stock_analyst

Stock Analyst Specialist

Overview

Wraps virattt/ai-hedge-fund and ZhuLinsen/daily_stock_analysis.

Orchestrates a three-stage fixture pipeline for a requested ticker string: synthetic fundamentals, synthetic technical indicators, and synthetic news sentiment. The layers are merged into a structured response with a demonstration stance signal.

No market data is loaded. Every price, ratio, indicator, article count, theme, and stance is generated deterministically from the ticker string and is unsuitable for trading or valuation.

Prerequisites

Use Python 3.11+ in a local OSS Agent Lab checkout and run pip install -e ..
Use a valid test ticker and treat the response strictly as fixture data.
Read the runtime contract for supported bounds and disclaimers.

Capabilities

analyze: Exercise the full synthetic pipeline for a ticker and period label.
finance: Return a finance-shaped fixture for integration tests.
stock_analysis: Generate price, market cap, P/E, sector, and recommendation fixtures.
technical_analysis: Generate RSI, MACD, moving-average, and signal fixtures.

Tools

Tool	Description	Side Effects
`analyze_ticker`	Generate synthetic fundamental fields	None
`technical_indicators`	Generate synthetic indicator fields	None
`news_sentiment`	Generate synthetic sentiment, count, and theme fields	None

Parameters

All parameters are passed via request.intent.parameters:

Parameter	Type	Default	Description
`ticker`	`str`	(query text)	Stock ticker symbol, e.g. `AAPL`
`period`	`str`	`"1y"`	Lookback period: `1m`, `3m`, `6m`, `1y`, `3y`, `5y`
`indicators`	`list[str]`	all	Subset of `rsi`, `macd`, `moving_averages`
`days`	`int`	`7`	News lookback window in calendar days

Instructions

Normalize a non-empty ticker and choose a supported period, indicator subset, and day count.
Run the local Python API or CLI.
Label all values and recommendations as synthetic test output.
For real analysis, obtain timestamped data from an authoritative provider and use qualified advice.

Examples

Python API

from agents.specialists.stock_analyst.agent import StockAnalystSpecialist
from oss_agent_lab.contracts import Intent, Query, SpecialistRequest

specialist = StockAnalystSpecialist()

request = SpecialistRequest(
    intent=Intent(
        action="analyze",
        domain="finance",
        confidence=0.95,
        parameters={"ticker": "AAPL", "period": "1y", "days": 14},
    ),
    query=Query(user_input="AAPL"),
    specialist_name="stock_analyst",
)

response = await specialist.execute(request)
print(response.result["summary"]["overall_stance"])  # "bullish" | "neutral" | "bearish"

CLI

oss-lab run stock_analyst "AAPL"

Output shape

{
  "ticker": "AAPL",
  "fundamental": {
    "price": 182.0,
    "market_cap": 295.4,
    "pe_ratio": 28.0,
    "recommendation": "hold",
    "sector": "Technology"
  },
  "technical": {
    "rsi": 54.0,
    "macd": {"line": 1.2, "signal": 0.8, "histogram": 0.4},
    "moving_averages": {"sma_20": 183.6, "sma_50": 185.2, "sma_200": 179.1},
    "signals": ["RSI neutral", "MACD bullish crossover", "Short-term trend above medium-term: bullish bias"]
  },
  "sentiment": {
    "overall_sentiment": "positive",
    "articles_analyzed": 23,
    "key_themes": ["earnings beat", "product launch"],
    "sentiment_score": 0.65
  },
  "summary": {
    "overall_stance": "bullish",
    "confidence": 0.715,
    "key_signals": ["Fundamental: hold (P/E 28.0)", "Sentiment: positive (+0.650)"],
    "risk_note": "Simulated outputs — not financial advice. Verify with live market data before acting."
  }
}

Output

The response combines synthetic fundamental, technical, sentiment, and summary dictionaries. The risk_note states that outputs are simulated; preserve it in every user-facing rendering.

Error Handling

Reject empty tickers, unsupported periods/indicators, and non-positive news windows.
Never infer that an upstream API, filing, exchange, or news source was contacted.
Refuse to frame the generated buy/hold/sell value as financial advice.

Resources