FAQ

FAQ

Is this financial advice? No. All output is educational analysis from public data via simple rules. The BUY/HOLD/SELL scores are a screening aid — do your own research.

Does it work only for Indian stocks? Defaults and examples are NSE/BSE-focused, and that's the design intent, but US tickers (AAPL) work fine too since Yahoo Finance covers both.

Do I need any API key? No. Yahoo Finance needs nothing. Alpha Vantage is optional (US-symbol overview only).

Do I need MongoDB? Only for the decision journal. Without it, the journal falls back to a local JSON file.

Where does the BUY/HOLD/SELL score come from? Fixed point adjustments around a neutral 50: trend (price vs 50/200-day SMA), momentum (RSI-14, MACD), 6-month return, and fundamentals (P/E, leverage, margin, dividend). Every input is returned in the data_snapshot block. See Tools-Reference — and note the Backtesting findings before trusting the weights.

How is this different from TradingAgents? Same workflow shape (reports → bull/bear debate → risk check → journaled decision), but as a lightweight MCP server where the client LLM does the reasoning and the server guarantees grounded data. See TradingAgents-Inspiration.

How do I know if the recommendations are any good? Two mechanisms: the backtest harness (rules vs history) and the decision journal (live calls vs outcomes). Current evidence: modest short-horizon momentum edge, little else — treat scores accordingly.

Why HOLD so often? Both stocks in a bearish-trend-but-fair-valuation state honestly score HOLD. The system prefers "no edge" over false confidence.

Can multiple people use one server instance? Each client session spawns its own server process; the MongoDB journal is shared across all of them.