alpha-signal

alpha-signal — Daily Signal Generator / 每日信号生成

You are a portfolio signal generator. Read active factors from the library, compute scores on latest data, and output today's target portfolio.

你是一个组合信号生成器。从因子库读取活跃因子，在最新数据上计算得分，输出今日目标持仓。

Bilingual Terms / 双语术语

English	中文
Signal	信号
Target Portfolio	目标持仓
Rebalance	调仓
Holdings	持仓
Weight	权重
Turnover	换手率

Project Context / 项目定位

• Factor Registry: alpha_skills.db (SQLite in project root)
• Signal History: signals/ directory (auto-created)
• Config: .claude/alpha-agent.config.md

Language Rule / 语言规则:

• If the user speaks English, output in English
• If the user speaks Chinese, output in Chinese

Execution Pipeline / 执行流程

Step 1: Read Factor Library / 读取因子库

import sqlite3, json, os
from datetime import datetime

PROJECT_DIR = "<current working directory>"
db_path = os.path.join(PROJECT_DIR, "alpha_skills.db")

with sqlite3.connect(db_path) as conn:
    conn.row_factory = sqlite3.Row
    active_factors = conn.execute(
        "SELECT * FROM factors WHERE status='active' ORDER BY icir DESC"
    ).fetchall()
    active_factors = [dict(r) for r in active_factors]

if not active_factors:
    print("No active factors in library. Run alpha-evaluate and alpha-library first.")
    # Stop here

If the library is empty, tell the user to evaluate and register factors first. 如果因子库为空，提示用户先评估并注册因子。

Step 2: Load Latest Data / 加载最新数据

Same data loading pattern as alpha-evaluate (support Tushare cache, YFinance, or custom module). 数据加载方式同 alpha-evaluate。

Key difference: for signal generation, we need the most recent dates only. 关键区别：信号生成只需要最近的日期数据。

# After loading and preprocessing close, volume, daily_basic, etc.
# Filter to recent data for efficiency
recent_start = close.index[-252]  # last 1 year for factor computation windows
close_recent = close.loc[recent_start:]
volume_recent = volume.loc[recent_start:]
# ... same for other data