v1.1: 3 clases + fix yfinance + forzar re-entrenamiento

app.py

@@ -140,7 +140,7 @@ async def get_regime(req: RegimeRequest):
     proba = _regime_model.predict_proba(last_row)[0]
     regime_id = int(np.argmax(proba))
 
-    REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE", 3: "BREAKOUT"}
+    REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE"}
 
     return RegimeResponse(
         regime=REGIME_NAMES[regime_id],

download_data.py

@@ -353,8 +353,8 @@ def download_macro_data(days: int) -> pd.DataFrame:
     for name, ticker in tickers.items():
         try:
             logger.info("📥 Descargando %s (%s)...", name, ticker)
-            data = yf.download(ticker, start=start_date, interval="1d",
-                               progress=False, auto_adjust=True)
+            tick = yf.Ticker(ticker)
+            data = tick.history(start=start_date, interval="1d")
             if not data.empty:
                 frames[name] = data["Close"].rename(f"macro_{name.lower()}")
         except Exception as e:

regime_detector.py

@@ -37,7 +37,7 @@ logger = logging.getLogger("AuroraBrain.Regime")
 DATA_DIR = os.path.join(os.path.dirname(__file__), "data")
 MODELS_DIR = os.path.join(os.path.dirname(__file__), "models")
 
-REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE", 3: "BREAKOUT"}
+REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE"}
 
 
 def select_features(df: pd.DataFrame, target_col: str = "regime",
@@ -116,7 +116,7 @@ def train_regime_model(df: pd.DataFrame, feature_cols: list[str],
         reg_alpha=0.1,
         reg_lambda=1.0,
         objective="multi:softprob",
-        num_class=4,
+        num_class=3,
         eval_metric="mlogloss",
         random_state=42,
         n_jobs=-1,
@@ -159,7 +159,7 @@ def train_regime_model(df: pd.DataFrame, feature_cols: list[str],
         cv_model = xgb.XGBClassifier(
             n_estimators=200, max_depth=6, learning_rate=0.05,
             subsample=0.8, colsample_bytree=0.8,
-            objective="multi:softprob", num_class=4,
+            objective="multi:softprob", num_class=3,
             random_state=42, n_jobs=-1,
         )
         cv_model.fit(X.iloc[train_idx], y.iloc[train_idx], verbose=False)

regime_labeler.py

@@ -30,12 +30,11 @@ logger = logging.getLogger("AuroraBrain.Labeler")
 DATA_DIR = os.path.join(os.path.dirname(__file__), "data")
 
 # Régimen IDs
-REGIME_TRENDING = 0
+REGIME_TRENDING = 0    # Incluye BREAKOUT (movimiento fuerte en una dirección)
 REGIME_RANGING = 1
 REGIME_VOLATILE = 2
-REGIME_BREAKOUT = 3
 
-REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE", 3: "BREAKOUT"}
+REGIME_NAMES = {0: "TRENDING", 1: "RANGING", 2: "VOLATILE"}
 
 
 def label_regimes(df: pd.DataFrame, horizon: int = 12,
@@ -68,7 +67,6 @@ def label_regimes(df: pd.DataFrame, horizon: int = 12,
     logger.info("🏷️ Etiquetando regímenes (horizon=%d velas)...", horizon)
 
     for i in range(n - horizon):
-        # Ventana de futuro
         future_prices = c.iloc[i + 1: i + 1 + horizon]
         future_volumes = v.iloc[i + 1: i + 1 + horizon]
 
@@ -79,59 +77,31 @@ def label_regimes(df: pd.DataFrame, horizon: int = 12,
         if current_price <= 0:
             continue
 
-        # Métricas de la ventana futura
         max_price = future_prices.max()
         min_price = future_prices.min()
         end_price = future_prices.iloc[-1]
 
-        # Retorno neto
         return_pct = (end_price - current_price) / current_price * 100
 
-        # Máximo drawdown en la ventana
         running_max = future_prices.cummax()
         drawdowns = (running_max - future_prices) / running_max * 100
         max_drawdown = drawdowns.max()
 
-        # Máximo rally en la ventana
-        running_min = future_prices.cummin()
-        rallies = (future_prices - running_min) / running_min * 100
-        max_rally = rallies.max()
-
-        # Rango total
         price_range = (max_price - min_price) / current_price * 100
 
-        # Volumen promedio futuro vs histórico
-        vol_sma = df["f_vol_sma_20"].iloc[i] if "f_vol_sma_20" in df.columns else v.iloc[max(0, i-20):i].mean()
-        avg_future_vol = future_volumes.mean()
-        vol_ratio = avg_future_vol / vol_sma if vol_sma > 0 else 1.0
-
-        # ── Clasificación (prioridad: VOLATILE > BREAKOUT > TRENDING > RANGING) ──
-
-        # VOLATILE: drawdown fuerte
+        # ── Clasificación simplificada: 3 clases ──
+        # VOLATILE: drawdown fuerte (prioridad máxima)
         if max_drawdown >= volatile_dd_threshold:
             regimes[i] = REGIME_VOLATILE
             continue
 
-        # BREAKOUT: movimiento grande + volumen alto
-        if price_range >= breakout_threshold and vol_ratio >= breakout_vol_mult:
-            regimes[i] = REGIME_BREAKOUT
-            continue
-
-        # TRENDING: avance sostenido en una dirección
-        if abs(return_pct) >= trend_threshold:
+        # TRENDING: movimiento fuerte en cualquier dirección (incluye breakouts)
+        if abs(return_pct) >= trend_threshold or price_range >= breakout_threshold:
             regimes[i] = REGIME_TRENDING
             continue
 
-        # RANGING: precio se quedó en un rango estrecho
-        if price_range <= range_threshold:
-            regimes[i] = REGIME_RANGING
-            continue
-
-        # Default para movimientos moderados
-        if abs(return_pct) >= trend_threshold * 0.6:
-            regimes[i] = REGIME_TRENDING
-        else:
-            regimes[i] = REGIME_RANGING
+        # RANGING: todo lo demás
+        regimes[i] = REGIME_RANGING
 
     # Últimas `horizon` velas no tienen etiqueta confiable
     regimes[-horizon:] = -1  # Marcar como desconocido

startup.sh

@@ -3,37 +3,25 @@ echo "========================================="
 echo "  🧠 AURORA BRAIN — Starting Pipeline"
 echo "========================================="
 
-# Paso 1: Descargar datos (si no existen)
-if [ ! -f "data/klines_BTCUSDT_4h.parquet" ]; then
-    echo "📥 Descargando datos históricos..."
-    python download_data.py --symbols BTCUSDT ETHUSDT SOLUSDT --days 1825
-else
-    echo "✅ Datos ya descargados"
-fi
+# Limpiar datos anteriores para re-entrenar con mejoras
+rm -rf data/*.parquet models/*.pkl models/*.json
+echo "🗑️ Datos anteriores limpiados — re-entrenando desde cero"
+
+# Paso 1: Descargar datos
+echo "📥 Descargando datos históricos..."
+python download_data.py --symbols BTCUSDT ETHUSDT SOLUSDT --days 1825
 
 # Paso 2: Generar features
-if [ ! -f "data/features_BTCUSDT_4h.parquet" ]; then
-    echo "🔧 Generando features..."
-    python feature_engine.py --all
-else
-    echo "✅ Features ya generadas"
-fi
+echo "🔧 Generando features..."
+python feature_engine.py --all
 
-# Paso 3: Etiquetar regímenes
-if [ ! -f "data/labeled_BTCUSDT_4h.parquet" ]; then
-    echo "🏷️ Etiquetando regímenes..."
-    python regime_labeler.py --all
-else
-    echo "✅ Regímenes ya etiquetados"
-fi
+# Paso 3: Etiquetar regímenes (ahora 3 clases: TRENDING/RANGING/VOLATILE)
+echo "🏷️ Etiquetando regímenes..."
+python regime_labeler.py --all
 
 # Paso 4: Entrenar modelo
-if [ ! -f "models/regime_model.pkl" ]; then
-    echo "🎯 Entrenando detector de régimen..."
-    python regime_detector.py --symbol BTCUSDT
-else
-    echo "✅ Modelo ya entrenado"
-fi
+echo "🎯 Entrenando detector de régimen..."
+python regime_detector.py --symbol BTCUSDT
 
 echo "========================================="
 echo "  🚀 Levantando API..."