AI-native trading and research platform for Indian markets.

Honba is a Rust-powered engine with a Python control plane, built for Indian equities, indices, index universes, mutual funds, ETFs, and options. It combines deterministic event-driven backtesting with a three-tier optimization pipeline and an AI feedback loop.

• honba - Core engine, Python control plane, Indian market module. https://github.com/honba-labs/honba

• honba-adapters - Broker adapters for Dhan, Zerodha, Angel One, Fyers, Upstox, Kotak, IIFL, and Motilal Oswal. https://github.com/honba-labs/honba-adapters

• honba-strategies - Community strategy catalog, organized by trading idea. https://github.com/honba-labs/honba-strategies

• honba-examples - Numbered learning-path examples. https://github.com/honba-labs/honba-examples

• honba-frontend - Dashboard and visual flow builder. https://github.com/honba-labs/honba-frontend

• honba-docs - Documentation book and site. https://github.com/honba-labs/honba-docs

Deterministic engine. The same event flow runs in backtest and live. Strategies never need to be reimplemented between research and production.

Three-tier backtesting. A fast vectorized pre-filter, a realistic event-driven simulator, and a concurrent parameter sweep for optimization.

Indian market first-class. NSE and BSE calendar, STT, GST and stamp-duty cost model, dynamic index constituents, weekly options expiry cycles.

AI feedback loop. A research journal that an LLM reads and refines, Monte Carlo and walk-forward validation automated on every strategy, and an MCP gateway for external agents.

git clone https://github.com/honba-labs/honba.git cd honba cargo build --workspace cd python && pip install -e ".[dev]"

text

Then follow the installation guide in honba-docs and run your first backtest using the examples in honba-examples.

Honba composes proven ideas from the open-source trading ecosystem:

• Nautilus Trader - deterministic event-driven engine
• Barter-rs - concurrent backtesting and latency simulation
• Jesse - statistical validation patterns
• QuantDinger - AI research loop patterns
• WonderTrader - context-aware caching and position merging
• StockSharp - multi-repo organizational structure

See the NOTICE file in the core repository for full attribution.

Under active development. The crate skeleton is complete, the dependency hierarchy is enforced in CI, and honba-messages, honba-entities, and honba-india are the current focus.

Apache-2.0. See the LICENSE file in each repository.