Blackjack simulator: A gamified R package

R Package Development · C++ Integration · Simulation

The idea

Blackjack is a game of probability. Every decision — hit or stand — has a mathematically optimal answer depending on your hand and the dealer’s visible card. So when our team was tasked with building a gamified R package, Blackjack felt like a natural fit. It’s simple enough to explain in one sentence, but deep enough to keep a statistician busy for hours.

The result is blackjackr — an interactive R package that lets you play Blackjack, simulate thousands of rounds, and test strategies, all from your R console.

My contributions

This was a team project with Andy Tran and Rong Xu. My specific contributions across both parts of the project were:

• C++ programming for the deal_hand() function for performance
• Building core functions and unit tests
• Writing the introduction vignette and README
• Debugging errors during R CMD checks
• Refining roxygen documentation comments

Installing the package

install.packages("devtools")
devtools::install_github("MonashARP/game-package-bilbies")

Playing a round

Once installed, you can play a full interactive round of Blackjack straight from the console. The game prompts you to hit, stand, double down, or split depending on your hand and balance.

library(blackjack)

# Play a single round with a starting bet of $100
play_blackjack(bet = 100, n_decks = 4)

You can also run a multiplayer session with up to several players each taking turns against the dealer:

play_blackjack_multi(num_players = 3, bet = 50)

Working with cards

The package includes a set of utility functions for inspecting card properties. These are useful for building game logic or debugging:

library(blackjack)

deck <- create_shuffled_deck()
card <- deck[1]

card_name(card)

[1] "A♠"

card_rank(card)

[1] "Ace"

card_value(card)

[1] 11

card_is_face(card)

[1] FALSE

Simulating strategy

The most analytically interesting part of the package is simulation_blackjack() — it lets you simulate thousands of rounds automatically and observe outcomes. This is where the probability really comes to life.

results <- simulation_blackjack(
  n_simulations = 1000,
  bet = 100,
  n_decks = 4
)

summary(results)

Number of cases in table: 1000 
Number of factors: 1 

What I learned

Building a package rather than just a script forced a completely different way of thinking. Every function needs documentation, every edge case needs a unit test, and everything needs to pass R CMD check before it can be shared. The C++ component for deal_hand() was particularly challenging — it was my first time writing compiled code inside an R package — but the performance improvement made it worth it.

The simulation function is what I’m most proud of. Watching the distribution of outcomes emerge over thousands of rounds is a genuinely satisfying demonstration of probability in action.

View on GitHub Package documentation