diff --git a/src/api/main.py b/src/api/main.py
index adb574c..e547e00 100644
--- a/src/api/main.py
+++ b/src/api/main.py
@@ -1770,6 +1770,214 @@ async def websocket_status():
     }
 
 
+# =============================================================================
+# Historical Signals Query Endpoints
+# =============================================================================
+
+class SignalQueryParams(BaseModel):
+    """Parameters for signal query"""
+    symbol: Optional[str] = None
+    direction: Optional[str] = None
+    start_date: Optional[datetime] = None
+    end_date: Optional[datetime] = None
+    min_confidence: Optional[float] = Field(default=None, ge=0.0, le=1.0)
+    amd_phase: Optional[str] = None
+    timeframe: Optional[str] = None
+    is_valid_only: bool = False
+    page: int = Field(default=1, ge=1)
+    page_size: int = Field(default=50, ge=1, le=200)
+
+
+class StoredSignalResponse(BaseModel):
+    """Single stored signal response"""
+    signal_id: str
+    symbol: str
+    direction: str
+    entry_price: float
+    stop_loss: float
+    take_profit: float
+    risk_reward_ratio: float
+    prob_tp_first: float
+    confidence_score: float
+    amd_phase: str
+    volatility_regime: str
+    timeframe: str
+    timestamp: datetime
+    valid_until: datetime
+    attention_score: Optional[float] = None
+    flow_class: Optional[int] = None
+    is_high_flow: Optional[bool] = None
+
+
+class SignalQueryResponse(BaseModel):
+    """Signal query response with pagination"""
+    signals: List[Dict[str, Any]]
+    total_count: int
+    page: int
+    page_size: int
+    has_more: bool
+    query_params: Optional[Dict[str, Any]] = None
+
+
+@app.get("/api/signals/history", response_model=SignalQueryResponse, tags=["Historical Signals"])
+async def query_historical_signals(
+    symbol: Optional[str] = Query(default=None, description="Filter by symbol"),
+    direction: Optional[str] = Query(default=None, description="Filter by direction: long, short, neutral"),
+    start_date: Optional[datetime] = Query(default=None, description="Start date filter"),
+    end_date: Optional[datetime] = Query(default=None, description="End date filter"),
+    min_confidence: Optional[float] = Query(default=None, ge=0.0, le=1.0, description="Minimum confidence"),
+    amd_phase: Optional[str] = Query(default=None, description="Filter by AMD phase"),
+    timeframe: Optional[str] = Query(default=None, description="Filter by timeframe"),
+    is_valid_only: bool = Query(default=False, description="Only return still-valid signals"),
+    page: int = Query(default=1, ge=1, description="Page number"),
+    page_size: int = Query(default=50, ge=1, le=200, description="Results per page")
+):
+    """
+    Query historical signals with filters.
+
+    Supports filtering by symbol, direction, date range, confidence, AMD phase, etc.
+    Returns paginated results sorted by timestamp (newest first).
+
+    Example queries:
+    - Get all XAUUSD signals: /api/signals/history?symbol=XAUUSD
+    - Get high confidence longs: /api/signals/history?direction=long&min_confidence=0.7
+    - Get signals from last 24h: /api/signals/history?start_date=2026-01-24T00:00:00
+    """
+    global prediction_service
+
+    if prediction_service is None:
+        prediction_service = get_prediction_service()
+
+    if not prediction_service.signal_store_available:
+        raise HTTPException(
+            status_code=503,
+            detail="Signal store not available"
+        )
+
+    result = prediction_service.query_signals(
+        symbol=symbol.upper() if symbol else None,
+        direction=direction,
+        start_date=start_date,
+        end_date=end_date,
+        min_confidence=min_confidence,
+        amd_phase=amd_phase,
+        timeframe=timeframe,
+        is_valid_only=is_valid_only,
+        page=page,
+        page_size=page_size
+    )
+
+    return SignalQueryResponse(**result)
+
+
+@app.get("/api/signals/latest", tags=["Historical Signals"])
+async def get_latest_signals(
+    symbol: Optional[str] = Query(default=None, description="Filter by symbol"),
+    limit: int = Query(default=10, ge=1, le=100, description="Number of signals")
+):
+    """
+    Get the most recent signals.
+
+    Returns the latest signals, optionally filtered by symbol.
+    """
+    global prediction_service
+
+    if prediction_service is None:
+        prediction_service = get_prediction_service()
+
+    if not prediction_service.signal_store_available:
+        raise HTTPException(
+            status_code=503,
+            detail="Signal store not available"
+        )
+
+    signals = prediction_service.get_latest_signals(
+        symbol=symbol.upper() if symbol else None,
+        limit=limit
+    )
+
+    return {
+        "count": len(signals),
+        "signals": signals
+    }
+
+
+@app.get("/api/signals/valid", tags=["Historical Signals"])
+async def get_valid_signals(
+    symbol: Optional[str] = Query(default=None, description="Filter by symbol")
+):
+    """
+    Get all currently valid signals.
+
+    Returns signals that have not yet expired (valid_until > now).
+    """
+    global prediction_service
+
+    if prediction_service is None:
+        prediction_service = get_prediction_service()
+
+    if not prediction_service.signal_store_available:
+        raise HTTPException(
+            status_code=503,
+            detail="Signal store not available"
+        )
+
+    signals = prediction_service.get_valid_signals(
+        symbol=symbol.upper() if symbol else None
+    )
+
+    return {
+        "count": len(signals),
+        "valid_signals": signals
+    }
+
+
+@app.get("/api/signals/{signal_id}", tags=["Historical Signals"])
+async def get_signal_by_id(signal_id: str):
+    """
+    Get a specific signal by its ID.
+
+    Signal IDs are in the format: SIG-XXXXXXXX
+    """
+    global prediction_service
+
+    if prediction_service is None:
+        prediction_service = get_prediction_service()
+
+    if not prediction_service.signal_store_available:
+        raise HTTPException(
+            status_code=503,
+            detail="Signal store not available"
+        )
+
+    signal = prediction_service.get_signal_by_id(signal_id)
+
+    if signal is None:
+        raise HTTPException(
+            status_code=404,
+            detail=f"Signal {signal_id} not found"
+        )
+
+    return signal
+
+
+@app.get("/api/signals/stats", tags=["Historical Signals"])
+async def get_signal_store_stats():
+    """
+    Get signal store statistics.
+
+    Returns information about stored signals, including counts by symbol
+    and direction, capacity, and retention settings.
+    """
+    global prediction_service
+
+    if prediction_service is None:
+        prediction_service = get_prediction_service()
+
+    stats = prediction_service.get_signal_store_stats()
+    return stats
+
+
 # Main entry point
 if __name__ == "__main__":
     import uvicorn
diff --git a/src/services/prediction_service.py b/src/services/prediction_service.py
index 08e2529..71f2eb0 100644
--- a/src/services/prediction_service.py
+++ b/src/services/prediction_service.py
@@ -45,6 +45,9 @@ from ..pipelines.hierarchical_pipeline import (
 from ..backtesting.rr_backtester import RRBacktester, BacktestConfig, BacktestResult
 from ..backtesting.metrics import TradingMetrics, TradeRecord
 
+# Signal store for historical queries
+from .signal_store import SignalStore, StoredSignal, get_signal_store
+
 
 class Direction(Enum):
     LONG = "long"
@@ -213,6 +216,7 @@ class PredictionService:
         self._attention_provider = None  # Level 0 attention models
         self._hierarchical_pipeline = None  # L0→L1→L2 pipeline
         self._backtester = None  # Backtesting engine
+        self._signal_store = None  # Historical signal store
         self._models_loaded = False
 
         # Symbol-specific trainers (nuevos modelos por símbolo/timeframe)
@@ -292,6 +296,19 @@ class PredictionService:
                 logger.warning(f"RRBacktester initialization failed: {e}")
                 self._backtester = None
 
+            # Initialize Signal Store for historical queries
+            try:
+                persistence_dir = os.path.join(self.models_dir, "..", "logs", "signals")
+                self._signal_store = get_signal_store(
+                    max_signals=10000,
+                    retention_hours=168,  # 7 days
+                    persistence_dir=persistence_dir
+                )
+                logger.info("✅ SignalStore initialized")
+            except Exception as e:
+                logger.warning(f"SignalStore initialization failed: {e}")
+                self._signal_store = None
+
             self._models_loaded = True
 
             # Cargar modelos por símbolo si el feature flag está activo
@@ -981,7 +998,7 @@ class PredictionService:
             adjusted_confidence = min(1.0, confidence * 1.15)  # 15% boost
             logger.debug(f"High flow detected, boosted confidence: {confidence:.3f} -> {adjusted_confidence:.3f}")
 
-        return TradingSignal(
+        signal = TradingSignal(
             signal_id=f"SIG-{uuid.uuid4().hex[:8].upper()}",
             symbol=symbol,
             direction=direction,
@@ -1005,6 +1022,142 @@ class PredictionService:
             }
         )
 
+        # Store signal for historical queries
+        self._store_signal(signal)
+
+        return signal
+
+    def _store_signal(self, signal: TradingSignal):
+        """Store a generated signal in the signal store"""
+        if not self._signal_store:
+            return
+
+        try:
+            stored = StoredSignal(
+                signal_id=signal.signal_id,
+                symbol=signal.symbol,
+                direction=signal.direction.value,
+                entry_price=signal.entry_price,
+                stop_loss=signal.stop_loss,
+                take_profit=signal.take_profit,
+                risk_reward_ratio=signal.risk_reward_ratio,
+                prob_tp_first=signal.prob_tp_first,
+                confidence_score=signal.confidence_score,
+                amd_phase=signal.amd_phase.value,
+                volatility_regime=signal.volatility_regime.value,
+                timeframe=signal.metadata.get("timeframe", "15m"),
+                timestamp=signal.timestamp,
+                valid_until=signal.valid_until,
+                attention_score=signal.attention.attention_score if signal.attention else None,
+                flow_class=signal.attention.flow_class if signal.attention else None,
+                is_high_flow=signal.attention.is_high_flow if signal.attention else None,
+                metadata=signal.metadata
+            )
+            self._signal_store.store(stored)
+        except Exception as e:
+            logger.debug(f"Failed to store signal: {e}")
+
+    @property
+    def signal_store_available(self) -> bool:
+        """Check if signal store is available"""
+        return self._signal_store is not None
+
+    def query_signals(
+        self,
+        symbol: Optional[str] = None,
+        direction: Optional[str] = None,
+        start_date: Optional[datetime] = None,
+        end_date: Optional[datetime] = None,
+        min_confidence: Optional[float] = None,
+        amd_phase: Optional[str] = None,
+        timeframe: Optional[str] = None,
+        is_valid_only: bool = False,
+        page: int = 1,
+        page_size: int = 50
+    ) -> Dict[str, Any]:
+        """
+        Query historical signals.
+
+        Args:
+            symbol: Filter by symbol
+            direction: Filter by direction ('long', 'short', 'neutral')
+            start_date: Filter by start date
+            end_date: Filter by end date
+            min_confidence: Minimum confidence score
+            amd_phase: Filter by AMD phase
+            timeframe: Filter by timeframe
+            is_valid_only: Only return signals still valid
+            page: Page number
+            page_size: Results per page
+
+        Returns:
+            Dict with signals and pagination info
+        """
+        if not self._signal_store:
+            return {
+                "signals": [],
+                "total_count": 0,
+                "page": page,
+                "page_size": page_size,
+                "has_more": False,
+                "error": "Signal store not available"
+            }
+
+        result = self._signal_store.query(
+            symbol=symbol,
+            direction=direction,
+            start_date=start_date,
+            end_date=end_date,
+            min_confidence=min_confidence,
+            amd_phase=amd_phase,
+            timeframe=timeframe,
+            is_valid_only=is_valid_only,
+            page=page,
+            page_size=page_size
+        )
+
+        return {
+            "signals": [s.to_dict() for s in result.signals],
+            "total_count": result.total_count,
+            "page": result.page,
+            "page_size": result.page_size,
+            "has_more": result.has_more,
+            "query_params": result.query_params
+        }
+
+    def get_latest_signals(self, symbol: Optional[str] = None, limit: int = 10) -> List[Dict]:
+        """Get the most recent signals"""
+        if not self._signal_store:
+            return []
+
+        signals = self._signal_store.get_latest(symbol=symbol, limit=limit)
+        return [s.to_dict() for s in signals]
+
+    def get_valid_signals(self, symbol: Optional[str] = None) -> List[Dict]:
+        """Get all currently valid signals"""
+        if not self._signal_store:
+            return []
+
+        signals = self._signal_store.get_valid(symbol=symbol)
+        return [s.to_dict() for s in signals]
+
+    def get_signal_by_id(self, signal_id: str) -> Optional[Dict]:
+        """Get a specific signal by ID"""
+        if not self._signal_store:
+            return None
+
+        signal = self._signal_store.get(signal_id)
+        return signal.to_dict() if signal else None
+
+    def get_signal_store_stats(self) -> Dict[str, Any]:
+        """Get signal store statistics"""
+        if not self._signal_store:
+            return {"available": False}
+
+        stats = self._signal_store.get_statistics()
+        stats["available"] = True
+        return stats
+
     def _determine_direction(
         self,
         amd: AMDDetection,
diff --git a/src/services/signal_store.py b/src/services/signal_store.py
new file mode 100644
index 0000000..c3f88cb
--- /dev/null
+++ b/src/services/signal_store.py
@@ -0,0 +1,462 @@
+"""
+Signal Store Service
+====================
+
+Stores and queries historical trading signals.
+Provides in-memory storage with optional persistence to database.
+
+Author: ML Pipeline
+Version: 1.0.0
+Created: 2026-01-25
+"""
+
+import asyncio
+from collections import deque
+from dataclasses import dataclass, asdict
+from datetime import datetime, timedelta
+from typing import Dict, List, Optional, Any
+from pathlib import Path
+import json
+from loguru import logger
+
+
+@dataclass
+class StoredSignal:
+    """A stored trading signal"""
+    signal_id: str
+    symbol: str
+    direction: str  # 'long', 'short', 'neutral'
+    entry_price: float
+    stop_loss: float
+    take_profit: float
+    risk_reward_ratio: float
+    prob_tp_first: float
+    confidence_score: float
+    amd_phase: str
+    volatility_regime: str
+    timeframe: str
+    timestamp: datetime
+    valid_until: datetime
+    # Optional attention info
+    attention_score: Optional[float] = None
+    flow_class: Optional[int] = None
+    is_high_flow: Optional[bool] = None
+    # Optional hierarchical info
+    hierarchical_available: bool = False
+    delta_high_final: Optional[float] = None
+    delta_low_final: Optional[float] = None
+    trade_quality: Optional[str] = None
+    # Metadata
+    metadata: Optional[Dict[str, Any]] = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        d = asdict(self)
+        d['timestamp'] = self.timestamp.isoformat()
+        d['valid_until'] = self.valid_until.isoformat()
+        return d
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> 'StoredSignal':
+        """Create from dictionary"""
+        data = data.copy()
+        if isinstance(data.get('timestamp'), str):
+            data['timestamp'] = datetime.fromisoformat(data['timestamp'])
+        if isinstance(data.get('valid_until'), str):
+            data['valid_until'] = datetime.fromisoformat(data['valid_until'])
+        return cls(**data)
+
+
+@dataclass
+class SignalQueryResult:
+    """Result of a signal query"""
+    signals: List[StoredSignal]
+    total_count: int
+    page: int
+    page_size: int
+    has_more: bool
+    query_params: Dict[str, Any]
+
+
+class SignalStore:
+    """
+    In-memory store for historical signals with query capabilities.
+
+    Features:
+    - Stores signals with automatic expiration
+    - Query by symbol, direction, date range, etc.
+    - Pagination support
+    - Optional persistence to JSON files
+    """
+
+    def __init__(
+        self,
+        max_signals: int = 10000,
+        retention_hours: int = 168,  # 7 days
+        persistence_dir: Optional[str] = None
+    ):
+        """
+        Initialize signal store.
+
+        Args:
+            max_signals: Maximum signals to keep in memory
+            retention_hours: Hours to retain signals (default 7 days)
+            persistence_dir: Optional directory for JSON persistence
+        """
+        self.max_signals = max_signals
+        self.retention_hours = retention_hours
+        self.persistence_dir = Path(persistence_dir) if persistence_dir else None
+
+        # Storage: deque for efficient FIFO, dict for fast lookup
+        self._signals: deque = deque(maxlen=max_signals)
+        self._index_by_id: Dict[str, StoredSignal] = {}
+        self._index_by_symbol: Dict[str, List[str]] = {}  # symbol -> list of signal_ids
+
+        # Statistics
+        self._stats = {
+            'total_stored': 0,
+            'total_queries': 0,
+            'signals_by_symbol': {},
+            'signals_by_direction': {'long': 0, 'short': 0, 'neutral': 0}
+        }
+
+        # Load persisted signals if directory exists
+        if self.persistence_dir:
+            self.persistence_dir.mkdir(parents=True, exist_ok=True)
+            self._load_persisted()
+
+        logger.info(f"SignalStore initialized (max={max_signals}, retention={retention_hours}h)")
+
+    def store(self, signal: StoredSignal) -> bool:
+        """
+        Store a new signal.
+
+        Args:
+            signal: Signal to store
+
+        Returns:
+            True if stored successfully
+        """
+        try:
+            # Check for duplicate
+            if signal.signal_id in self._index_by_id:
+                logger.debug(f"Signal {signal.signal_id} already exists, updating")
+                self._remove_signal(signal.signal_id)
+
+            # Add to storage
+            self._signals.append(signal)
+            self._index_by_id[signal.signal_id] = signal
+
+            # Update symbol index
+            if signal.symbol not in self._index_by_symbol:
+                self._index_by_symbol[signal.symbol] = []
+            self._index_by_symbol[signal.symbol].append(signal.signal_id)
+
+            # Update stats
+            self._stats['total_stored'] += 1
+            self._stats['signals_by_symbol'][signal.symbol] = \
+                self._stats['signals_by_symbol'].get(signal.symbol, 0) + 1
+            self._stats['signals_by_direction'][signal.direction] = \
+                self._stats['signals_by_direction'].get(signal.direction, 0) + 1
+
+            logger.debug(f"Stored signal {signal.signal_id} for {signal.symbol}")
+            return True
+
+        except Exception as e:
+            logger.error(f"Failed to store signal: {e}")
+            return False
+
+    def get(self, signal_id: str) -> Optional[StoredSignal]:
+        """Get a signal by ID"""
+        return self._index_by_id.get(signal_id)
+
+    def query(
+        self,
+        symbol: Optional[str] = None,
+        direction: Optional[str] = None,
+        start_date: Optional[datetime] = None,
+        end_date: Optional[datetime] = None,
+        min_confidence: Optional[float] = None,
+        max_confidence: Optional[float] = None,
+        amd_phase: Optional[str] = None,
+        timeframe: Optional[str] = None,
+        is_valid_only: bool = False,
+        page: int = 1,
+        page_size: int = 50,
+        sort_by: str = 'timestamp',
+        sort_desc: bool = True
+    ) -> SignalQueryResult:
+        """
+        Query signals with filters.
+
+        Args:
+            symbol: Filter by symbol (e.g., 'XAUUSD')
+            direction: Filter by direction ('long', 'short', 'neutral')
+            start_date: Filter by start date
+            end_date: Filter by end date
+            min_confidence: Minimum confidence score
+            max_confidence: Maximum confidence score
+            amd_phase: Filter by AMD phase
+            timeframe: Filter by timeframe
+            is_valid_only: Only return signals that are still valid
+            page: Page number (1-indexed)
+            page_size: Number of results per page
+            sort_by: Field to sort by
+            sort_desc: Sort descending
+
+        Returns:
+            SignalQueryResult with matching signals
+        """
+        self._stats['total_queries'] += 1
+
+        # Start with all signals or symbol-filtered
+        if symbol and symbol.upper() in self._index_by_symbol:
+            signal_ids = self._index_by_symbol[symbol.upper()]
+            signals = [self._index_by_id[sid] for sid in signal_ids if sid in self._index_by_id]
+        else:
+            signals = list(self._signals)
+
+        now = datetime.utcnow()
+
+        # Apply filters
+        filtered = []
+        for s in signals:
+            # Symbol filter (if not already applied)
+            if symbol and s.symbol.upper() != symbol.upper():
+                continue
+
+            # Direction filter
+            if direction and s.direction != direction:
+                continue
+
+            # Date range filter
+            if start_date and s.timestamp < start_date:
+                continue
+            if end_date and s.timestamp > end_date:
+                continue
+
+            # Confidence filter
+            if min_confidence is not None and s.confidence_score < min_confidence:
+                continue
+            if max_confidence is not None and s.confidence_score > max_confidence:
+                continue
+
+            # AMD phase filter
+            if amd_phase and s.amd_phase != amd_phase:
+                continue
+
+            # Timeframe filter
+            if timeframe and s.timeframe != timeframe:
+                continue
+
+            # Valid only filter
+            if is_valid_only and s.valid_until < now:
+                continue
+
+            filtered.append(s)
+
+        # Sort
+        if sort_by == 'timestamp':
+            filtered.sort(key=lambda x: x.timestamp, reverse=sort_desc)
+        elif sort_by == 'confidence':
+            filtered.sort(key=lambda x: x.confidence_score, reverse=sort_desc)
+        elif sort_by == 'symbol':
+            filtered.sort(key=lambda x: x.symbol, reverse=sort_desc)
+
+        # Pagination
+        total_count = len(filtered)
+        start_idx = (page - 1) * page_size
+        end_idx = start_idx + page_size
+        page_signals = filtered[start_idx:end_idx]
+        has_more = end_idx < total_count
+
+        return SignalQueryResult(
+            signals=page_signals,
+            total_count=total_count,
+            page=page,
+            page_size=page_size,
+            has_more=has_more,
+            query_params={
+                'symbol': symbol,
+                'direction': direction,
+                'start_date': start_date.isoformat() if start_date else None,
+                'end_date': end_date.isoformat() if end_date else None,
+                'min_confidence': min_confidence,
+                'amd_phase': amd_phase,
+                'timeframe': timeframe
+            }
+        )
+
+    def get_latest(self, symbol: Optional[str] = None, limit: int = 10) -> List[StoredSignal]:
+        """Get the latest signals"""
+        if symbol:
+            symbol = symbol.upper()
+            if symbol in self._index_by_symbol:
+                signal_ids = self._index_by_symbol[symbol][-limit:]
+                signals = [self._index_by_id[sid] for sid in reversed(signal_ids) if sid in self._index_by_id]
+                return signals[:limit]
+            return []
+
+        signals = list(self._signals)[-limit:]
+        return list(reversed(signals))
+
+    def get_valid(self, symbol: Optional[str] = None) -> List[StoredSignal]:
+        """Get all currently valid signals"""
+        now = datetime.utcnow()
+        result = self.query(
+            symbol=symbol,
+            is_valid_only=True,
+            page_size=1000
+        )
+        return result.signals
+
+    def get_statistics(self) -> Dict[str, Any]:
+        """Get store statistics"""
+        now = datetime.utcnow()
+        valid_count = len([s for s in self._signals if s.valid_until > now])
+
+        return {
+            'total_signals': len(self._signals),
+            'valid_signals': valid_count,
+            'expired_signals': len(self._signals) - valid_count,
+            'unique_symbols': len(self._index_by_symbol),
+            'symbols': list(self._index_by_symbol.keys()),
+            'total_stored_all_time': self._stats['total_stored'],
+            'total_queries': self._stats['total_queries'],
+            'signals_by_symbol': self._stats['signals_by_symbol'],
+            'signals_by_direction': self._stats['signals_by_direction'],
+            'max_capacity': self.max_signals,
+            'retention_hours': self.retention_hours
+        }
+
+    def cleanup_expired(self) -> int:
+        """Remove expired signals based on retention policy"""
+        cutoff = datetime.utcnow() - timedelta(hours=self.retention_hours)
+        removed = 0
+
+        # Find signals to remove
+        to_remove = []
+        for s in self._signals:
+            if s.timestamp < cutoff:
+                to_remove.append(s.signal_id)
+
+        # Remove them
+        for signal_id in to_remove:
+            if self._remove_signal(signal_id):
+                removed += 1
+
+        if removed > 0:
+            logger.info(f"Cleaned up {removed} expired signals")
+
+        return removed
+
+    def _remove_signal(self, signal_id: str) -> bool:
+        """Remove a signal by ID"""
+        if signal_id not in self._index_by_id:
+            return False
+
+        signal = self._index_by_id[signal_id]
+
+        # Remove from deque
+        try:
+            self._signals.remove(signal)
+        except ValueError:
+            pass
+
+        # Remove from indices
+        del self._index_by_id[signal_id]
+
+        if signal.symbol in self._index_by_symbol:
+            try:
+                self._index_by_symbol[signal.symbol].remove(signal_id)
+            except ValueError:
+                pass
+
+        return True
+
+    def persist(self, filename: Optional[str] = None) -> bool:
+        """Persist signals to JSON file"""
+        if not self.persistence_dir:
+            logger.warning("No persistence directory configured")
+            return False
+
+        try:
+            if filename is None:
+                filename = f"signals_{datetime.utcnow().strftime('%Y%m%d_%H%M%S')}.json"
+
+            filepath = self.persistence_dir / filename
+            data = {
+                'saved_at': datetime.utcnow().isoformat(),
+                'count': len(self._signals),
+                'signals': [s.to_dict() for s in self._signals]
+            }
+
+            with open(filepath, 'w') as f:
+                json.dump(data, f, indent=2, default=str)
+
+            logger.info(f"Persisted {len(self._signals)} signals to {filepath}")
+            return True
+
+        except Exception as e:
+            logger.error(f"Failed to persist signals: {e}")
+            return False
+
+    def _load_persisted(self) -> int:
+        """Load signals from persistence directory"""
+        if not self.persistence_dir or not self.persistence_dir.exists():
+            return 0
+
+        loaded = 0
+        try:
+            # Find most recent file
+            files = sorted(self.persistence_dir.glob("signals_*.json"), reverse=True)
+            if not files:
+                return 0
+
+            latest_file = files[0]
+            with open(latest_file, 'r') as f:
+                data = json.load(f)
+
+            for signal_data in data.get('signals', []):
+                try:
+                    signal = StoredSignal.from_dict(signal_data)
+                    if self.store(signal):
+                        loaded += 1
+                except Exception as e:
+                    logger.debug(f"Failed to load signal: {e}")
+
+            logger.info(f"Loaded {loaded} persisted signals from {latest_file}")
+
+        except Exception as e:
+            logger.error(f"Failed to load persisted signals: {e}")
+
+        return loaded
+
+    def clear(self):
+        """Clear all signals"""
+        self._signals.clear()
+        self._index_by_id.clear()
+        self._index_by_symbol.clear()
+        logger.info("Signal store cleared")
+
+
+# Global singleton instance
+_signal_store: Optional[SignalStore] = None
+
+
+def get_signal_store(
+    max_signals: int = 10000,
+    retention_hours: int = 168,
+    persistence_dir: Optional[str] = None
+) -> SignalStore:
+    """Get or create the global signal store instance"""
+    global _signal_store
+
+    if _signal_store is None:
+        _signal_store = SignalStore(
+            max_signals=max_signals,
+            retention_hours=retention_hours,
+            persistence_dir=persistence_dir
+        )
+
+    return _signal_store