2        A3    -15       0
#> 11   1    3       REG        B1        B3     17       0
#> 12   1    3       REG        B2        B4     -3       0