A four-workflow multi-agent system that produces institutional-style stock research every week. The main orchestrator runs on a Schedule Trigger every Monday at 06:20, loops over a 15-ticker watchlist with Loop Over Items, and a form trigger ("Manual Entry: Stock Analysis") allows on-demand analysis of any single ticker. Each ticker is fed through Execute Workflow into an AI Agent on the Google Gemini chat model, which must call all three sub-workflow tools in order — Technical Analysis Tool, then Trends Analysis Tool, then Fundamental Analysis Tool — and synthesizes everything into a strict JSON report: recommendation, sentiment counts, chart URL, technical write-up, top articles, hot topics. The technical sub-workflow validates the ticker via TwelveData, pulls 180 days of price history with Bollinger Bands and MACD, renders a TradingView chart image via Chart-IMG, computes Fibonacci levels and support/resistance in a Code node, and has GPT-4o Vision read the chart. The fundamental sub-workflow pulls company Overview, Income Statement, and Balance Sheet from Alpha Vantage (with 12-second waits between calls to respect rate limits) and extracts MarketCap, P/E, EPS, and two-year revenue and net-income trends. The trends sub-workflow fetches Alpha Vantage news sentiment and counts bullish, neutral, and bearish articles with hot-topic ranking. Back in the main workflow, if the recommendation title contains "Buy," a $7,000 paper buy order fires on Alpaca; Generate HTML renders the report into a styled email, Adjust HTML Colors recolors sections by sentiment, and Send Stock Analysis emails it via SMTP. These are research aids for study and decision support, not financial advice — and the buy branch can be removed to keep the system reports-only.
The retail investor's Monday watchlist review
An investor tracking fifteen US stocks used to spend Sunday evenings assembling research from five websites. The Monday 06:20 run does it while he sleeps: by breakfast, one styled HTML email covers all fifteen tickers with the same structure — technicals, fundamentals, news sentiment, recommendation. The sentiment-colored sections from Adjust HTML Colors let him triage visually before reading a word. Practical tip: edit the tickers array in the Stock List node — any NASDAQ or NYSE symbols work, and the TwelveData validation step rejects bad tickers before they waste API calls.
The investment club's shared research desk
An eight-member investment club argues from different data every month because everyone researches differently. The strict JSON report schema fixes that: every ticker gets the same sections, the same sentiment counts, the same structure, so the club debates conclusions instead of data. The top articles and hot topics fields give the meeting its discussion agenda. Because the sub-workflows are separate imported files called as tools, the club's technically inclined member can improve the technical analysis once and every future report benefits. Practical tip: move the Schedule Trigger to Sunday evening so the reports are waiting when the club meets Monday night.
The finance newsletter writer's research assistant
A newsletter author covering markets needs a research draft per stock without spending a day per stock. The form trigger is her tool: type a ticker into "Manual Entry: Stock Analysis" and get the full three-agent report on demand, with the technical write-up and top articles sections serving as the draft's backbone. She rewrites in her own voice — the workflow is a research starting point, not publishable copy. Practical tip: delete or deactivate the Buy? to Buy in Alpaca branch. A writer has no use for paper orders, and reports-only mode keeps the workflow simpler and cheaper to run.
The paper trader testing strategy ideas
A swing trader wants to test whether acting on the agent's recommendations would have worked, without risking money. The Alpaca integration already points at Alpaca's paper API: when the recommendation title contains "Buy," Buy in Alpaca places a $7,000 notional paper order, creating a real track record of the strategy's calls. The notional value is editable to match his actual position sizing. Practical tip: tighten the Buy? condition — for example, also require the overall sentiment to be Bullish — so the paper portfolio only acts on high-conviction calls, which is the version of the strategy worth evaluating.
The finance student learning real analysis
A finance student finds textbooks abstract and real analyst reports opaque. Watching this system work teaches the structure of equity research: the agent must call technical, then trends, then fundamental analysis, mirroring how professionals triangulate price action, news flow, and financials. The Fibonacci and support/resistance Code node shows how levels are computed, not just asserted, and the sentiment-colored HTML makes the relationship between news tone and recommendation visible. Practical tip: use the "swap the brain" customization — point the Google Gemini Chat Model at different Gemini models and compare how the recommendations change, which is itself a lesson in model sensitivity.