feat: 多周期牛熊判定模块 — 方案一矩阵展示 + 四法投票 + 多TF策略

- engine/indicators/regime.py: RegimeDetector(四法投票) + MultiTimeframeRegime(多周期并行) 四法: EMA200斜率 / 价格vsEMA200 / ATH回撤 / 窄幅盘整(<3%振幅) 全部 O(1)/bar 增量计算,适用于回测和实时 - engine/example/regime_display.py: 多周期牛熊矩阵展示脚本独立加载各周期数据 → 运行判定 → 日线对齐矩阵 + 详细拆解 + 统计输出 engine/backtest/REGIME_MATRIX_BTCUSDT.md - engine/example/regime_mtf_strategy.py: 多周期共识策略 + 四策略对比回测 MTF Consensus: 1w定方向 + 1d确认 + 4h EMA入场 vs Old Regime(单TF基线) vs Long/Short(无过滤) - engine/indicators/__init__.py: 导出 RegimeDetector, MultiTimeframeRegime
2026-06-17 11:30:19 +08:00
parent 626acb20d3
commit 4294ec401d
5 changed files with 1202 additions and 0 deletions
@@ -0,0 +1,268 @@
+"""
+多周期牛熊判定模块
+
+基于 EMA200 斜率、价格 vs EMA200、ATH 回撤三法投票，支持单周期判定
+和多周期并行分析。每根 bar O(1) 增量更新，适用于回测和实时分析。
+
+用法:
+    # 单周期
+    det = RegimeDetector()
+    for price in closes:
+        det.update(price)
+    regime = det.detect(closes[-1], len(closes) - 1)
+
+    # 多周期
+    mtf = MultiTimeframeRegime(['1h', '4h', '1d', '1w'])
+    for price in closes:
+        mtf.update(price)
+    matrix = mtf.detect_all(closes[-1], len(closes) - 1)
+"""
+
+from .incremental import EmaInc
+
+
+class RegimeDetector:
+    """单周期牛熊判定器 — 四法投票
+
+    四个独立判定维度：
+        1. EMA200 斜率：近 20 根 bar 的 EMA200 变化率 → bull/bear/sideways
+        2. 价格 vs EMA200：当前价在 EMA200 上方/下方 → bull/bear
+        3. ATH 回撤：当前价距历史高点的回撤幅度 → bull/bear/sideways
+        4. 窄幅盘整：近 20 根 bar 收盘价振幅 < 3% → sideways（否则 unknown，不投票）
+
+    最终判定：四选二投票（>=2 票即定论）。
+    """
+
+    def __init__(self, range_period: int = 20, range_threshold: float = 0.03):
+        self._ath = 0.0
+        self._e200 = EmaInc(200)
+        self._range_period = range_period
+        self._range_threshold = range_threshold
+        self._recent_high: list[float] = []
+        self._recent_low: list[float] = []
+
+    def update(self, price: float) -> None:
+        """每根 bar 调一次：更新 ATH + EMA200 + 价格区间窗口"""
+        if price > self._ath:
+            self._ath = price
+        self._e200.update(price)
+        # 滚动窗口：维持最近 range_period 根 bar 的最高/最低（只用 close 做代理）
+        self._recent_high.append(price)
+        self._recent_low.append(price)
+        if len(self._recent_high) > self._range_period:
+            self._recent_high.pop(0)
+            self._recent_low.pop(0)
+
+    def detect(self, price: float, idx: int) -> str:
+        """返回当前 bar 的牛熊判定：bull / bear / sideways"""
+        r1 = self._ema200_slope(idx)
+        r2 = self._price_vs_ema200(price, idx)
+        r3 = self._ath_drawdown(price)
+        r4 = self._price_range()
+
+        b = sum(1 for r in [r1, r2, r3, r4] if r == "bull")
+        br = sum(1 for r in [r1, r2, r3, r4] if r == "bear")
+        sx = sum(1 for r in [r1, r2, r3, r4] if r == "sideways")
+
+        if b >= 2:
+            return "bull"
+        if br >= 2:
+            return "bear"
+        if sx >= 2:
+            return "sideways"
+        return "sideways"
+
+    def detect_detail(self, price: float, idx: int) -> dict:
+        """返回详细判定（含各子法的结果 + 三态票数）"""
+        r1 = self._ema200_slope(idx)
+        r2 = self._price_vs_ema200(price, idx)
+        r3 = self._ath_drawdown(price)
+        r4 = self._price_range()
+        b = sum(1 for r in [r1, r2, r3, r4] if r == "bull")
+        br = sum(1 for r in [r1, r2, r3, r4] if r == "bear")
+        sx = sum(1 for r in [r1, r2, r3, r4] if r == "sideways")
+        if b >= 2:
+            final = "bull"
+        elif br >= 2:
+            final = "bear"
+        elif sx >= 2:
+            final = "sideways"
+        else:
+            final = "sideways"
+        return {
+            "final": final,
+            "ema200_slope": r1,
+            "price_vs_ema200": r2,
+            "ath_drawdown": r3,
+            "price_range": r4,
+            "bull_votes": b,
+            "bear_votes": br,
+            "sideways_votes": sx,
+            "range_pct": self._range_pct(),
+        }
+
+    def _ema200_slope(self, idx: int) -> str:
+        if idx < 220:
+            return "unknown"
+        e200 = self._e200
+        if e200[idx - 20] == 0:
+            return "unknown"
+        slope = (e200[idx] - e200[idx - 20]) / e200[idx - 20]
+        if slope > 0.002:
+            return "bull"
+        if slope < -0.002:
+            return "bear"
+        return "sideways"
+
+    def _price_vs_ema200(self, price: float, idx: int) -> str:
+        if idx < 210:
+            return "unknown"
+        e = self._e200[idx]
+        if e == 0:
+            return "unknown"
+        return "bull" if price > e else "bear"
+
+    def _ath_drawdown(self, price: float) -> str:
+        if self._ath == 0:
+            return "unknown"
+        dd = (price - self._ath) / self._ath
+        if dd > -0.20:
+            return "bull"
+        if dd < -0.40:
+            return "bear"
+        return "sideways"
+
+    def _price_range(self) -> str:
+        """窄幅盘整检测：近 range_period 根 bar 振幅 < 阈值 → sideways"""
+        if len(self._recent_high) < self._range_period:
+            return "unknown"
+        highest = max(self._recent_high)
+        lowest = min(self._recent_low)
+        if highest == 0:
+            return "unknown"
+        range_pct = (highest - lowest) / highest
+        return "sideways" if range_pct < self._range_threshold else "unknown"
+
+    def _range_pct(self) -> float:
+        """当前波动区间百分比（用于展示）"""
+        if len(self._recent_high) < self._range_period:
+            return 0.0
+        highest = max(self._recent_high)
+        if highest == 0:
+            return 0.0
+        return (highest - min(self._recent_low)) / highest
+
+    @property
+    def ema200(self) -> float:
+        """最新 EMA200 值"""
+        return self._e200.current
+
+    @property
+    def ath(self) -> float:
+        """历史最高价"""
+        return self._ath
+
+    @property
+    def ready(self) -> bool:
+        """是否已有足够数据做判定"""
+        return len(self._e200) >= 220
+
+
+class MultiTimeframeRegime:
+    """多周期牛熊并行分析
+
+    管理多个 RegimeDetector 实例，每个对应一个周期。
+    所有检测器共用同一价格序列更新，保证 O(1)/bar 的增量计算。
+
+    Attributes:
+        timeframes: 周期列表（如 ['1h', '4h', '1d', '1w']）
+        detectors: 周期 → RegimeDetector 映射
+        _prices: 价格序列（所有检测器共享）
+    """
+
+    def __init__(self, timeframes: list[str]):
+        self.timeframes = timeframes
+        self.detectors: dict[str, RegimeDetector] = {
+            tf: RegimeDetector() for tf in timeframes
+        }
+        self._prices: list[float] = []
+
+    def update(self, price: float) -> None:
+        """更新所有周期检测器"""
+        self._prices.append(price)
+        for det in self.detectors.values():
+            det.update(price)
+
+    def detect_all(self) -> dict[str, str]:
+        """返回各周期当前判定（需至少 220 根 bar）"""
+        n = len(self._prices)
+        if n < 220:
+            return {tf: "unknown" for tf in self.timeframes}
+        price = self._prices[-1]
+        return {
+            tf: self.detectors[tf].detect(price, n - 1)
+            for tf in self.timeframes
+        }
+
+    def detect_all_detail(self) -> dict[str, dict]:
+        """返回各周期详细判定"""
+        n = len(self._prices)
+        if n < 220:
+            return {tf: {"final": "unknown"} for tf in self.timeframes}
+        price = self._prices[-1]
+        return {
+            tf: self.detectors[tf].detect_detail(price, n - 1)
+            for tf in self.timeframes
+        }
+
+    def detect_at(self, idx: int) -> dict[str, str]:
+        """返回指定 bar 位置各周期的判定"""
+        if idx < 0 or idx >= len(self._prices):
+            raise IndexError(f"idx {idx} out of range [0, {len(self._prices)})")
+        price = self._prices[idx]
+        result = {}
+        for tf in self.timeframes:
+            if idx < 220:
+                result[tf] = "unknown"
+            else:
+                result[tf] = self.detectors[tf].detect(price, idx)
+        return result
+
+    def score(self) -> float:
+        """加权综合评分 [-1, 1]
+
+        各周期赋分：bull=+1, sideways=0, bear=-1
+        权重：周期越长权重越大
+
+        Returns:
+            -1.0 (强熊) ~ +1.0 (强牛)
+        """
+        n = len(self._prices)
+        if n < 220:
+            return 0.0
+
+        # 周期权重：1h=1, 4h=2, 1d=3, 1w=4, 1mon=5
+        tf_weights = {
+            "1m": 1, "3m": 1, "5m": 1, "15m": 2, "30m": 2,
+            "1h": 3, "2h": 3, "4h": 4, "6h": 4, "8h": 4,
+            "1d": 5, "1w": 6, "1mon": 7,
+        }
+
+        score_map = {"bull": 1.0, "sideways": 0.0, "bear": -1.0, "unknown": 0.0}
+        regimes = self.detect_all()
+        total_weight = 0.0
+        weighted_score = 0.0
+
+        for tf, regime in regimes.items():
+            w = tf_weights.get(tf, 1)
+            weighted_score += score_map[regime] * w
+            total_weight += w
+
+        if total_weight == 0:
+            return 0.0
+        return weighted_score / total_weight
+
+    @property
+    def ready(self) -> bool:
+        """所有检测器是否都已就绪"""
+        return len(self._prices) >= 220