I've been building DepthSight for the past year - a self-hosted algorithmic trading platform that you run entirely on your own hardware, so your API keys and trading logic never touch a third-party server (like 3Commas or Veles).
Instead of writing code, it features a drag-and-drop node editor (40+ logic blocks) where you can build complex strategies, cross-reference indicators, and manage risk dynamically. It also includes an AI assistant that can generate strategy structures from text prompts or even screenshots of chart setups.
Since Lemmy is built on federation, I thought you guys might appreciate the architecture: DepthSight nodes can opt-in to connect to a central Federation Hub. This creates a community network where self-hosted nodes can share verified strategy templates, discuss trading ideas, and form a global node topology map.
Features out of the box:
- Visual strategy builder
- Federation Hub
- Dual backtesting engines
- Dynamic risk management that adapts per trading pair
- Multi-tenant support (JWT, Redis quotas) if you want to host it for others
Stack: FastAPI, Celery, Redis, PostgreSQL, React (with a mobile-optimized PWA). AGPLv3 licensed. Runs with a single deployment script (curl | bash).
Github: https://github.com/DepthSight-Pro/DepthSight
Built this solo (heavily leveraging LLMs for the heavy lifting in code generation). Would love to get some feedback from the self-hosted community, especially on the architecture or features you'd want to see added!
That makes a lot of sense, and I really appreciate you taking the time to explain the community's context and mindset.If the overarching sentiment is anti-corporate, then taking the core infrastructure out of the hands of proprietary, rent-seeking SaaS companies (who charge monthly fees just to hold your data) and open-sourcing it so anyone can run it on their own hardware feels like it aligns with that ethos-even if the domain itself is controversial.But putting the financial aspect completely aside: the reason I chose to build this is that the trading domain offers a level of raw engineering complexity that you rarely find in standard self-hosted apps. Handling real-time exchange WebSocket streams, fanning out market data via Redis, managing stateful Celery workers, and ensuring strict multi-tenant isolation (with JWT and Redis ACL quotas) is a massive architectural challenge. Even if you or others here have absolutely zero interest in participating in algorithmic trading, I was hoping the community might appreciate the architecture and the implementation itself.