// the take
OctoBot is a Python trading bot that's been running since 2018, supporting grid, DCA, and TradingView webhook strategies across 15+ exchanges via CCXT. It ships with paper trading, backtesting, a web UI, Telegram interface, and a mobile app. Aimed at individual crypto investors who want automation without writing code.
The plugin architecture (called 'tentacles') is genuinely well-designed — strategies, evaluators, and exchange connectors are all swappable without touching core code. CCXT as the exchange abstraction layer is the right call; you get 15+ exchanges without maintaining 15+ API clients. The built-in backtesting engine runs against historical exchange data rather than synthetic candles, which gives more honest results. Eight years of active development means the project has survived multiple market cycles and API-breaking exchange changes — that's not nothing.
The 'AI trading bot' feature is LLM-prompt-to-trade-decision, which is statistically a coin flip dressed up in GPT branding — the README doesn't warn you that language models have no edge on price prediction. The codebase is split across a dozen separate repos (OctoBot-Trading, OctoBot-Commons, OctoBot-Backtesting, etc.), making local development setup genuinely painful; the contributing guide basically says 'use our Gitpod config' to avoid dealing with it. No mention of position sizing, risk management, or drawdown controls in the documented strategies — a grid bot on a 90% drawdown coin will just keep buying all the way down. The cloud platform (octobot.cloud) and the open-source bot are increasingly divergent; several features shown in the README screenshots are cloud-only, which isn't clearly labeled.
We read the README, the directory tree and a key source file in June 2026. Nobody pays to be listed. How we judge →