Polymarket AI Trading Bot: ML Ensemble, LLM, and Copy Trading in Rust

⚡ The gist in 5 seconds - The polytutor-ai/polymarket-ai-trading-bot project has been published on GitHub — a trading bot for Polymarket prediction markets. - Decisions are made by a combination of an ML ensemble and an OpenAI LLM; a separate module copies trades from the leaderboard. - Infrastructure: Rust services, a Postgres ledger, and a Python component (the repository description cuts off at "a Python mo…"). ### 🔍 What was found The polytutor-ai/polymarket-ai-trading-bot repository hosts an AI bot for trading on Polymarket, a prediction market platform. According to the project description, decision-making is handled by an ML ensemble paired with an OpenAI LLM, alongside a separate copy-trading bot that mirrors trades from the leaderboard. The backend is built from Rust services, operations are tracked in a Postgres ledger, and a Python module is also mentioned — the full name of the component is cut off in the description. The mechanics are described as combined: ML models generate a signal, the language model participates in decision selection, and the copy-trading bot translates the behavior of traders from the leaderboard into its own orders. How the ensemble weights are structured, which models are used, and what logic selects leaders for copying is not clear from the description. This is more of an open-source framework than a ready-to-run "black box": it won't work without configuring OpenAI keys and connecting to Polymarket. ### 💡 Why it matters Prediction markets like Polymarket have long attracted automation enthusiasts, but most solutions are closed-source. Here, a public repository combines an AI stack (ML ensemble + LLM) with Rust and Postgres infrastructure — lowering the barrier to entry for anyone who wants to dissect such an architecture piece by piece. The leaderboard-based copy trading is of particular interest: it's an attempt to port a social strategy from stock trading to prediction markets. Whether this approach is profitable cannot be judged from the repository description — there is no data on results, backtests, or metrics in the post.