Built on AKShare, AKShare One Derivative provides:

• Unified input/output formats across 6 data sources (EastMoney, Sina, XueQiu, SSE, Exchange, AKShare)
• Two-track architecture: Track A (data fetching) + Track B (derivatives analytics)
• Built-in MCP server with 43 tools for LLM Agent integration
• Covers A-shares, futures, ETF options, and commodity options

Installation

pip install akshare-one-derivative

# With MCP server support
pip install akshare-one-derivative[mcp]

# With TA-Lib technical indicators
pip install akshare-one-derivative[talib]

Requires Python >= 3.10.

Quick Start

from akshare_one import get_hist_data, get_realtime_data

# Stock historical data (EastMoney, forward-adjusted)
df = get_hist_data("600000", interval="day", adjust="qfq")

# Futures main contract daily data (Sina)
from akshare_one import get_futures_hist_data
df = get_futures_hist_data(symbol="AG", contract="main", interval="day")

# ETF options chain (Sina)
from akshare_one import get_options_chain
df = get_options_chain(underlying_symbol="510300")

# Commodity options chain (Sina, auto-detect main contract by OI)
from akshare_one import get_commodity_options_chain
df = get_commodity_options_chain(commodity_symbol="黄金期权", contract="main")

# Calculate Greeks (Black-76)
from akshare_one.options_indicators import calculate_greeks
greeks = calculate_greeks(S=3.2, K=3.0, T=0.25, r=0.02, sigma=0.2)

# Historical volatility with Yang-Zhang estimator
from akshare_one.futures_indicators import get_historical_volatility
hv = get_historical_volatility(ohlcv_df, method="yang_zhang", window=20)

API Overview

Track A — Data Fetching

All functions return standardized pd.DataFrame with unified column names.

Track B — Derivatives Analytics

Local computation engines operating on Track A data.

Options Indicators (akshare_one.options_indicators):

Futures Indicators (akshare_one.futures_indicators):

Technical Indicators (akshare_one.indicators):

34 indicators including SMA, EMA, RSI, MACD, Bollinger Bands, Stochastic, ATR, CCI, ADX, OBV, MFI, SAR, and more. Supports TA-Lib and pure-pandas backends.

MCP Server

The built-in MCP server exposes 43 tools for LLM agents (Claude, GPT, etc.), organized in three categories:

compute_* tools are designed for LLM agents — they accept simple params (symbol, date) and handle data fetching internally, so agents never need to pass DataFrames.

Run the server

# stdio mode (for Claude Desktop, Cursor, etc.)
uvx --from "akshare-one-derivative[mcp]" akshare-one-derivative

# HTTP mode
uvx --from "akshare-one-derivative[mcp]" akshare-one-derivative --transport http --port 8000

Claude Desktop config

{
  "mcpServers": {
    "akshare-one": {
      "command": "uvx",
      "args": ["--from", "akshare-one-derivative[mcp]", "akshare-one-derivative"]
    }
  }
}

Architecture

akshare_one/
├── __init__.py              # Track A public API (28 functions)
├── options_indicators.py    # Track B options (8 functions)
├── futures_indicators.py    # Track B futures (9 functions)
├── indicators.py            # Technical indicators (34 functions)
├── mcp/
│   └── server.py            # FastMCP server (43 tools)
└── modules/
    ├── derivatives/          # Track B engines (Greeks, options, futures)
    ├── historical/           # Stock historical data providers
    ├── realtime/             # Stock realtime data providers
    ├── financial/            # Financial statement providers
    ├── futures/              # Futures data providers (Sina, AKShare)
    ├── options/              # Options data providers (ETF + commodity)
    ├── indicators/           # Technical indicator calculators
    ├── info/                 # Stock info providers
    ├── insider/              # Insider trade providers
    ├── news/                 # News providers
    └── cache.py              # TTL cache (cachetools)

All modules follow the Factory + ABC pattern: base.py (abstract interface) → factory.py (registry) → {source}.py (implementation).

Documentation

Full API documentation: https://zwldarren.github.io/akshare-one/

License

MIT