The Autonomous Quant Research Platform

alphabench is an AI-native platform for quantitative trading research. It enables traders and researchers to discover, backtest, and iteratively refine algorithmic trading strategies through natural language conversation.

Describe a strategy idea in plain English. The platform analyzes your intent, fetches historical market data, compiles and executes the backtest, and returns professional-grade performance analytics -- all in under a minute.

• Natural Language Strategy Development. Express trading ideas conversationally. The agentic AI interprets your intent, selects the right instruments, and generates executable strategy logic without requiring you to write code.

• High-Performance Backtesting. Strategies are compiled into a domain-specific language and executed by RaptorBT, our open-source Rust-based backtesting engine. It delivers deterministic results with 30+ industry-standard performance metrics including Sharpe ratio, maximum drawdown, CAGR, profit factor, and win rate.

• Broad Strategy Coverage. Support for single-instrument, basket, pairs trading, options (spreads, straddles, butterflies, iron condors), calendar spreads, and multi-strategy portfolios. 25+ technical indicators are available out of the box -- SMA, EMA, RSI, MACD, Bollinger Bands, Supertrend, VWAP, ADX, and more.

• Iterative Refinement. Each research session is a conversation. Adjust parameters, swap indicators, change instruments, or pivot strategy types through follow-up queries. The platform maintains full context across iterations.

• Performance Analytics. Every backtest produces equity curves benchmarked against a reference index, detailed trade-level breakdowns, drawdown analysis, and risk-adjusted return metrics.

alphabench is built for professionals and researchers in systematic trading and quantitative finance:

• Quantitative Analysts looking to accelerate strategy prototyping and reduce time spent on boilerplate data wrangling.
• Systematic Traders who want to rapidly test and iterate on trading ideas across equities, derivatives, and multi-asset portfolios.
• Portfolio Managers who need to evaluate strategy performance and risk characteristics without writing code.
• Research Teams at hedge funds, asset management firms, and proprietary trading desks who require a shared, auditable research workflow.

The core backtesting engine that powers alphabench is fully open source.

A high-performance, event-driven backtesting engine written in Rust with Python bindings via PyO3. RaptorBT is designed for speed and correctness -- benchmarked at 5,800x faster than comparable Python-based engines on equivalent workloads, with a sub-10MB footprint and fully deterministic execution.

Available on PyPI:

pip install raptorbt

The quantitative research lifecycle is powerful but fragmented. Data acquisition, strategy formulation, backtesting, and performance analysis are typically spread across disconnected tools, languages, and workflows. This friction slows down iteration and limits who can participate in systematic investing.

alphabench exists to collapse that workflow into a single, intelligent interface. Our goal is to make institutional-grade quantitative research accessible to anyone with a trading idea -- and to give experienced quants a faster path from hypothesis to validated result.

alphabench | The Speed of Thought, Quantified.