# AI use cases for Vibe Trading (HKUDS/Vibe-Trading) Repo: [HKUDS/Vibe-Trading](https://github.com/HKUDS/Vibe-Trading) | Use cases: 3 | Page: https://dedupedai.com/use-cases/hkuds-vibe-trading/ Vibe-Trading is HKUDS's personal trading agent: you describe research or a strategy in plain English and it fetches market data, writes and runs backtests, and reports return, drawdown and other metrics. Recent releases add daily paper trading against completed bars with saved paper accounts, a large Alpha Zoo of factors, email and chat reports, many LLM providers and an MCP server so other agents can use its tools. Not financial advice: paper trade first, and remember you can lose money. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Backtest a trading strategy described in plain English: https://dedupedai.com/use-cases/hkuds-vibe-trading/backtest-a-strategy-from-plain-english/ - Paper trade a strategy on daily bars: https://dedupedai.com/use-cases/hkuds-vibe-trading/paper-trade-a-strategy-on-daily-bars/ - Give your coding agent trading research tools over MCP: https://dedupedai.com/use-cases/hkuds-vibe-trading/give-your-agent-trading-research-tools-over-mcp/ ## How to backtest a trading strategy described in plain English with Vibe Trading Repo: [HKUDS/Vibe-Trading](https://github.com/HKUDS/Vibe-Trading) | Difficulty: Intermediate | Time: 20 minutes | Basis: From the README Install the package, run the setup, and ask for a backtest in one sentence. **Who it is for:** A developer with a strategy idea who does not want to write data loading and backtest code by hand. **The problem:** Testing even a simple moving-average idea means wiring data, signals and metrics together first. **The outcome:** A finished backtest with a summary of return and drawdown, and the generated code saved in the run directory. ### Steps 1. Install with `pip install -U vibe-trading-ai` 2. Run the interactive setup (vibe-trading init in the README quick start) to write your .env with an LLM key 3. Ask for a backtest with vibe-trading run and a prompt, for example a BTC-USDT 20/50 moving-average strategy for 2024 4. Read the run card: settings it did not recognise are reported, and values it cannot compute come back as null instead of a guess ### Prompt to paste into your agent ```text Install HKUDS/Vibe-Trading and backtest on for . Report return, max drawdown and number of trades, and list any backtest setting the run card flagged as unrecognised. ``` ### Caveat A backtest is not a forecast, and the project warns it has never launched a token, so ignore any coin using its name. Not financial advice: paper trade first, and remember you can lose money. Risk, credentials: Your LLM key lives in the runtime .env, so keep it private and out of shared folders. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/hkuds-vibe-trading/backtest-a-strategy-from-plain-english/ ## How to paper trade a strategy on daily bars with Vibe Trading Repo: [HKUDS/Vibe-Trading](https://github.com/HKUDS/Vibe-Trading) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Save a paper account and step it forward against completed daily bars without ever reaching a broker. **Who it is for:** A learner who backtested a strategy and wants to see how it behaves going forward before any real money. **The problem:** Backtests look good in hindsight; you need a forward test that cannot peek at the future. **The outcome:** A saved paper account that moves forward one completed day at a time and never touches a broker. ### Steps 1. Install or upgrade with `pip install -U vibe-trading-ai` 2. Create and manage saved paper accounts with `vibe-trading paper`, or from the REST API or Web page 3. Let the agent match orders or a strategy against completed daily bars with the paper_trade_step tool 4. Check the account each day; saved accounts only move forward ### Prompt to paste into your agent ```text Using HKUDS/Vibe-Trading, create a saved paper account and paper trade on against completed daily bars. Each day report positions, cash and PnL, and confirm no broker connector is enabled. ``` ### Caveat Paper fills on completed daily bars ignore real slippage and liquidity, and live broker trading is marked experimental. Not financial advice: paper trade first, and remember you can lose money. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/hkuds-vibe-trading/paper-trade-a-strategy-on-daily-bars/ ## How to give your coding agent trading research tools over MCP with Vibe Trading Repo: [HKUDS/Vibe-Trading](https://github.com/HKUDS/Vibe-Trading) | Difficulty: Intermediate | Time: 20 minutes | Basis: From the README Run the Vibe Trading MCP server so Claude Code, Codex or another MCP client can fetch data and run backtests. **Who it is for:** An agent user who wants market data, factor research and backtests inside the assistant they already use. **The problem:** General agents have no reliable market data or backtest engine of their own. **The outcome:** Your MCP client calling Vibe Trading tools for research and backtests, with trading connectors left off. ### Steps 1. Install with `pip install -U vibe-trading-ai` 2. Add an MCP server entry whose command is `vibe-trading-mcp` to your client config 3. Set any environment variables in the client's env block, since a shell export does not reach a spawned server 4. Start with a research-only smoke test: discover tools and run one market data or backtest request 5. Only select a trading connector profile later, and prefer a paper profile ### Prompt to paste into your agent ```text Add HKUDS/Vibe-Trading as an MCP server (command vibe-trading-mcp) to , list the tools it exposes, and run one backtest of through it. Do not enable any trading connector. ``` ### Caveat Shell tools stay off unless you opt in at startup; keep it that way for research. Not financial advice: paper trade first, and remember you can lose money. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/hkuds-vibe-trading/give-your-agent-trading-research-tools-over-mcp/