Launching today
godzilla.dev
Self-hosted crypto funding rate arbitrage & market making
9 followers
Self-hosted crypto funding rate arbitrage & market making
9 followers
godzilla.dev is an open-source C++/Python framework for self-hosted crypto funding rate arbitrage and ultra low-latency market making, with enterprise private deployment. Unlike cloud bot platforms, strategies, API keys, and execution all run on your own infrastructure. Python for strategy logic, C++ for the execution path. Install via pip (godzilla.dev on PyPI), extend with your own signals, deploy anywhere. In production use at institutional trading desks.
A backtesting harness that replays historical funding rate data against the C++ execution path would save so much time. Right now it feels like you have to be live to really validate anything, and a solid simulated venue for paper trading the latency-sensitive strategy logic side-by-side with real market snapshots would make the whole framework way more approachable for newer contributors.
@serhatbyo Great point — and you're the second person in this thread to ask for exactly this, so it's clearly a real gap. The design goal would be replaying historical funding rates and book snapshots through the same C++ execution path used live, so the only difference between paper and production is the venue adapter. It's on the roadmap now;
Have you considered adding built-in support for funding rate forecasting models out of the box, maybe something like a simple LSTM or linear regression wrapper that plugs into the Python strategy layer? Would save a lot of boilerplate for anyone coming in fresh and wanting to test signal ideas quickly.
@ensar33961 Interesting idea. My instinct is to keep forecasting models out of core — signal research styles vary too much and I don't want to bless one approach — but a clean interface where any predictor plugs into the Python strategy layer, plus a linear-regression reference implementation as a template, feels right. Boilerplate reduction without opinion lock-in. Would that cover what you had in mind?
Congrats on the launch, running strategies on your own infra is a big deal for anyone worried about custody. One thing that would help adoption is a built-in backtest harness that replays historical funding rates and order book snapshots against the same C++ execution path, so you can validate signal changes without touching live capital. Would also make new strategy onboarding way less stressful.
@turgay76925 Thanks! You and Serhat independently landed on the same request, which tells me the priority. Same execution path for backtest and live is exactly the right constraint — otherwise the backtest lies to you.
Love that the execution path lives in C++ while strategy logic stays in Python, that's the right split for latency-sensitive work without making iteration a pain. Self-hosting the keys and strategies feels overdue for this space.
@aylindurma83884 Thanks! The C++/Python split was the single most-debated decision early on — glad it resonates. "Self-hosting feels overdue" is pretty much the founding thesis.