In khailper/mtggplot: ggplot2 color palettes based on Magic: the Gathering

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(tidyverse)
library(scales)
library(cowplot)
library(gganimate)
library(mtggplot)

A classic problem is should you keep a hand that has important spells, but not the right lands to cast them. Limited is my typical non-causual format, so we'll start by calculating the probabilites for 40 card decks.

create_hits_distibution <- function(land_count, deck_size){
  tibble(turn = 2:7) %>% 
    # mental model by anology: on turn 3, prob that you drew a land on either 
    # of your two draw steps, plus the probability you drew two lands between
    # the two draw steps
    mutate(total_hits = map(turn, ~ seq(1, .x - 1))) %>%
    unnest() %>%
    mutate(prob = dhyper(x = total_hits,
                         m = land_count,
                         # non-land cards (or off-color lands)
                         # assume we're on the play
                         n = deck_size - 7 - (land_count),
                         k = turn - 1),
           possible_hits = land_count
    ) %>%
    group_by(turn, possible_hits) %>%
    summarise(prob_of_at_least_one_hit = sum(prob)) %>%
    identity()
}

limited_data <- map_dfr(5:12, create_hits_distibution, deck_size = 40)

limited_gif <-  ggplot(limited_data, 
                       aes(turn,  
                           y = prob_of_at_least_one_hit)) + 
  geom_area(fill = "white") +
  geom_hline(yintercept = 0.5, color = "red")+ 
  geom_text(x = 4.5, 
            y = 0.5, 
            label = "It's a coin toss.", 
            vjust = -0.5,
            color = "red",
            size = 6) +
  geom_text(x = 4.5, y = 0.8, label = ":(", color = "white", size = 10) + 
  geom_text(x = 4.5, y = 0.1, label = ":)", color = "black", size = 10) + 
  transition_states(possible_hits, transition_length = 1, state_length = 2) +
  theme_cowplot() +
  theme(panel.background = element_rect(fill = 'black')) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_y_continuous(labels = percent_format(), 
                     limits = c(0, 1),
                     expand = c(0, 0)) +
  labs(y = "Probability that you find the land you need...", 
       x = "... by turn X (on the play).",
       title = "Can you expect to draw the land your hand needs?",
       subtitle = "On-color lands in your Limited deck: {next_state}",
       caption = "Created by Karl Hailperin (@khailper)")

limited_gif

constructed_data <- map_dfr(12:19, create_hits_distibution, deck_size = 60)

constructed_gif <-  ggplot(constructed_data, 
                       aes(turn,  
                           y = prob_of_at_least_one_hit)) + 
  geom_area(fill = "white") +
  geom_hline(yintercept = 0.5, color = "red")+ 
  geom_text(x = 4.5, 
            y = 0.5, 
            label = "It's a coin toss.", 
            vjust = -0.5,
            color = "red",
            size = 6) +
  geom_text(x = 4.5, y = 0.8, label = ":(", color = "white", size = 10) + 
  geom_text(x = 4.5, y = 0.1, label = ":)", color = "black", size = 10) + 
  transition_states(possible_hits, transition_length = 1, state_length = 2) +
  theme_cowplot() +
  theme(panel.background = element_rect(fill = 'black')) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_y_continuous(labels = percent_format(), 
                     limits = c(0, 1),
                     expand = c(0, 0)) +
  labs(y = "Probability that you find the land you need...", 
       x = "... by turn X (on the play).",
       title = "Can you expect to draw the land your hand needs?",
       subtitle = "On-color lands in your Constructed deck: {next_state}",
       caption = "Created by Karl Hailperin (@khailper)")

constructed_gif