yonatangross Global Stock Analysis
Instructions
Step 1 — Resolve the ticker and market
Confirm the symbol and its exchange (NASDAQ, NYSE, or S&P 500 membership). US symbols are alphabetic (AAPL, NVDA). If the user gives a company name, resolve to a ticker first. Any Israeli (.TA) symbol is out of scope — route those to tase-stock-analysis.
Step 2 — Fetch market data
Run: python scripts/fetch_market_data.py --ticker AAPL --period 1y This returns OHLCV JSON from yfinance. yfinance is unofficial and rate-limited; cache results and back off on HTTP 429. Add --fundamentals for P/E, P/S, margins, and growth.
Step 3 — Compute technical indicators
See references/indicators.md for formulas and thresholds. Compute RSI(14), MACD(12,26,9), Bollinger(20,2), EMA 20/50/200, CCI(20). Flag: RSI > 70 overbought / RSI < 30 oversold; price vs EMA200 = trend.
Step 4 — Layer fundamentals
P/E, P/S, gross/operating margin, revenue growth (YoY). Compare to the sector median when available.
Step 5 — Synthesize a rationale
Give a buy / hold / sell view with the 2-3 signals that drove it and one contrary signal. Always state assumptions and data freshness.
Examples
User says: "Is NVDA overbought right now?" Result: RSI(14) read + MACD histogram direction + the level to watch, with a one-line trend call from EMA200.
User says: "Fundamental snapshot of MSFT vs its sector" Result: P/E, P/S, margins, growth table with sector-median deltas.
User says: "Show me AAPL's last year with the moving averages" Result: a text trend read (price vs EMA stack) plus an optional interactive HTML chart — Close + EMA 20/50/200 and an RSI(14) panel — written via scripts/chart.py. See examples/technical-read.md and examples/fundamentals-snapshot.md.
Visualization (optional)
A chart is a bonus, not a default. Render one only when it makes the answer clearer or the user asks to "see", "show", or "chart" the data — otherwise the text analysis stands on its own. There is no ASCII fallback, and the chart is not forced onto answers that read fine as prose.
When a chart helps, run: python scripts/chart.py --ticker AAPL --period 1y --out AAPL.html This writes one self-contained interactive HTML file (inline SVG, hover crosshair + tooltip, sortable table view, light/dark toggle) via scripts/chart.py, which chains scripts/provider.py -> scripts/indicators.py -> scripts/viz.py. Use --period 1y or longer so EMA200 has enough history to draw. The colours come from a colour-blind-safe palette and the tooltip DOM is built with textContent (no injected markup).
Gotchas
- yfinance quotes are delayed and unofficial; never present as real-time.
- Historical prices are split-adjusted — intraday levels differ from a broker.
- After-hours / pre-market moves are not in the daily bar.
- S&P 500 / NASDAQ membership drifts; the reference list is dated.
Troubleshooting
Error: "Ticker not found"
Cause: wrong exchange suffix, or a foreign line quoted as an ADR. Solution: confirm the US listing symbol; for non-US firms use the ADR ticker.
Error: "Rate limited (429)"
Cause: yfinance throttling on rapid requests. Solution: back off, batch, and cache EOD data locally.
Bundled Resources
- scripts/fetch_market_data.py — OHLCV + fundamentals via yfinance.
- scripts/provider.py — robust market-data fetch: retry-with-backoff then raise (transient errors are not swallowed), multi-key resolve(), and the only yfinance importer so a paid source can swap in untouched.
- scripts/indicators.py — pure-Python, fully typed indicators (SMA, EMA, RSI, MACD, Bollinger, CCI) shared by the fetch and chart paths.
- scripts/viz.py — self-contained interactive HTML/SVG renderer (hover crosshair + tooltip, table view, light/dark toggle); no third-party deps, colour-blind-safe palette.
- scripts/chart.py — optional-chart orchestrator: provider -> indicators -> viz.
- references/indicators.md — indicator formulas, periods, interpretation.
- references/data-sources.md — yfinance limits + provider-abstraction note.
- examples/technical-read.md, examples/fundamentals-snapshot.md — two worked end-to-end walkthroughs.