A python package to run all of the necessary elements of a Craps table. The package follows some natural logic:
- a
CrapsTable
hasPlayer
(s) andDice
on it - the
Player
(s) haveBet
(s) on theCrapsTable
as prescribed by their strategies.
With these building blocks, crapssim supports
- running 1 session with 1 player/strategy to test a realistic day at the craps table,
- running many sessions with 1 player/strategy to understand how a strategy performs in the long term, or
- running many sessions with many players/strategies to simulate how they compare to each other
These powerful options can lead to some unique analysis of the game of craps, such as the following figure comparing 4 strategies with a budget of $200:
I will post results from this simulator on my site: https://pages.stat.wisc.edu/~kent/blog.
Current blog posts include:
- One Surprising Feature of Dark Side Betting
- 5 Systems to Try at the Craps Table
- Craps: Best Strategies on a Budget
- All Bets Are Off: Re-learning the Pass Line Bet in Craps
You can install crapssim with
pip install crapssim
This requires Python >=3.6 and pip to be installed on your computer. You will also need a recent version of numpy, which can be installed by pip install numpy
.
There's a few good resources for getting started:
- Try the interactive google collab notebook to test a single strategy and see how the interface works
- Check out Corey Brown's scripts to define strategies and compare them: https://github.com/coreyabrown/CoreyCrapsSim
- Looks at the minimal working examples below
To see how a single session might play out for you using a pass line bet with double odds, over 20 rolls, one might run:
import crapssim as craps
table = craps.Table()
your_strat = craps.strategy.passline_odds2
you = craps.Player(bankroll=200, bet_strategy=your_strat)
table.add_player(you)
table.run(max_rolls=20)
To evaluate a couple of strategies across many table sessions, you can run:
import crapssim as craps
n_sim = 20
bankroll = 300
strategies = {
"place68": craps.strategy.place68,
"ironcross": craps.strategy.ironcross
}
for i in range(n_sim):
table = craps.Table()
for s in strategies:
table.add_player(craps.Player(bankroll, strategies[s], s))
table.run(max_rolls=float("inf"), max_shooter=10, verbose=False)
for s in strategies:
print(f"{i}, {s}, {table._get_player(s).bankroll}, {bankroll}, {table.dice.n_rolls}")
For more advanced strategies, you need to write a custom function that can perform the strategy. Some building blocks and examples can be found in strategy.py
If you discover something interesting using this simulator, please let me know so that I can highlight those results here. You can find me at [email protected].
Those looking to contribute to this project are welcome to do so. Currently, the top priority is to improve
- Supported bets (see bet.py)
- Supported strategies (see strategy.py)
- Documentation