search

MatthewPeterKelly / poker-stats

Small library for computing poker statistics using Monte-Carlo simulation. The primary focus for the project is learning Rust.
MIT License
2 stars 1 forks source link

readme

Poker Stats with Rust

This small project is primarily a way for me to learn rust. The idea is to use Monte-Carlo simulation to compute the probability of drawing each of the different scoring hands. These probabilities can easily be computed analytically, so we have a good way to verify the results of the program.

Usage Examples:

Show top-level help file for the command-line interface:

cargo run -- --help

Quick default demonstration: draws a five card hand and scores it:

cargo run

Run statistics for drawing 5-card hands, using 100000 samples and 8 threads:

cargo run -- statistics 5 100000000
HandScore:
  high_card         100000000 (100.000%)
  pair              49287061  ( 49.287%)
  two_pair          4755262   (  4.755%)
  three_of_a_kind   2278276   (  2.278%)
  straight          354474    (  0.354%)
  flush             198043    (  0.198%)
  full_house        142993    (  0.143%)
  four_of_a_kind    23990     (  0.024%)
  straight_flush    1365      (  0.001%)

Draw a single 7-card hand:

cargo run -- draw-hand 7
Hand: 9♣, 8♥, 4♥, 6♦, 2♣, 2♦, 8♣
HandStats:
  Count: 7
  Suits: [♣]: 3, [♦]: 2, [♥]: 2
  Ranks: [2]: 2, [4]: 1, [6]: 1, [8]: 2, [9]: 1
HandScore:
  high_card         1 (100.000%)
  pair              1 (100.000%)
  two_pair          1 (100.000%)
  three_of_a_kind   0 (  0.000%)
  straight          0 (  0.000%)
  flush             0 (  0.000%)
  full_house        0 (  0.000%)
  four_of_a_kind    0 (  0.000%)
  straight_flush    0 (  0.000%)

Design:

My overall goal is learning, so I've tried to write idiomatic Rust code as much as possible. I'm coming from a C++ background, so perhaps I missed a few things.

I've tried to make the "inner loop" code as efficient as possible, with a side goal of optimizing the time to draw and score each hand. So far I've mostly approached this from a design standpoint, rather than using real profiling tools.

Organization:

This repository is set up with a small library that provides all of the classes and utility functions needed to run the analysis, along with a simple main function that exercises the library. Each file in the library includes unit tests.

Contributing

If you're interested in contributing to this project just send me and email or open a PR! This could include new features, bug-fixes, implementations for suggestions in the issues, or just style improvements to make the code more idiomatic.