feat: 测温机制定稿——三态骨架+温度(rsi)连续维度(数据验证) 回填market_regime 10年(2016-2026 2319条); 温度分档temp_band(panic/fear/neutral/greed/euphoria); strategy_router v2三态选策略+温度乘数; 废弃四态thermometer; 策略-适用温度表(v_weak=choppy×fear, v_oversold=trend_down×panic, v_next4=trend_up 63%)

2026-08-13 09:13:04 +08:00
parent 9310149171
commit 06fbdf5961
4 changed files with 286 additions and 232 deletions
@@ -0,0 +1,111 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""temp_band.py — 市场温度分档模块（2026-08-13）
+
+结论（数据验证）：三态骨架 + 温度(rsi)连续维度。
+- 三态（trend_up/choppy/trend_down）= market_regime.py 已有（策略类型分类）
+- 温度（rsi 分档）= 本模块（力度维度，仓位乘数）
+
+数据依据（2026-08-13 分市场状态检验）：
+  v_weak choppy: 恐慌<45 -> 71%胜率 / 中性45-60 -> 46% / 亢奋>60 -> 37%
+  v_oversold trend_down: 恐慌<35 -> 64% / 偏弱35-50 -> 53% / 中性>50 -> 0%
+  s2_panic trend_down 恐慌<35: 86%胜率 +16.6%
+
+档位：panic(<35) / fear(35-45) / neutral(45-60) / greed(60-70) / euphoria(>70)
+用法：
+  from temp_band import get_market_temp, temp_multiplier
+  t = get_market_temp()      # {rsi, band, regime}
+  mult = temp_multiplier(t["band"], strategy_family="mr")
+"""
+import sqlite3
+from pathlib import Path
+
+DB = Path("/home/hmo/MoFin/data/mofin.db")
+INDEX = "sh000001"
+
+
+def calc_rsi(series, n=14):
+    result = [None] * len(series)
+    if len(series) < n + 1:
+        return result
+    gains, losses = [], []
+    for i in range(1, len(series)):
+        ch = series[i] - series[i - 1]
+        gains.append(max(ch, 0))
+        losses.append(max(-ch, 0))
+        if i >= n:
+            avg_g = sum(gains[i - n:i]) / n
+            avg_l = sum(losses[i - n:i]) / n
+            rs = avg_g / avg_l if avg_l > 0 else 100
+            result[i] = 100 - 100 / (1 + rs)
+    return result
+
+
+def temp_band(rsi):
+    """连续 rsi -> 温度档位"""
+    if rsi is None:
+        return "unknown"
+    if rsi < 35:
+        return "panic"
+    if rsi < 45:
+        return "fear"
+    if rsi < 60:
+        return "neutral"
+    if rsi < 70:
+        return "greed"
+    return "euphoria"
+
+
+def temp_multiplier(band, strategy_family="mr"):
+    """温度 -> 仓位乘数。mr=均值回复（恐慌重仓），trend=趋势（恐慌回避）"""
+    if strategy_family == "trend":
+        return {
+            "panic": 0.0, "fear": 0.0, "neutral": 0.5,
+            "greed": 1.0, "euphoria": 0.5, "unknown": 0.5,
+        }.get(band, 0.5)
+    return {
+        "panic": 1.5, "fear": 1.0, "neutral": 0.8,
+        "greed": 0.5, "euphoria": 0.3, "unknown": 0.5,
+    }.get(band, 0.5)
+
+
+def get_market_temp(db_path=None):
+    """读取最新市场温度（rsi + 档位 + regime）。"""
+    db = db_path or DB
+    conn = sqlite3.connect(str(db), timeout=5)
+    try:
+        # 上证最近 30 日收盘算 rsi
+        rows = conn.execute(
+            "SELECT date, close FROM stock_daily WHERE code=? ORDER BY date DESC LIMIT 30",
+            (INDEX,)
+        ).fetchall()
+        # 最新 regime
+        reg = conn.execute(
+            "SELECT date, regime FROM market_regime ORDER BY date DESC LIMIT 1"
+        ).fetchone()
+    finally:
+        conn.close()
+    if not rows or len(rows) < 15:
+        return {"rsi": None, "band": "unknown", "regime": "unknown", "date": ""}
+    rows = list(reversed(rows))
+    closes = [r[1] for r in rows]
+    dates = [r[0] for r in rows]
+    rsi_series = calc_rsi(closes)
+    rsi_now = rsi_series[-1] if rsi_series else None
+    band = temp_band(rsi_now)
+    regime = reg[1] if reg else "unknown"
+    reg_date = reg[0] if reg else ""
+    return {
+        "rsi": round(rsi_now, 1) if rsi_now is not None else None,
+        "band": band,
+        "regime": regime,
+        "date": dates[-1],
+        "regime_date": reg_date,
+    }
+
+
+if __name__ == "__main__":
+    t = get_market_temp()
+    print(f"温度: rsi={t['rsi']} band={t['band']} regime={t['regime']} ({t['date']})")
+    print(f"  mr 乘数: {temp_multiplier(t['band'], 'mr')}")
+    print(f"  trend 乘数: {temp_multiplier(t['band'], 'trend')}")