Add advanced indicators: MTF analysis, ADX, StochRSI, OBV, CMF, pivot points

- config.py: Add HTF_TIMEFRAME (1h) and HTF_CANDLE_LIMIT (50)
- data_fetcher.py: Fetch both 5m and 1h candles per symbol
- indicators.py: Add ADX, StochRSI, OBV+slope, CMF to 5m indicators;
  new functions for HTF indicators, pivot points, and their summaries
- main.py: Wire up HTF data flow (1h indicators + pivots → LLM summary)
- llm_analyzer.py: Rewrite prompt with MTF filter (1h trend alignment),
  require 2+ confirmations for BUY, confidence scoring guide

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

config.py

@@ -32,6 +32,8 @@ SYMBOL_NAMES = {
 # Candle settings
 CANDLE_TIMEFRAME = "5m"
 CANDLE_LIMIT = 100
+HTF_TIMEFRAME = "1h"
+HTF_CANDLE_LIMIT = 50
 
 # Trading parameters
 MIN_ORDER_USDT = 5

data_fetcher.py

@@ -169,17 +169,20 @@ def fetch_active_orders() -> list[dict]:
 # ---------------------------------------------------------------------------
 
 def fetch_all_market_data(symbols: list[str] | None = None) -> dict:
-    """Fetch tickers + candles for all symbols. Returns dict keyed by symbol."""
+    """Fetch tickers + candles (5m and 1h) for all symbols. Returns dict keyed by symbol."""
     symbols = symbols or config.TOP_15_SYMBOLS
     tickers = fetch_tickers(symbols)
     data = {}
     for sym in symbols:
         try:
             candles = fetch_candles(sym)
+            candles_htf = fetch_candles(sym, timeframe=config.HTF_TIMEFRAME, limit=config.HTF_CANDLE_LIMIT)
             data[sym] = {
                 "ticker": tickers.get(sym, {}),
                 "candles": candles,
+                "candles_htf": candles_htf,
             }
+            time.sleep(0.15)  # Avoid Bitfinex rate limits
         except Exception as e:
             logger.warning("Failed to fetch data for %s: %s", sym, e)
     return data

indicators.py

@@ -15,7 +15,7 @@ def calculate_indicators(candles_df: pd.DataFrame) -> pd.DataFrame:
     Returns the DataFrame with additional indicator columns.
     """
     df = candles_df.copy()
-    if df.empty or len(df) < 26:
+    if df.empty or len(df) < 30:
         logger.warning("Not enough candle data to compute indicators (got %d rows)", len(df))
         return df
 
@@ -54,9 +54,73 @@ def calculate_indicators(candles_df: pd.DataFrame) -> pd.DataFrame:
     # Volume moving average (20)
     df["vol_ma20"] = volume.rolling(window=20).mean()
 
+    # ADX(14) — trend strength
+    df["adx"] = ta.trend.adx(high, low, close, window=14)
+
+    # Stochastic RSI(14, 14, 3, 3)
+    df["stoch_rsi_k"] = ta.momentum.stochrsi_k(close, window=14, smooth1=3, smooth2=3)
+    df["stoch_rsi_d"] = ta.momentum.stochrsi_d(close, window=14, smooth1=3, smooth2=3)
+
+    # OBV (On-Balance Volume) + slope
+    df["obv"] = ta.volume.on_balance_volume(close, volume)
+    df["obv_slope"] = df["obv"].diff(5)
+
+    # CMF (Chaikin Money Flow, 20-period)
+    df["cmf"] = ta.volume.chaikin_money_flow(high, low, close, volume, window=20)
+
     return df
 
 
+def calculate_htf_indicators(candles_df: pd.DataFrame) -> pd.DataFrame:
+    """Calculate trend-context indicators on higher timeframe (1h) candles."""
+    df = candles_df.copy()
+    if df.empty or len(df) < 30:
+        return df
+
+    close = df["close"]
+    high = df["high"]
+    low = df["low"]
+
+    df["ema9"] = ta.trend.ema_indicator(close, window=9)
+    df["ema21"] = ta.trend.ema_indicator(close, window=21)
+    df["adx"] = ta.trend.adx(high, low, close, window=14)
+    df["rsi"] = ta.momentum.rsi(close, window=14)
+
+    return df
+
+
+def calculate_pivot_points(htf_df: pd.DataFrame) -> dict | None:
+    """Calculate classic pivot points from previous day's price action using 1h candles."""
+    if htf_df.empty or len(htf_df) < 24:
+        return None
+
+    df = htf_df.copy()
+    df["date"] = df["timestamp"].dt.date
+
+    dates = sorted(df["date"].unique())
+    if len(dates) < 2:
+        return None
+
+    prev_day = df[df["date"] == dates[-2]]
+    if prev_day.empty:
+        return None
+
+    h = prev_day["high"].max()
+    l = prev_day["low"].min()
+    c = prev_day["close"].iloc[-1]
+
+    pivot = (h + l + c) / 3
+    return {
+        "pivot": pivot,
+        "r1": 2 * pivot - l,
+        "r2": pivot + (h - l),
+        "r3": h + 2 * (pivot - l),
+        "s1": 2 * pivot - h,
+        "s2": pivot - (h - l),
+        "s3": l - 2 * (h - pivot),
+    }
+
+
 def summarize_indicators(df: pd.DataFrame, symbol: str) -> str:
     """Produce a concise text summary of the latest indicator values for LLM consumption."""
     if df.empty or len(df) < 2:
@@ -89,6 +153,61 @@ def summarize_indicators(df: pd.DataFrame, symbol: str) -> str:
         lines.append(f"ATR(14): {last['atr']:.6g}")
     if pd.notna(last.get("vwap")):
         lines.append(f"VWAP: {last['vwap']:.6g}")
+    if pd.notna(last.get("adx")):
+        adx_label = "趨勢" if last["adx"] > 25 else "盤整" if last["adx"] < 20 else "弱"
+        lines.append(f"ADX(14): {last['adx']:.1f} ({adx_label})")
+    if pd.notna(last.get("stoch_rsi_k")):
+        lines.append(f"StochRSI: K={last['stoch_rsi_k']:.2f} D={last['stoch_rsi_d']:.2f}")
+    if pd.notna(last.get("obv_slope")):
+        obv_dir = "流入" if last["obv_slope"] > 0 else "流出"
+        lines.append(f"OBV趨勢: {obv_dir} (slope={last['obv_slope']:.0f})")
+    if pd.notna(last.get("cmf")):
+        cmf_label = "買壓" if last["cmf"] > 0.05 else "賣壓" if last["cmf"] < -0.05 else "中性"
+        lines.append(f"CMF(20): {last['cmf']:.3f} ({cmf_label})")
+
+    return "\n".join(lines) + "\n"
+
+
+def summarize_htf(htf_by_symbol: dict[str, pd.DataFrame]) -> str:
+    """Produce a compact 1h trend context summary for LLM."""
+    lines = ["## 1小時趨勢背景"]
+    for sym in sorted(htf_by_symbol):
+        df = htf_by_symbol[sym]
+        if df.empty or len(df) < 2:
+            continue
+        last = df.iloc[-1]
+        name = config.SYMBOL_NAMES.get(sym, sym)
+
+        trend = "多頭" if last.get("ema9", 0) > last.get("ema21", 0) else "空頭"
+        adx_val = last.get("adx", 0)
+        strength = "強趨勢" if adx_val > 25 else "盤整" if adx_val < 20 else "弱趨勢"
+        rsi_1h = last.get("rsi", 50)
+
+        lines.append(f"- {name}: {trend} ({strength}, ADX={adx_val:.0f}, RSI={rsi_1h:.0f})")
+    return "\n".join(lines) + "\n"
+
+
+def summarize_pivots(pivots_by_symbol: dict[str, dict], current_prices: dict) -> str:
+    """Summarize pivot point levels relative to current price."""
+    lines = ["## 關鍵支撐/阻力 (日線樞紐)"]
+    for sym in sorted(pivots_by_symbol):
+        p = pivots_by_symbol[sym]
+        if not p:
+            continue
+        name = config.SYMBOL_NAMES.get(sym, sym)
+        price = current_prices.get(sym, 0)
+        if price <= 0:
+            continue
+
+        supports = [("S3", p["s3"]), ("S2", p["s2"]), ("S1", p["s1"]), ("P", p["pivot"])]
+        resistances = [("P", p["pivot"]), ("R1", p["r1"]), ("R2", p["r2"]), ("R3", p["r3"])]
+
+        nearest_sup = max([(n, v) for n, v in supports if v < price], key=lambda x: x[1], default=None)
+        nearest_res = min([(n, v) for n, v in resistances if v > price], key=lambda x: x[1], default=None)
+
+        sup_str = f"{nearest_sup[0]}={nearest_sup[1]:.6g}" if nearest_sup else "N/A"
+        res_str = f"{nearest_res[0]}={nearest_res[1]:.6g}" if nearest_res else "N/A"
+        lines.append(f"- {name}: 支撐 {sup_str} | 阻力 {res_str}")
 
     return "\n".join(lines) + "\n"
 

llm_analyzer.py

@@ -14,24 +14,41 @@ def analyze_market(indicator_summary: str, account_status: str) -> list[dict]:
 ## 帳戶狀態
 {account_status}
 
-## 目前市場指標摘要（5分鐘K線最新值）
+## 目前市場指標摘要
 {indicator_summary}
 
 請根據以下策略分析每個幣種並給出交易建議：
 
 ## 交易策略
-1. **趨勢確認**: EMA(9) > EMA(21) 為多頭，反之為空頭
-2. **進場訊號 (BUY)**:
-   - RSI < 35 且 MACD histogram 由負轉正（超賣反彈）
-   - 價格觸及 Bollinger 下軌且 RSI < 40（均值回歸）
-   - MACD 金叉 + EMA(9) 上穿 EMA(21)（趨勢啟動）
-3. **進場訊號 (SELL)**:
-   - RSI > 70 且 MACD histogram 由正轉負（超買回落）
-   - 價格觸及 Bollinger 上軌且 RSI > 65
-   - MACD 死叉 + EMA(9) 下穿 EMA(21)
-4. **過濾條件**:
+
+### 最高優先：多時間框架過濾（現貨，只做多）
+- 1小時趨勢為多頭（EMA9 > EMA21）時，才考慮 BUY
+- 1小時為空頭時，只持有或平倉（SELL），不開新倉
+- 1小時 ADX < 20（盤整）時，避免進場
+- 1小時 ADX > 25 且方向一致 = 高信心交易
+
+### 進場訊號 (BUY) — 需至少2個確認
+1. **超賣反彈**: StochRSI K < 0.2 且 K 上穿 D + RSI < 40
+2. **均值回歸**: 價格觸及 BB 下軌 + RSI < 40 + CMF > 0（有買壓）
+3. **趨勢啟動**: MACD 金叉 + EMA9 上穿 EMA21 + ADX > 20（有趨勢）
+4. **支撐反彈**: 價格接近樞紐支撐位（S1/S2）+ OBV 流入 + RSI < 45
+
+### 出場訊號 (SELL)
+1. **超買回落**: StochRSI K > 0.8 且 K 下穿 D + RSI > 65
+2. **阻力拒絕**: 價格接近樞紐阻力位（R1/R2）+ OBV 流出
+3. **趨勢反轉**: MACD 死叉 + EMA9 下穿 EMA21 + CMF < 0
+
+### 過濾條件（必須全部滿足）
 - 成交量需高於 20 期平均（確認動能）
-   - ATR 過高時降低倉位（波動風控）
+- 5分鐘 ADX > 15（排除極度盤整）
+- OBV 方向需與交易方向一致（量價確認）
+- ATR 過高時降低 suggested_amount_pct（波動風控）
+
+### 信心分數指引
+- 0.8+: 多時間框架對齊 + 3個以上確認指標
+- 0.6-0.8: 多時間框架對齊 + 2個確認
+- 0.4-0.6: 單一時間框架訊號，降低倉位
+- < 0.4: 不建議交易，回傳 HOLD
 
 請以 JSON 格式回傳，每個幣種一個物件：
 ```json
@@ -40,8 +57,8 @@ def analyze_market(indicator_summary: str, account_status: str) -> list[dict]:
     "symbol": "tBTCUST",
     "action": "BUY" | "SELL" | "HOLD",
     "confidence": 0.0-1.0,
-    "reason": "簡短理由",
-    "suggested_amount_pct": 0.10-0.20
+    "reason": "簡短理由（含觸發的指標）",
+    "suggested_amount_pct": 0.05-0.20
   }}
 ]
 ```

main.py

@@ -179,7 +179,7 @@ def run_cycle():
         slack_notifier.send_error_alert(f"Market data fetch failed: {e}")
         return
 
-    # 4. Calculate indicators
+    # 4. Calculate indicators (5m)
     indicators_by_symbol = {}
     current_prices = {}
     for sym, md in market_data.items():
@@ -191,6 +191,15 @@ def run_cycle():
         if ticker:
             current_prices[sym] = ticker.get("last_price", 0)
 
+    # 4b. Calculate HTF indicators (1h)
+    htf_by_symbol = {}
+    pivots_by_symbol = {}
+    for sym, md in market_data.items():
+        candles_htf = md.get("candles_htf")
+        if candles_htf is not None and not candles_htf.empty:
+            htf_by_symbol[sym] = indicators.calculate_htf_indicators(candles_htf)
+            pivots_by_symbol[sym] = indicators.calculate_pivot_points(candles_htf)
+
     # 5. Cache data (persisted for next crontab run)
     try:
         data_fetcher.cache_market_data(market_data, indicators_by_symbol)
@@ -301,8 +310,12 @@ def run_cycle():
                 logger.info("Take profit executed: %s, P&L: %.2f USDT (%.2f%%)",
                             sym, realized_pnl or 0, realized_pnl_pct or 0)
 
-    # 7. Build indicator summary for LLM
+    # 7. Build indicator summary for LLM (5m + 1h context + pivots)
     indicator_summary = indicators.summarize_all(indicators_by_symbol)
+    if htf_by_symbol:
+        indicator_summary += "\n" + indicators.summarize_htf(htf_by_symbol)
+    if pivots_by_symbol:
+        indicator_summary += "\n" + indicators.summarize_pivots(pivots_by_symbol, current_prices)
 
     # 8. Build account status string for LLM
     account_str = _build_account_string(port, current_prices)

stop_orders.json

@@ -1,15 +1,9 @@
 {
-  "tADAUST": {
-    "order_id": 233273720461,
-    "stop_price": 0.2626,
-    "entry_price": 0.2707080505480357,
-    "amount": 201.94477684
-  },
   "tDOGE:UST": {
     "order_id": 233282701621,
-    "stop_price": 0.09394531,
+    "stop_price": 0.093946,
     "entry_price": 0.09685083602369626,
-    "amount": 144.10790954030196
+    "amount": 144.10790954
   },
   "tSUIUST": {
     "order_id": 233265250698,