Hi everyone,
I'm an independent developer, and I'd like to share something I've built: MicroFlow-L2, a C++23 market-data and feature engine for Binance USD-M Futures, with a Python/NumPy interface.
The idea is simple: keep your strategy in Python, while a separate C++ process handles market-data ingestion and feature calculations.
I'm offering a paid commercial license with source-code access, but I've also made a free technical demo so you can try the interface before deciding whether it fits your project.
WHAT I BUILT
The engine reads Binance's public top-five order-book snapshots and aggregate trades, calculates features in C++, and publishes the results through local shared memory.
Your Python code reads the published data as a NumPy matrix.
The full version includes 42 selectable feature identifiers and capacity for up to 64 configured symbols. These cover quotes, spreads, book imbalance, trade flow, returns, and cross-asset calculations.
LOW-LEVEL DESIGN
I focused on fixed-capacity data structures, SIMD calculations, and separating ingestion from Python consumers.
The project includes compile-time AVX2/AVX-512 paths, benchmark and verification material, and documentation explaining what each feature actually calculates.
In my documented tests, the engine showed fast local processing. The included benchmark material describes the tested conditions so you can evaluate the results and test it on your own setup.
These measurements cover local processing, not end-to-end latency from Binance. Performance depends on hardware, configuration, and workload.
Python can calculate these features too. What I'm offering is a separate C++ ingestion and calculation component, with source access and a ready-to-use Python interface.
TRY THE DEMO
GitHub demo:
https://github.com/microflowl2/MicroFlow-L2-DemoPrebuilt demo binaries are available for Windows and Linux. Download the package for your operating system, follow the included setup instructions, and run it. No compilation or custom integration is needed to try the demo.
The engine connects to Binance's public market-data streams, processes the data, and writes the calculated outputs to local shared memory. The included Python scripts read those outputs so you can inspect what the engine is publishing without writing your own client.
You can also monitor the program's network connections and local resource usage with your own tools while it runs.
The demo is limited to BTC, ETH, and SOL, a subset of features, and 15 minutes per launch.
No trading API keys or wallet connection are required. It does not place trades or access your exchange account.
WHAT THE LICENSE INCLUDES
- C++23 engine source code.
- Python shared-memory client.
- Configuration and CMake build files.
- Feature definitions and implementation limitations.
- Benchmark and verification material.
- Private GitHub repository access.
- Available v1.x corrective releases and asynchronous text-based support within the documented scope.
PRICE
USD 350, one-time, per purchaser entity, for internal use under the supplied license.
Payment is accepted in USDC or USDT on Arbitrum One, with the exact token and payment instructions agreed privately.
You receive the README and full license terms before paying.
IMPORTANT SCOPE
This is developer software, not a turnkey trading bot. It does not place orders, provide validated trading signals, or promise financial results.
It processes top-five snapshots, not a complete incremental order book. Evaluation and publication run on a nominal 100 ms schedule.
The commercially covered build configuration is Windows 11 x64 with AVX2/FMA. A Linux demo binary and Linux-related source are also provided, but Linux is outside the build and support commitment for this release.
If you use Python for trading research and want a separate C++ ingestion and feature component, try the demo first. If it fits your workflow, email me with your operating system and intended use, and I'll send you the full documentation and license terms before any payment.
Technical questions and feedback are welcome here.
Contact:
microflowl2@proton.meIndependent project. Not affiliated with or endorsed by Binance.
