| --- |
| license: apache-2.0 |
| language: |
| - en |
| tags: |
| - finance |
| - reinforcement-learning |
| - gym |
| - trading |
| - backtesting |
| - rust |
| pretty_name: Chapaty Environments |
| --- |
| |
| # Chapaty Environments |
|
|
| [](https://discord.gg/MmMAB6NCuK) |
| [](https://github.com/LenWilliamson/chapaty) |
| [](https://crates.io/crates/chapaty) |
|
|
| This dataset repository hosts the pre-compiled financial environments for [**Chapaty**](https://github.com/LenWilliamson/chapaty), a Rust library for building quantitative trading agents in financial markets. |
|
|
| Inspired by OpenAI Gymnasium, these datasets provide standardized, easily reproducible simulation states for training and backtesting agents. |
|
|
| ## What are these files? |
|
|
| The datasets here are serialized using `.postcard` (a highly efficient binary format for Rust). They contain environment configurations based on the version of the crate you are using. |
|
|
| ## Versioning |
|
|
| This repository uses **branches** to align with Chapaty crate releases (e.g., `v1.1.0`). |
| When you use Chapaty in Rust, the library automatically binds to the correct branch matching your crate version, ensuring your agents are always evaluated on consistent data. |
|
|
| ## Usage in Rust |
|
|
| **Fast Track:** Use the [**Chapaty Starter Template**](https://github.com/LenWilliamson/chapaty-template) to instantly bootstrap a new project. It includes pre-configured AI prompts for backtesting with a LLM of your choice, built-in dashboard setups with [Quantstats](https://github.com/ranaroussi/quantstats), and best-practice strategy examples. |
|
|
| You do not need to download these files manually. Use the Chapaty crate to load them directly into your async Rust application: |
|
|
| ```rust |
| use anyhow::{Context, Result}; |
| use chapaty::prelude::*; |
| |
| #[tokio::main] |
| async fn main() -> Result<()> { |
| // Select a preset (e.g., Daily BTC/USDT with SMA 20 & 50) |
| let preset = EnvPreset::BinanceBtcUsdt1dSma20Sma50; |
| |
| // Configure I/O to fetch seamlessly from Hugging Face |
| let file_stem = preset.to_string(); |
| let loc = StorageLocation::HuggingFace { version: None }; |
| let cfg = IoConfig::new(loc).with_file_stem(&file_stem); |
| |
| // Load the environment (downloads and caches locally on first run) |
| let mut env = chapaty::load(preset, &cfg) |
| .await |
| .context("Failed to load trading environment")?; |
| |
| println!("Successfully loaded environment for {}!", preset); |
| |
| Ok(()) |
| } |
| ``` |
|
|
| ## Available Presets |
|
|
| Presets encode exact data source IDs to reproduce identical environments. Current presets include: |
|
|
| * `binance_btc_usdt1d`: BTC/USDT Daily Spot |
| * `ninja_trader_cme6eh6_1m_5m_us_emp_high`: EUR/USD 1m/5m Futures with US Employment News |
| * `binance_btc_usdt1h_1m_volume_profile1d_100_usdt`: BTC/USDT Multi-resolution Spot with Volume Profile |
| * *(And many more—see the [Chapaty Crate Documentation](https://docs.rs/chapaty) for the full `EnvPreset` list).* |
|
|
| ## Disclaimer |
|
|
| **Trading and investing involve substantial risk. You may lose some or all of your capital.** |
| These datasets and the Chapaty library are provided for **research and educational purposes only** and do not constitute financial advice. |
|
|
