dr-manhattan

Real-time data is critical for prediction market trading, and dr-manhattan provides built-in WebSocket support for streaming live market data. This feature enables developers to receive continuous updates on order book changes, trade executions, and market price movements without constant polling. The WebSocket connections are automatically managed by the library, ensuring reliable, low-latency data feeds. For applications that require up-to-the-second information—such as high-frequency trading bots or real-time dashboard displays—this eliminates the overhead and complexity of manually implementing WebSocket handlers. dr-manhattan’s WebSocket implementation is integrated directly into the unified interface, so developers can subscribe to streams using the same exchange-agnostic patterns they use for REST calls. This seamless transition between request-response and streaming modes empowers developers to build responsive, event-driven systems that react instantly to market changes.

dr-manhattan includes a Strategy Framework that provides a base class for building trading strategies. This framework handles common infrastructure tasks such as order tracking, position management, and event logging, so developers can focus on strategy logic. It also integrates with the Order Management system, which offers standardized methods to create, cancel, and track orders across all exchanges. Error handling is unified, meaning that regardless of which exchange returns an error, the developer sees a consistent error format. The combination of these features allows users to rapidly prototype and backtest strategies, then deploy them to live markets with minimal friction. Additionally, the library is fully type-safe, with extensive type hints that catch errors at compile time and provide superior autocomplete in modern IDEs, further reducing bugs and development time.

Using dr-manhattan begins with installation via the uv package manager: uv pip install -e git+https://github.com/guzus/dr-manhattan. Once installed, developers import the library and initialize an exchange object by passing configuration options such as API keys and timeouts. They can then call methods like fetch_markets() to retrieve all available markets from that exchange. The library’s clean architecture makes it easy to add support for new exchanges: one simply implements the abstract methods defined in the base exchange class. The entire codebase is open source under the MIT license, hosted on GitHub, with a community-supported ecosystem including a Telegram group and a Twitter account. This design encourages contributions and ensures that the library evolves to meet the needs of the prediction market community.

Developers leverage dr-manhattan to build trading bots that automatically execute strategies across multiple prediction markets. For example, a quant trader might use the framework to implement a sentiment-based strategy that monitors Polymarket’s election markets and Kalshi’s economic indicators, executing trades when conditions align. Market researchers can use the unified API to aggregate data from all exchanges for analysis, constructing dashboards that show price discrepancies or liquidity trends. Another use case is arbitrage: by fetching order books from several platforms simultaneously, a bot can identify and exploit price differences before they disappear. The outcome is a significant acceleration in development speed, reduced code complexity, and the ability to adapt quickly to new market opportunities. These capabilities empower users to focus on strategy and analysis rather than API integration.

dr-manhattan is designed for developers, quantitative analysts, and cryptocurrency or prediction market enthusiasts who need a robust, flexible API layer. Target users include individual traders building personal bots, startups creating prediction market applications, and researchers analyzing market behavior. The library supports Polymarket, Kalshi, Opinion, Limitless, and Predict.fun, with the potential to add more through its extensible design. It is completely open source under the MIT license, with no pricing or subscription required—developers can use it freely and even contribute to its development. In summary, dr-manhattan delivers on its promise of being the "CCXT for prediction markets": a simple, scalable unified API that dramatically simplifies cross-platform trading and data access, enabling the prediction market community to build richer, more powerful applications with less effort.