AI Trading Workshop: How to Use AI to Find Trades and Build Strategies

Welcome to the replay of our Composer Webinar: AI Trading Workshop with the Composer Team Hosted by: Michael Liu, Product Marketing Manager at Composer Guest Speakers: Ronny Li, Co-founder and Chief Data Officer at Composer Update: The Composer MCP Server mentioned in this video was sunset on April 17, 2026 and is no longer available. Here's what replaced it: • Composer API: For pulling data and building tools on top of Composer. Full docs at https://api.composer.trade/docs/index... • Trade with AI: If you were using MCP to build and test symphonies, Trade with AI does this directly in the app without any setup. Login to your Composer account to get started. Reach out to [email protected] with any questions. 00:00 Intro + Why does AI matter in trading 10:00 How to set up the Composer MCP 12:00 Using AI help you find strategies tailored to your goals 19:43 Using AI help you analyze strategies in your portfolio 28:25 More ideas on how to use AI to trade 31:10 Using AI to build an uncorrelated, multi-strategy portfolio 34:41 Using AI to generate options trade ideas to hedge your portfolio 37:41 Q&A and Wrap up Prompts used: 1. Search through the Composer symphony database and give me a quick list of the top 20 symphonies that are at least 1 year out of sample and with the lowest drawdowns but have better returns than the SPY. 2. Compare my live performance for each symphony against the backtest over the same time period. Also compare the live performance against SPY over the same time period (the backtest should already return SPY's performance in the `benchmarks` field). Use time-weighted returns to compare the performance without the impact of deposits and withdrawals. Return the results as a simple markdown table with two additional columns: "Deviation from backtest" = (live time weighted return - backtest return) "Deviation from SPY" = (live time weighted return - SPY time weighted return) Sort the results by the "Deviation from SPY" column in descending order so the best performers are at the top. Notify me about any symphonies that are severely underperforming their backtest and/or SPY. Relevant Composer tools: list_accounts get_aggregate_symphony_stats backtest_symphony_by_id (call this for each symphony returned by `get_aggregate_symphony_stats`) Tips: Avoid using your "analyze" tool because it is buggy. Note that some symphonies contain the | (pipe) character in their name, which causes issues with markdown tables. Since different clients use different markdown renderers, the simplest solution is to replace all | with -. 3. You’re running a small hedge fund and want to create a resilient multi-strategy portfolio. Select 3–5 promising strategies from the Composer database that look different in style (e.g., momentum, mean reversion, sector rotation), give me the link to them: Summarize each strategy’s strengths and weaknesses. Suggest how they might complement each other to reduce risk and increase diversification. Create an allocation plan that aims for low correlation between strategies. 4. How could I protect my portfolio with Composer options against a worst-case scenario where everything drops more than 10% in the next 90 days? Helpful Resources: Composer MCP Server: https://github.com/invest-composer/co... Article on how to automate AI using n8n: https://www.composer.trade/blog/how-t...