Core pub/sub and API gateway are stable.
Mirror every trade, stream, and exchange in a live event-driven simulation. Test strategies against production data — without production risk.
Quantitative developers and small trading firms who write their own algorithms — and are tired of rebuilding exchange integrations, stitching together vendor tools, and maintaining parallel production and backtest stacks.
Most teams are running a fragile patchwork of vendor services held together by bespoke code and institutional knowledge.
Every exchange, broker, and data feed speaks its own protocol. With Virtufin, only the edge is venue-specific: a WebSocket connection plus a market-data worker that maps the venue's messages onto shared contracts. Strategies, indicators and executors don't change when you switch venues.
The contract between services lives in Slack threads and Postman collections. Virtufin bakes the contract into a published specification — every service knows the rules.
Adding a new exchange, strategy language, or transport is a six-month project. Virtufin makes it a configuration change. The platform handles the rest.
The same live market data that drives your production stack also feeds a parallel simulation environment — your digital twin.
Every market event flows through a single pub/sub fabric. Each event belongs to a world — act for the live market, hyp.<name> for a named hypothetical one — and each strategy run to a scenario (sc.<id>), so production and simulation never collide.
Virtufin is not a framework. It's a platform — pluggable, language-agnostic, and governed by a published specification.
Binance, Alpaca, any WebSocket source
Pub/sub fabric · act/hyp worlds · API
.NET, Python, or native via a C ABI
Write workers in .NET or Python, or in C, C++ or Rust as a shared library against the native C ABI. .NET and native workers run inside the worker runtime process; Python and C# source workers run in a child process.
Binance and Alpaca today, any WebSocket-based feed tomorrow. Adding a venue means three small pieces: a WebSocketManager connection, a naming fragment for its topics, and a market-data worker that decodes its messages into the shared candle, trade and order-book contracts. Everything downstream consumes those contracts unchanged.
All inter-service traffic flows as CloudEvents on a Dapr pub/sub bus. A scenario puts the market, portfolio and strategy each in the live world or a named hypothetical one, so a strategy can trade against a simulated executor without touching production.
Every order, fill, and signal emits a structured event. OpenTelemetry traces every call through the gateway. No black boxes. No mystery latency.
Start with the docs, deploy a sample worker, and see your first digital twin. Needs Dapr and the .NET SDK.