Built an open-core trading bot framework — free on GitHub, paid version on Gumroad

I've been running live crypto futures bots for a while, and kept rebuilding the same infrastructure (order execution, position sizing, exchange-state sync, PnL tracking) every time I tried a new strategy. So I extracted it into a framework that separates strategy logic from infrastructure.

Design choices I made:

- strategy_base.py is the only interface a new strategy needs to implement — signal generation is fully decoupled from order execution

- DRY_RUN mode works fully with no real exchange keys needed, so you can validate strategy logic before ever touching live orders

- Handles the boring-but-critical stuff: duplicate-entry prevention, exchange-state resync after restarts, shadow signal tracking for strategies you haven't gone live with yet

It's open-core: the framework and an example MA-crossover strategy are fully open on GitHub. The real order-execution implementation is a paid add-on ($59) on Gumroad for anyone who wants the finished version instead of building it themselves.

Happy to answer questions about the architecture or trade-offs I ran into.