DOCSEARCHbyAlgolia
Submit Docs

Vibe-Trading

repository·main·Indexed 12 days ago

https://github.com/hkuds/vibe-trading

A natural-language finance research AI agent with backtesting, featuring a multi-stack architecture with a FastAPI backend, React frontend, and Electron desktop shell. It includes a valuation engine for DCF and Comps, a quantlib module with 265 financial mathematics functions, and a Strategy Development Manager for converting research into registered factors. Supports institutional data tools, options strategy analysis, and multiple IM channel adapters.

Tokens
375.8K
Snippets
730
Records
1.5K
Agent score
96%

What's inside Vibe-Trading

  1. Overview of the Candlestick Pattern Recognition Skill

    main
    The candlestick skill is a strategy engine that uses a pure pandas vectorized implementation to identify 15 classic candlestick patterns. It categorizes patterns into Single-Candle, Double-Candle, and Triple-Candle types to generate a composite trading signal based on bullish and bearish scores.

  2. Overview of Social Media Intelligence skill

    main
    The social-media-intelligence skill is a tool designed for financial signal extraction. It integrates collection methods and quantitative applications across four major platforms: Twitter/X, Telegram, Discord, and Reddit. It is intended to support sentiment-driven trading strategies by monitoring ecosystems like FinTwit, crypto intelligence channels, quant communities, and retail sentiment barometers.

  3. Overview of the Private-Company Research Skill

    main

    The private-company-research skill is a deep research framework designed for analyzing unlisted/pre-IPO companies (e.g., SpaceX, Stripe, ByteDance). Because information is scarce for private entities, the framework uses a multi-lens approach to recover a company's true value rather than just its market valuation.

    Core Principles:

    • Information Scarcity Management: Every data point must include a confidence label (🟢high, 🟡medium, 🔴low). Inference must be explicitly separated from verifiable facts.
    • Bias Mitigation: The framework actively counters AI biases like 'false conservatism' (being too vague due to lack of data), 'false precision' (guessing to fill templates), and 'comparables trap' (applying public-market logic to private companies).
    • Output: Produces a fair-value range, exit-path analysis, and an information-gap map.

  4. Overview of Eastmoney (东方财富) Data Source

    main

    The eastmoney skill provides free, unauthenticated access to market and disclosure data via Vibe-Trading's internal wrappers. It covers A-shares, Hong Kong stocks, and US stocks, providing data on fund flows, dragon-tiger lists, margin trading, block trades, shareholder counts, lock-up expiries, industry sectors, research reports, financial news, and financial statements.

    Important: Rate Limiting Eastmoney enforces rate limits by source IP and may temporarily ban IPs for sudden bursts of requests. All tools use a shared per-host throttling layer (backtest.loaders._http). Do not bypass the tools to make raw HTTP requests directly to the endpoints.

    You can adjust the minimum request interval using the environment variable: VIBE_TRADING_EASTMONEY_MIN_INTERVAL (default is 1.0 second).

  5. Overview of SEC EDGAR Filing Analysis Skill

    main

    The edgar-sec-filings skill provides a framework for analyzing US public company filings from the SEC EDGAR system. It is designed to extract fundamental insights, risk signals, and investment-relevant information from various filing types.

    Supported Filing Types:

    • 10-K: Annual reports (comprehensive fundamentals).
    • 10-Q: Quarterly reports (trend/inflection detection).
    • 8-K: Current events (catalysts/risk triggers).
    • DEF 14A: Proxy statements (governance quality).
    • Form 4: Insider transactions (insider conviction).
    • 13F: Institutional holdings (smart money positioning).
    • SC 13D/G: Ownership stake disclosures (activist signals).

  6. Overview of ashare-pre-st-filter skill

    main

    The ashare-pre-st-filter skill is a risk-analysis framework designed to predict whether an A-share stock (Main Board, ChiNext, or STAR Market) will be flagged with ST/*ST risk in the next fiscal year. It evaluates risks based on revenue, profit, net assets, and dividends, while also incorporating regulatory penalty records from Sina as independent evidence.

    Key Constraints:

    • Scope: Only applies to A-shares (Main Board, ChiNext, STAR Market). It explicitly rejects Hong Kong stocks, US stocks, Crypto, Commodities, and the Beijing Stock Exchange (8xxxxx.BJ).
    • Purpose: Forward-looking prediction of ST/*ST risk; it does not predict financial fraud.
    • Output: Provides a dual-axis result: Risk Level (based on threshold breaches) and Prediction Confidence.
    • Dependencies: Requires the tushare skill for primary data. If data is missing, it falls back to the akshare skill.

  7. Overview of Vibe-Trading

    main
    Vibe-Trading is a personalized trading agent designed to provide comprehensive trading capabilities through a single command. It is built using a modern stack including Python 3.11+, FastAPI for the backend, and React 19 for the frontend. The core AI logic is available via the vibe-trading-ai package on PyPI.

  8. Overview of yfinance skill

    main

    The yfinance skill provides a global market data interface for retrieving OHLCV (Open, High, Low, Close, Volume) and research data for US, HK, and Canadian stocks, ETFs, and indices via Yahoo Finance. It is completely free and requires no API key.

    Key usage modes:

    • Backtesting: Use source: "yfinance" or source: "auto" in your configuration to invoke the built-in yfinance DataLoader.
    • Agent/Swarm Work: Use the get_market_data tool for OHLCV bars. This tool normalizes symbols and returns strict JSON.
    • Research/Deep Data: Use direct yfinance calls for data outside OHLCV coverage, such as company info, financial statements, options, holders, and insider transactions.

  9. SEC EDGAR Coverage and Data Lag Limitations

    main

    When using SEC EDGAR data, be aware of the following constraints:

    • Geographic Scope: Coverage is limited to the United States only. Non-U.S. symbols will fail to resolve to a CIK.
    • Data Lag:
      • Filings appear only after submission.
      • XBRL companyfacts data typically trails the actual filing.
      • For specific nuances regarding reporting lag for Form 4 (insider) and 13F filings, refer to the edgar-sec-filings methodology skill.

  10. Use QVeris for premium market data

    main

    QVeris is a paid capability marketplace providing global multi-asset data (market, fundamental, macro, derivatives, crypto, China/HK, news, filings, and alternative data).

    When to use QVeris:

    • When built-in free Vibe-Trading sources lack coverage or depth.
    • When a premium provider is required (e.g., options Greeks, analyst/calendar feeds, or specific China/HK coverage).
    • When the user explicitly requests QVeris.

    Default Behavior: For ordinary OHLCV (Open, High, Low, Close, Volume) data, keep source: "auto" to use free loaders unless the user explicitly selects source: "qveris".

  11. Disclaimer and Risk Warning for Vibe-Trading

    main

    Vibe-Trading is a research and trading software. It is not investment advice. The software does not hold funds or operate exchanges.

    Key safety and risk considerations:

    • Broker Channels: Trading is performed only through broker channels explicitly authorized by the user (e.g., Robinhood Agentic Trading).
    • Control: Trading can be stopped at any time within the limits of the configured restrictions.
    • Experimental Nature: Broker trading features are experimental and have not been verified by the developers on actual broker accounts. Use at your own risk.
    • Performance: Past performance does not guarantee future results.