Initial commit: cc-cursor 全链路量化研究平台

7 Sprints 全部完成: Sprint 0: 基础设施 (DataManager + MariaDB) Sprint 1: 因子引擎 (34因子/12分类) Sprint 2: VectorBT 回测 (5策略+截面) Sprint 3: Optuna 优化 (+Walk-Forward) Sprint 4: ML 模型 (LightGBM+CatBoost) Sprint 5: Qwen 情绪因子 (三源新闻+日期对齐) Sprint 6: Agent 系统 (4Agent+日报.md/.html) 生产加固 (15项): Tushare双源fallback, SSH自动恢复, pool_pre_ping, save_daily先删后插, load_dotenv绝对路径, 日报5d/20d修复, RiskAgent改上证指数, 昨日对比+数据截止, mac_report utf8mb4, CLAUDE-*.md 9条已知Bug, demo全参数化, djapi数据源归一化, indexDatas API修正 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-07 15:59:05 +08:00
commit 271a9343a5
293 changed files with 59598 additions and 0 deletions
@@ -0,0 +1,115 @@
+"""
+信号生成工具函数。
+
+因子值 → 交易信号的桥梁，纯函数无副作用。
+"""
+
+import pandas as pd
+
+
+def factor_to_threshold_signal(
+    factor_series: pd.Series,
+    buy_threshold: float,
+    sell_threshold: float | None = None,
+    cross_direction: str = "up",
+) -> pd.Series:
+    """
+    因子阈值交叉信号。
+
+    参数:
+        factor_series: 因子值 Series
+        buy_threshold: 买入阈值（如 RSI < 30 则买）
+        sell_threshold: 卖出阈值（如 RSI > 70 则卖），None 表示平所有仓
+        cross_direction: 'up'=因子向上穿越阈值时触发, 'down'=向下穿越
+
+    返回:
+        信号 Series：1=买入, 0=平仓
+    """
+    signals = pd.Series(0, index=factor_series.index)
+
+    if cross_direction == "down":
+        buys = factor_series < buy_threshold
+    else:
+        buys = factor_series > buy_threshold
+
+    signals[buys] = 1
+
+    if sell_threshold is not None:
+        if cross_direction == "down":
+            sells = factor_series > sell_threshold
+        else:
+            sells = factor_series < sell_threshold
+        signals[sells] = 0
+
+    # 过滤连续信号
+    signals = _filter_consecutive(signals)
+
+    return signals
+
+
+def factor_to_quantile_signal(
+    factor_series: pd.Series,
+    top_quantile: float = 0.8,
+    bottom_quantile: float = 0.2,
+) -> pd.Series:
+    """
+    因子分位数信号 — 按滚动分位数判断。
+
+    参数:
+        factor_series: 因子值
+        top_quantile: 高于此分位买入
+        bottom_quantile: 低于此分位平仓
+
+    返回:
+        信号 Series
+    """
+    top = factor_series.quantile(top_quantile)
+    bottom = factor_series.quantile(bottom_quantile)
+
+    signals = pd.Series(0, index=factor_series.index)
+    signals[factor_series > top] = 1
+    signals[factor_series < bottom] = 0
+
+    return _filter_consecutive(signals)
+
+
+def cross_signal(
+    fast: pd.Series,
+    slow: pd.Series,
+) -> pd.Series:
+    """
+    金叉/死叉信号。
+
+    fast 上穿 slow → 买入(1)
+    fast 下穿 slow → 平仓(0)
+    """
+    fast = fast.dropna()
+    slow = slow.dropna()
+    common_idx = fast.index.intersection(slow.index)
+    fast, slow = fast[common_idx], slow[common_idx]
+
+    signals = pd.Series(-1, index=common_idx)
+    above = (fast > slow).fillna(False)
+    above = above.infer_objects(copy=False)
+    # 交叉点：今天 above=True 且昨天 above=False → 金叉
+    prev = above.shift(1).fillna(False)
+    prev = prev.infer_objects(copy=False)
+    cross_up = above & ~prev
+    cross_down = ~above & prev
+
+    signals[cross_up] = 1
+    signals[cross_down] = 0
+
+    return _filter_consecutive(signals)
+
+
+def _filter_consecutive(signals: pd.Series) -> pd.Series:
+    """过滤连续相同信号，只保留首次出现的信号。"""
+    result = signals.copy()
+    prev = None
+    for i in range(len(result)):
+        if result.iloc[i] == prev:
+            result.iloc[i] = -1  # 标记为不操作
+        else:
+            prev = result.iloc[i]
+    return result[result != -1].reindex(signals.index).fillna(-1)